Over time, a whole bunch of drivers have come up
with their own scheme to delay the configure_filter
operation to a workqueue. To be able to simplify
things, allow configure_filter to sleep, and add
a new prepare_multicast callback that drivers that
need the multicast address list implement. This new
callback must be atomic, but most drivers either
don't care or just calculate a hash which can be
done atomically and then uploaded to the hardware
non-atomically.
A cursory look suggests that at76c50x-usb, ar9170,
mwl8k (which is actually very broken now), rt2x00,
wl1251, wl1271 and zd1211 should make use of this
new capability.
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This patch fixes a bug in ath9k_stop() where the HW
was not put into FULL_SLEEP state. Not doing so will
cause issues in suspend-resume and the HW will not respond
to chip resets.
Signed-off-by: Sujith <Sujith.Manoharan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
PHY counters are available in all chipsets supported
by ath9k. Remove the check.
Signed-off-by: Sujith <Sujith.Manoharan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Turns out ath5k and ath9k can share the same helper to
allocates RX skbs. We allocate skbs aligned to the cache line
size. This requirement seems to have come from AR5210; when
this was not done it seems sometimes we'd get bogus data. I'm
also told it may have been a performance enhancement
consideration. In the end I can't be sure we can remove this
on new hardware so just keep this and start sharing it through
ath.ko.
Make ath9k start using this, ath5k is next.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
In the the patch:
ath9k: use new FIF_PSPOLL configure filter
I forgot to add the new FIF_PSPOLL to the supported mask.
Without this we mask out the FIF_PSPOLL flag therefore
we'd ignore it from mac80211.
Cc: Jouni Malinen <jouni.malinen@atheros.com>
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
The keycache reset will not fail as right above we ensure
to set the sc->keymax to be <= ah->caps.keycache_size. Just remove
this dangling check.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
If hw initialization fails (ath9k_hw_init()) on ath_init_softc()
we bail out and call ath9k_hw_detach(). The call ath9k_hw_detach()
is conditional though as ath9k_hw_init() could itself have called
ath9k_hw_detach(). Just describing this is itself a brain twister.
Avoid this nonsense by removing ath9k_hw_detach() from ath9k_hw_init().
Upon hw initialization failure we expect the callers to take care of
the cleanup.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
We re-label the device driver initialization routines from the
ath_softc, the "Software Carrier" fillers. This should make it
clearer what each of these do.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
During initialization ath9k tends to use "attach" to when we
initialize hardware due to the fact we used to attach a "HAL".
The notion of a HAL is long gone, so lets just be clear on what
we are doing.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
We're now propagating the -ENOMEM error so there is no need to
keep a debug message there now.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
The softc is cached and set within the ath_hw struct.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
We do this in case attach and friends try to get back to
ah from the softc somehow.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This lets us trim one argument off of hw initializer routines.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This lets us simplify attach code and arguments passed.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
We were never propagating hw initialization errors, lets
do that now and also use -EOPNOTSUPP when device revision is
not supported yet.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
The mac80211 workqueue exists to enable mac80211 and drivers
to queue their own work on a single threaded workqueue. mac80211
takes care to flush the workqueue during suspend but we never
really had requirements on drivers for how they should use
the workqueue in consideration for suspend.
We extend mac80211 to document how the mac80211 workqueue should
be used, how it should not be used and finally move raw access to
the workqueue to mac80211 only. Drivers and mac80211 use helpers
to queue work onto the mac80211 workqueue:
* ieee80211_queue_work()
* ieee80211_queue_delayed_work()
These helpers will now warn if mac80211 already completed its
suspend cycle and someone is trying to queue work. mac80211
flushes the mac80211 workqueue prior to suspend a few times,
but we haven't taken the care to ensure drivers won't add more
work after suspend. To help with this we add a warning when
someone tries to add work and mac80211 already completed the
suspend cycle.
Drivers should ensure they cancel any work or delayed work
in the mac80211 stop() callback.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
We should be cancelling our work at the stop callback since
we are borrowing the mac80211 workqueue for our work. As it
stands mac80211 expects this for suspend purposes.
