Commit Graph

455103 Commits

Author SHA1 Message Date
Eliad Peller
50675360a8 iwlwifi: mvm: disable beacon filtering escape timer in d0i3
The beacon filtering configuration in d0i3 currently uses
the max value defined as 1024. However, with beacon interval
of 100ms we end up with too-frequent wakeups.

Instead, configure the escape timer to 0, asking the fw
to disable it altogether.

Signed-off-by: Eliad Peller <eliadx.peller@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:38 +03:00
Emmanuel Grumbach
34c8b24ff2 iwlwifi: mvm: BT Coex - avoid the shared antenna for management frames
If BT is active, we need to refrain from using the shared
antenna. This logic is done in the firmware when we use the
link quality tables. But for management frames, the rate is
written in the Tx command by the driver.
Hence the driver needs to make sure not use the shared
antenna when BT is active for any frames that don't use
the rate scale table such as management frames or multicast.

Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:37 +03:00
Emmanuel Grumbach
8286d9f50a iwlwifi: mvm: BT Coex - add reduced Tx power thresholds to constants
This really belongs to the constants file.

Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:36 +03:00
Emmanuel Grumbach
35fbf5d08e iwlwifi: mvm: BT Coex - don't limit rate control if TTC is on
When the firmware enables TxTxCorunning, we can lift the
constaints on the rate control.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:36 +03:00
Emmanuel Grumbach
7fa4fa0c44 iwlwifi: mvm: BT Coex - allow MIMO in more cases
We can Tx in MIMO rates when we are in TxTx Disallow mode
just like we can when we are in Tight mode.
Same if we are in 5Ghz regardless of the mode we are.
Change the code to allow MIMO in these cases.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:35 +03:00
Emmanuel Grumbach
a39979a8c3 iwlwifi: mvm: BT Coex - allow to force the antenna allocation
This can be used for testing.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:34 +03:00
Liad Kaufman
5daddc9902 iwlwifi: mvm: assure no overflows occur while reading otp
Just in case sizes change in the OTP without proper SW
updating, an additional check is inserted when reading OTP
sections to assure no overflows occur.

Signed-off-by: Liad Kaufman <liad.kaufman@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:34 +03:00
Liad Kaufman
f251c07c88 iwlwifi: nvm: update maximal parsed values in external nvm
Some of the maximum values of the parsed external NVM file
in the B-step of the 8000 HW family were updated, so this
updates these values.

Signed-off-by: Liad Kaufman <liad.kaufman@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:33 +03:00
Liad Kaufman
b513ee7fd6 iwlwifi: update trans->hw_rev 8000 hw family format
The format of the CSR_HW_REV register has changed in 8000
HW family. To keep backwards compatibility, we store the
value of this register as usual in trans->hw_rev, only we
store it in the old format in this variable.

Signed-off-by: Liad Kaufman <liad.kaufman@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:32 +03:00
Eran Harary
6a68a39fcd iwlwifi: mvm: fix bug in set_hw_address function
Don't use nvm_hw pointer if it is NULL. Print an error
message if the MAC address isn't valid.

Signed-off-by: Eran Harary <eran.harary@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:31 +03:00
Andy Lutomirski
c9e38e7709 iwlwifi: dvm: add a force_cam module parameter to fully disable power saving
iwldvm stalls are often blamed on power management.  Add an option
to force it all the way off.

Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:31 +03:00
Andy Lutomirski
f40faf6237 iwlwifi: pcie: improve debugfs queue info
This adds need_update and write_actual to rx_queue and need_update
and an HCMD indicator to tx_queue.

On my card, rx_queue now looks like:

read: 181
write: 180
write_actual: 176
need_update: 0
free_count: 40
closed_rb_num: 181

tx_queue now looks like:

hwq 00: read=29 write=30 use=1 stop=0 need_update=0
hwq 01: read=0 write=0 use=1 stop=0 need_update=0
hwq 02: read=128 write=128 use=1 stop=0 need_update=0
hwq 03: read=0 write=0 use=1 stop=0 need_update=0
hwq 04: read=94 write=94 use=1 stop=0 need_update=0 HCMD
hwq 05: read=0 write=0 use=0 stop=0 need_update=0
hwq 06: read=0 write=0 use=0 stop=0 need_update=0
hwq 07: read=0 write=0 use=0 stop=0 need_update=0
hwq 08: read=0 write=0 use=0 stop=0 need_update=0
hwq 09: read=0 write=0 use=0 stop=0 need_update=0
hwq 10: read=0 write=0 use=0 stop=0 need_update=0
hwq 11: read=0 write=0 use=0 stop=0 need_update=0
hwq 12: read=0 write=0 use=0 stop=0 need_update=0
hwq 13: read=0 write=0 use=0 stop=0 need_update=0
hwq 14: read=0 write=0 use=0 stop=0 need_update=0
hwq 15: read=0 write=0 use=0 stop=0 need_update=0
hwq 16: read=0 write=0 use=0 stop=0 need_update=0
hwq 17: read=0 write=0 use=0 stop=0 need_update=0
hwq 18: read=0 write=0 use=0 stop=0 need_update=0
hwq 19: read=0 write=0 use=0 stop=0 need_update=0

This may help with debugging queue stalls.

Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:30 +03:00
Andy Lutomirski
d536c32b45 iwlwifi: pcie: log when waking the NIC for hcmd submission fails
I've never seen this happen, but it's useful to rule it out.

Signed-off-by: Andy Lutomirski <luto@amacapital.net>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:29 +03:00
David Spinadel
9ee6dace73 iwlwifi: fix NVM channel attribute map.
Fix NVM channel attributes. Add indoor-only and GO Concurrent
bits. Remove DFS channel bit which is overlapped with radar.

Signed-off-by: David Spinadel <david.spinadel@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:29 +03:00
Eran Harary
52ab4c86a7 iwlwifi: mvm: fixes for 8000 NVM flow
The nvm_file should be loaded by default for SDIO
procucts only. Change the configuration accordingly.
While at it, fix a typo in the device name.

Signed-off-by: Eran Harary <eran.harary@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:28 +03:00
Emmanuel Grumbach
c2d202017d iwlwifi: pcie: add firmware monitor capabilities
This allows to use the firmware monitor. This capability
uses a lot of contiguous memory (up to 64MB), so make its
usage module parameter dependent.

The driver will try to allocate as much contiguous memory
as possible downgrading its requirements until the
allocation succeeds.

Dump this data into the fw-error dump file when an error
happens.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:27 +03:00
Emmanuel Grumbach
1fa1605648 iwlwifi: rename iwl_mvm_fw_error_next_data
This is not related to mvm. Rename to iwl_fw_error_next_data.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:26 +03:00
Emmanuel Grumbach
06ddbf5ada iwlwifi: add device / firmware to fw-error-dump file
This can be useful later for parsing since the parsing may
differ based on the device's family / bus.
Also add the human readable version of the firmware.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:26 +03:00
Eliad Peller
5788ab75ec iwlwifi: remove MCS32 support declaration
MCS32 is currently not supported, so don't
declare support for it.

Signed-off-by: Eliad Peller <eliadx.peller@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:25 +03:00
David Spinadel
762533ba9c iwlwifi: mvm: don't use hardcoded num of scan channels
Use num of scan channels as advertised by fw TLV.

Signed-off-by: David Spinadel <david.spinadel@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 21:55:24 +03:00
Emmanuel Grumbach
35a4e18171 Merge remote-tracking branch 'iwlwifi-fixes/master' into iwlwifi-next 2014-06-24 15:05:09 +03:00
Johannes Berg
ab48003032 iwlwifi: mvm: add back support for low-priority scan
The low-priority scan feature can be useful, e.g. for OBSS
scans (if those are required by the AP); add back support
for it, restoring the maximum out time to the value it was
for low-priority scan before that was removed.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 15:04:39 +03:00
David Spinadel
0564679bb6 iwlwifi: mvm: use ksize to memset scan_command
Use ksize to get scan command size instead of calculating it, to
avoid nasty bugs.

Signed-off-by: David Spinadel <david.spinadel@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 15:04:14 +03:00
Emmanuel Grumbach
14ca0f389c iwlwifi: bump API version for 8000 devices
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 14:31:41 +03:00
Eliad Peller
89879413eb iwlwifi: mvm: rework sched scan channel configuration
The current sched scan channel configuration code
configures all the supported channels for scanning.

However, this can result in SYSASSERT in some cases,
when the configured channel is disabled.

Instead, configure only the channels given in the
req struct, and set the channel_count field appropriately.

While on it, change the code to use channel->hw_value
instead of recalculating the channel number.

Signed-off-by: Eliad Peller <eliadx.peller@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
2014-06-24 14:30:56 +03:00
Arik Nemtsov
ee10f2c779 mac80211: protect TDLS discovery session
After sending a TDLS discovery-request, we expect a reply to arrive on
the AP's channel. We must stay on the channel (no PSM, scan, etc.), since
a TDLS setup-response is a direct packet not buffered by the AP.
Add a new mac80211 driver callback to allow discovery session protection.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:28:19 +02:00
Arik Nemtsov
7adc3e4664 mac80211: make sure TDLS peer STA exists during setup
Make sure userspace added a TDLS peer station before invoking the
transmission of the first setup frame. This ensures packets to the peer
won't go through the AP path.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:28:18 +02:00
Arik Nemtsov
c887f0d3a0 mac80211: add API to request TDLS operation from userspace
Write a mac80211 to the cfg80211 API for requesting a userspace TDLS
operation. Define TDLS specific reason codes that can be used here.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:28:17 +02:00
Arik Nemtsov
db67d661e8 mac80211: implement proper Tx path flushing for TDLS
As the spec mandates, flush data in the AP path before transmitting the
first setup frame. Data packets transmitted during setup are already
dropped in the Tx path.