The ath9k specific virtual wiphy stuff need only be
cancelled only when the we have no secondary virtual wiphys.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
We do this as we'll be moving the cancel elsewhere later.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
ath9k uses the mac80211 workqueue for 4 different types of work:
* Led blink work
* TX hang monitoring work
* internal wiphy schedular work
* channel change work done for internal wiphy schedular
Since the internal wiphy schedular can end up kicking off some
channel channel change work we should first cancel the wiphy
schedular work and then the channel change work.
The TX hang work can be cancelled second since we're going down
anyway.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Our version goes by the kernel release.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
There is a small window where the mac80211 changes the IEEE80211_CONF_PS
flag, and then informs the driver about the change. We have a race
condition if we are checking the flag in the same time. Avoid it by
introducing a local variable, and using that instead of checking the
IEEE80211_CONF_PS flag directly.
This fix the problem reported by Luis:
http://article.gmane.org/gmane.linux.kernel.wireless.general/34363
Changes-licensed-under: ISC
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Cleanup aggregation start/stop function interfaces.
Signed-off-by: Sujith <Sujith.Manoharan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
mac80211 will have disabled the queues for us when
needed.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This uses the new configuration changes indicated up by
mac80211 when all interfaces are marked idle. We need to do
a little more work as we have our own set of virtual
wiphys within ath9k.
Only when all virtual wiphys are inactive do we allow an idle
state change for a wiphy to trigger disabling the radio.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Because ath9k_setpower is called from various contexts, we have to
protect it against concurrent calls.
Changes-licensed-under: ISC
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
The hardware doesn't generate interrupts in some cases and so work
around this by monitoring the TX status periodically and reset the
chip if required.
This behavior of the hardware not generating the TX interrupts can
be noticed through ath9k debugfs interrupt statistics when heavy
traffic is being sent from STA to AP. One can easily see this behavior
when the STA is transmitting at a higher rates. The interrupt statistics
in the debugfs interface clearly shows that only RX interrupts alone
being generated and TX being stuck.
TX should be monitored through a timer and reset the chip only when
frames are queued to the hardware but TX interrupts are not generated
for the same even after one second. Also, we shouldn't remove holding
descriptor from AC queue if it happens to be the only descriptor and
schedule TX aggregation regarless of queue depth as it improves
scheduling of AMPDUs from software to hardware queue.
Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: Senthil Balasubramanian <senthilkumar@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
RSSI reported by the RX descriptor requires little manipulation.
Manipulate and report the correct RSSI to the stack. This will
fix the improper signal levels reported by iwconfig iw dev wlanX
station dump. Also the Link Quality reported seems to be varying
(falls to zero also sometimes) when iperf is run from STA to AP.
Also use the default noise floor for now as the one reported
during the caliberation seems to be wrong.