For the teardown flow, flush all packets in the direct path before
transmitting the teardown frame. Un-authorize the peer sta after teardown
is sent, forcing all subsequent Tx to the peer through the AP.

Make sure to flush the queues when disabling the link to get the
teardown packet out.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
[adjust to Luca's new quuee API and stop only vif queues]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:27:58 +02:00
Arik Nemtsov
191dd46905 mac80211: split tdls_mgmt function
There are setup/teardown specific actions to be done that accompany
the sending of a TDLS management packet. Split the main function to
simplify future additions.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:24:55 +02:00
Arik Nemtsov
2fb6b9b8e5 mac80211: use TDLS initiator in tdls_mgmt operations
The TDLS initiator is set once during link setup. If determines the
address ordering in the link identifier IE.
Use the value from userspace in order to have a correct teardown packet.
With the current code, a teardown from the responder side fails the TDLS
MIC check because of a bad link identifier IE.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:24:55 +02:00
Arik Nemtsov
31fa97c5de cfg80211: pass TDLS initiator in tdls_mgmt operations
The TDLS initiator is set once during link setup. If determines the
address ordering in the link identifier IE.

Fix dependent drivers - mwifiex and mac80211.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:24:55 +02:00
Arik Nemtsov
17e6a59a36 mac80211: cleanup TDLS state during failed setup
When setting up a TDLS session, register a delayed work to remove
the peer if setup times out. Prevent concurrent setups to support this
capacity.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:24:55 +02:00
Arik Nemtsov
68885a54cd mac80211: set auth flags after other station info
For TDLS, the AUTHORIZED flag arrives with all other important station
info (supported rates, HT/VHT caps, ...). Make sure to set the station
state in the low-level driver after transferring this information to
the mac80211 STA entry.
This aligns the STA information during sta_state callbacks with the
non-TDLS case.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:24:55 +02:00
Arik Nemtsov
9deba04d0f mac80211: clarify TDLS Tx handling
Rename the flags used in the Tx path and add an explanation for the
reasons to drop, send directly or through the AP.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:24:54 +02:00
Luciano Coelho
a46992b441 mac80211: stop only the queues assigned to the vif during channel switch
Instead of stopping all the hardware queues during channel switch,
which is especially bad when we have large CSA counts, stop only the
queues that are assigned to the vif that is performing the channel
switch.

Additionally, check for (sdata->csa_block_tx) instead of calling
ieee80211_csa_needs_block_tx(), which can now be removed.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:22:29 +02:00
Luciano Coelho
26da23b695 mac80211: add functions to stop and wake all queues assigned to a vif
In some cases we may want to stop the queues of a single vif (for
instance during a channel-switch).  Add a function that stops all the
queues that are assigned to a vif.  If a queue is assigned to more
than one vif, the corresponding netdev subqueue of the other vif(s)
will also be stopped.  If the HW doesn't set the
IEEE80211_HW_QUEUE_CONTROL flag, then all queues are stopped.

Also add a corresponding function to wake the queues of a vif back.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:22:27 +02:00
Luciano Coelho
cca07b00a5 mac80211: introduce refcount for queue_stop_reasons
Sometimes different vifs may be stopping the queues for the same
reason (e.g. when several interfaces are performing a channel switch).
Instead of using a bitmask for the reasons, use an integer that holds
a refcount instead.  In order to keep it backwards compatible,
introduce a boolean in some functions that tell us whether the queue
stopping should be refcounted or not.  For now, use not refcounted for
all calls to keep it functionally the same as before.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:22:25 +02:00
Luciano Coelho
59f48fe22f mac80211: don't stop all queues when flushing
There is no need to stop all queues when we want to flush specific
queues, so stop only the queues that will be flushed.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:22:24 +02:00
Thomas Gleixner
8d7b70fb7b net: Mac80211: Remove silly timespec dance
Converting time from one format to another seems to give coders a warm
and fuzzy feeling.

Use the proper interfaces.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: John W. Linville <linville@tuxdriver.com>
[fix compile error]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:22:21 +02:00
Johannes Berg
3283e286b8 mac80211_hwsim: clean up own address matching
Using perm_addr is always wrong, it may be reassigned by
anyone using standard netdev APIs. Remove that from the
match function and also use the match function where only
the perm_addr was used now.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:22:20 +02:00
Thomas Gleixner
181715203b mac80211: Use ktime_get_ts()
do_posix_clock_monotonic_gettime() is a leftover from the initial
posix timer implementation which maps to ktime_get_ts().