The Signal and Link Quality before this patch (taken while TX is
in progress from STA to AP)
09:59:13.285428037 Link Quality=29/70 Signal level=-81 dBm
09:59:13.410660084 Link Quality=20/70 Signal level=-90 dBm
09:59:13.586864392 Link Quality=21/70 Signal level=-89 dBm
09:59:13.710296281 Link Quality=21/70 Signal level=-89 dBm
09:59:13.821683064 Link Quality=25/70 Signal level=-85 dBm
09:59:13.933402989 Link Quality=24/70 Signal level=-86 dBm
09:59:14.045839276 Link Quality=26/70 Signal level=-84 dBm
09:59:14.193926673 Link Quality=23/70 Signal level=-87 dBm
09:59:14.306230262 Link Quality=31/70 Signal level=-79 dBm
09:59:14.419459667 Link Quality=26/70 Signal level=-84 dBm
09:59:14.530711167 Link Quality=37/70 Signal level=-73 dBm
09:59:14.642593962 Link Quality=29/70 Signal level=-81 dBm
09:59:14.754361169 Link Quality=21/70 Signal level=-89 dBm
09:59:14.866217355 Link Quality=21/70 Signal level=-89 dBm
09:59:14.976963623 Link Quality=28/70 Signal level=-82 dBm
09:59:15.089149809 Link Quality=26/70 Signal level=-84 dBm
09:59:15.205039887 Link Quality=27/70 Signal level=-83 dBm
09:59:15.316368003 Link Quality=23/70 Signal level=-87 dBm
09:59:15.427684036 Link Quality=36/70 Signal level=-74 dBm
09:59:15.539756380 Link Quality=21/70 Signal level=-89 dBm
09:59:15.650549093 Link Quality=22/70 Signal level=-88 dBm
09:59:15.761171672 Link Quality=32/70 Signal level=-78 dBm
09:59:15.872793750 Link Quality=23/70 Signal level=-87 dBm
09:59:15.984421694 Link Quality=22/70 Signal level=-88 dBm
09:59:16.097315093 Link Quality=21/70 Signal level=-89 dBm
The link quality and signal level after this patch (take while
TX is in progress from STA to AP)
17:21:25.627848091 Link Quality=65/70 Signal level=-45 dBm
17:21:25.762805607 Link Quality=65/70 Signal level=-45 dBm
17:21:25.875521888 Link Quality=66/70 Signal level=-44 dBm
17:21:25.987468448 Link Quality=66/70 Signal level=-44 dBm
17:21:26.100628151 Link Quality=66/70 Signal level=-44 dBm
17:21:26.213129671 Link Quality=66/70 Signal level=-44 dBm
17:21:26.324923070 Link Quality=65/70 Signal level=-45 dBm
17:21:26.436831357 Link Quality=65/70 Signal level=-45 dBm
17:21:26.610356973 Link Quality=65/70 Signal level=-45 dBm
17:21:26.723340047 Link Quality=65/70 Signal level=-45 dBm
17:21:26.835715293 Link Quality=64/70 Signal level=-46 dBm
17:21:26.949542748 Link Quality=64/70 Signal level=-46 dBm
17:21:27.062261613 Link Quality=65/70 Signal level=-45 dBm
17:21:27.174511563 Link Quality=64/70 Signal level=-46 dBm
17:21:27.287616232 Link Quality=64/70 Signal level=-46 dBm
17:21:27.400598119 Link Quality=64/70 Signal level=-46 dBm
17:21:27.511381404 Link Quality=64/70 Signal level=-46 dBm
17:21:27.624530421 Link Quality=65/70 Signal level=-45 dBm
17:21:27.737807109 Link Quality=64/70 Signal level=-46 dBm
17:21:27.850861352 Link Quality=65/70 Signal level=-45 dBm
17:21:27.963369436 Link Quality=64/70 Signal level=-46 dBm
17:21:28.076582289 Link Quality=64/70 Signal level=-46 dBm
Signed-off-by: Senthil Balasubramanian <senthilkumar@atheros.com>
Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This has no functional change and just cleans up the code
to be more legible and removes a useless variable for
Multi Rate Retry.
For regular frames we use 2 retries for MRR segments [0-2].
For the last MRR segment [3] we use 4.
MRR[0] = 2
MRR[1] = 2
MRR[2] = 2
MRR[3] = 4
Cc: Derek Smithies <derek@indranet.co.nz>
Cc: Chittajit Mitra <Chittajit.Mitra@Atheros.com>
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This patch fixes an issue when the TX and RX streams are different.
Signed-off-by: Senthil Balasubramanian <senthilkumar@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
ANI calibration shouldn't be done when we are not on our home channel.
This is already verified. However, it is racy. Fix this by proper
spin locks.
Signed-off-by: Senthil Balasubramanian <senthilkumar@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
ANI is not required when the STA is disconnected. So stop it and enable
ANI for adhoc and monitor mode.
Signed-off-by: Senthil Balasubramanian <senthilkumar@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Remove unncessary STATION mode check in ath9k_bss_assoc_info() as
it is called only for STATION mode.
Signed-off-by: Senthil Balasubramanian <senthilkumar@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
We want to put the chip into FULL SLEEP state, when we are disabling the
radio, but the the current code always change it to AWAKE/NETWORK SLEEP.
Changes-licensed-under: ISC
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
The ath9k_ps_{wakeup,restore} calls must be in balance.
Changes-licensed-under: ISC
Signed-off-by: Gabor Juhos <juhosg@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This ports the ath9k rfkill code to the new API offered by
cfg80211 and thus removes a lot of useless stuff.
("With this series a kernel panic, which is a regression, during module
unload disappears." -- Vasanthakumar Thiagarajan <vasanth@atheros.com>
Other patches in the series:
ath9k: Add helper to get ath9k specific current channel
ath9k: Make sure we have current channel in ah_curchan before rf
disable/enable
-- JWL)
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Cc: Luis Rodriguez <mcgrof@gmail.com>
Tested-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This patch completely rewrites the rfkill core to address
the following deficiencies:
* all rfkill drivers need to implement polling where necessary
rather than having one central implementation
* updating the rfkill state cannot be done from arbitrary
contexts, forcing drivers to use schedule_work and requiring
lots of code
* rfkill drivers need to keep track of soft/hard blocked
internally -- the core should do this
* the rfkill API has many unexpected quirks, for example being
asymmetric wrt. alloc/free and register/unregister
* rfkill can call back into a driver from within a function the
driver called -- this is prone to deadlocks and generally
should be avoided
* rfkill-input pointlessly is a separate module
* drivers need to #ifdef rfkill functions (unless they want to
depend on or select RFKILL) -- rfkill should provide inlines
that do nothing if it isn't compiled in
* the rfkill structure is not opaque -- drivers need to initialise
it correctly (lots of sanity checking code required) -- instead
force drivers to pass the right variables to rfkill_alloc()
* the documentation is hard to read because it always assumes the
reader is completely clueless and contains way TOO MANY CAPS
* the rfkill code needlessly uses a lot of locks and atomic
operations in locked sections
* fix LED trigger to actually change the LED when the radio state
changes -- this wasn't done before
Tested-by: Alan Jenkins <alan-jenkins@tuffmail.co.uk>
Signed-off-by: Henrique de Moraes Holschuh <hmh@hmh.eng.br> [thinkpad]
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Some APs seem to drift away from the expected TBTT (timestamp %
beacon_int_in_usec differs quite a bit from zero) which can result in
us waking up way too early to receive a Beacon frame. In order to work
around this, re-configure the Beacon timers after having received a
Beacon frame from the AP (i.e., when we know the offset between the
expected TBTT and the actual time the AP is sending out the Beacon
frame).
Signed-off-by: Jouni Malinen <jouni.malinen@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
This is always discarded anyway but lets just set this to our
safest lowest.
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
ath_regd_init() needs to be called with the wiphy already
properly set with the bands. Without this the custom regulatory
settings were not taking effect, and the device would get
the default channel settings from ath9k_[25]ghz_chantable.
Reported-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
mac80211 does not set PM field for normal data frames, so we need to
update that based on the current PS mode when using PS-Poll
(timeout=0) power save mode.
This allows the AP to remain in sync with our PS state. However, there
is still a potential race condition between PS state changes when
multiple TX queues are used and nullfunc and PS-Poll frames use
different queue. That corner case may need to be handled separately by
changing which queue is used either in ath9k or mac80211.
Signed-off-by: Jouni Malinen <jouni.malinen@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
We must make sure the chip is awake when changing the RX filter
parameters. This could have caused problems, e.g., when changing the
interface to promiscuous mode while in sleep mode.
Signed-off-by: Jouni Malinen <jouni.malinen@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
If the chip complains about TSF sync, make sure we remain awake to
sync with the next Beacon frame. In theory, this should not be needed
since we are currently trying to receive all Beacon frames, anyway,
better have this code ready should we ever change that.
Signed-off-by: Jouni Malinen <jouni.malinen@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Johannes Berg