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:22:18 +02:00
Michal Kazior
10d78f2782 mac80211: use csa counter offsets instead of csa_active
vif->csa_active is protected by mutexes only. This
means it is unreliable to depend on it on codeflow
in non-sleepable beacon and CSA code. There was no
guarantee to have vif->csa_active update be
visible before beacons are updated on SMP systems.

Using csa counter offsets which are embedded in
beacon struct (and thus are protected with single
RCU assignment) is much safer.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:22:16 +02:00
Michal Kazior
af296bdb8d mac80211: move csa counters from sdata to beacon/presp
Having csa counters part of beacon and probe_resp
structures makes it easier to get rid of possible
races between setting a beacon and updating
counters on SMP systems by guaranteeing counters
are always consistent against given beacon struct.

While at it relax WARN_ON into WARN_ON_ONCE to
prevent spamming logs and racing.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
[remove pointless array check]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 14:22:06 +02:00
Janusz Dziedzic
b49328361b mac80211: allow tx via monitor iface when DFS
Allow send frames using monitor interface
when DFS chandef and we pass CAC (beaconing
allowed).

This fix problem when old kernel and new backports used,
in such case hostapd create/use also monitor interface.
Before this patch all frames hostapd send using monitor
iface were dropped when AP was configured on DFS channel.

Signed-off-by: Janusz Dziedzic <janusz.dziedzic@tieto.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 11:05:35 +02:00
Johannes Berg
b7ffbd7ef6 cfg80211: make ethtool the driver's responsibility
Currently, cfg80211 tries to implement ethtool, but that doesn't
really scale well, with all the different operations. Make the
lower-level driver responsible for it, which currently only has
an effect on mac80211. It will similarly not scale well at that
level though, since mac80211 also has many drivers.

To cleanly implement this in mac80211, introduce a new file and
move some code to appropriate places.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 11:05:33 +02:00
Johannes Berg
ba9030c20a mac80211: remove weak WEP IV accounting
Since WEP is practically dead, there seems very little
point in keeping WEP weak IV accounting.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 11:05:31 +02:00
Antonio Ospite
b314c66990 trivial: net/mac80211/mesh.c: fix typo s/Substract/Subtract/
Signed-off-by: Antonio Ospite <ao2@ao2.it>
Cc: Luis Carlos Cobo <luisca@cozybit.com>
Cc: Johannes Berg <johannes@sipsolutions.net>
Cc: linux-wireless@vger.kernel.org
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 11:05:29 +02:00
Bob Copeland
2b470c39e8 mac80211: remove ignore_plink_timer flag
The mesh_plink code is doing some interesting things with the
ignore_plink_timer flag.  It seems the original intent was to
handle this race:

cpu 0                           cpu 1
-----                           -----
                                start timer handler for state X
acquire sta_lock
change state from X to Y
mod_timer() / del_timer()
release sta_lock
                                acquire sta_lock
                                execute state Y timer too soon

However, using the mod_timer()/del_timer() return values to
detect these cases is broken.  As a result, timers get ignored
unnecessarily, and stations can get stuck in the peering state
machine.

Instead, we can detect the case by looking at the timer expiration.
In the case of del_timer, just ignore the timers in the following
(LISTEN/ESTAB) states since they won't have timers anyway.

Signed-off-by: Bob Copeland <me@bobcopeland.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 11:05:27 +02:00
Johannes Berg
5ac2e35030 mac80211: fix station/driver powersave race
It is currently possible to have a race due to the station PS
unblock work like this:
 * station goes to sleep with frames buffered in the driver
 * driver blocks wakeup
 * station wakes up again
 * driver flushes/returns frames, and unblocks, which schedules
   the unblock work
 * unblock work starts to run, and checks that the station is
   awake (i.e. that the WLAN_STA_PS_STA flag isn't set)
 * we process a received frame with PM=1, setting the flag again
 * ieee80211_sta_ps_deliver_wakeup() runs, delivering all frames
   to the driver, and then clearing the WLAN_STA_PS_DRIVER and
   WLAN_STA_PS_STA flags

In this scenario, mac80211 will think that the station is awake,
while it really is asleep, and any TX'ed frames should be filtered
by the device (it will know that the station is sleeping) but then
passed to mac80211 again, which will not buffer it either as it
thinks the station is awake, and eventually the packets will be
dropped.

Fix this by moving the clearing of the flags to exactly where we
learn about the situation. This creates a problem of reordering,
so introduce another flag indicating that delivery is being done,
this new flag also queues frames and is cleared only while the
spinlock is held (which the queuing code also holds) so that any
concurrent delivery/TX is handled correctly.

Reported-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2014-06-23 11:05:25 +02:00