There's no reason to have one synchronize_net() for each
removed station, refactor the code slightly to have just
a single synchronize_net() for all stations.

Note that this is currently useless as hostapd removes
stations one by one and this coalescing never happens.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

There's no reason to do this inside the sta key removal
since the keys can only be reached through the sta (and
not by the driver at all) so once the sta can no longer
be reached, the keys are safe.

This will allow further optimisation opportunities with
multiple stations.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

If we can assume that stations are never referenced by the
driver after sta_state returns (and this is true since the
previous iwlmvm patch and for all other drivers) then we
don't need to delay station destruction, and don't need to
play tricks with rcu_barrier() etc.

This should speed up some scenarios like hostapd shutdown.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

The pointer should be cleared before synchronize_rcu() so that the
consequently dead station won't be found by any lookups in the TX
or RX paths.

Also check that the station is actually the one being removed, the
check is not needed because each 4-addr VLAN can only have a single
station and non-4-addr VLANs always have a NULL pointer there, but
the code is clearer this way (and we avoid the memory write.)
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Currently, mac80211 allows drivers to keep RCU-protected station
references that are cleared when the station is removed from the
driver and consequently needs to synchronize twice, once before
removing the station from the driver (so it can guarantee that
the station is no longer used in TX towards the driver) and once
after the station is removed from the driver.

Add a new pre-RCU-synchronisation station removal operation to
the API to allow drivers to clear/invalidate their RCU-protected
station pointers before the RCU synchronisation.

This will allow removing the second synchronisation by changing
the driver API so that the driver may no longer assume a valid
RCU-protected pointer after sta_remove/sta_state returns.

The alternative to this would be to synchronize_rcu() in all the
drivers that currently rely on this behaviour (only iwlmvm) but
that would defeat the purpose.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

This fixes a regression introduced by my patch "mac80211: don't cancel
csa finalize work within stop_ap", which added sdata locks to
ieee80211_csa_finalize_work() without removing the locking for
ieee80211_ibss_finish_csa(), which is called by the former, resulting
in a deadlock due to nested locking.
Signed-off-by: NSimon Wunderlich <sw@simonwunderlich.de>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

On scan completion we try start any pending roc.

However, if scan was just pending (and not actually started)
there is no point in trying to start the roc, as it might
have started already.

This solves the following warning:
WARNING: CPU: 0 PID: 3552 at net/mac80211/offchannel.c:269 ieee80211_start_next_roc+0x164/0x204 [mac80211]()
[<c001cd38>] (unwind_backtrace+0x0/0xf0)
[<c00181d0>] (show_stack+0x10/0x14)
[<c05c0d8c>] (dump_stack+0x78/0x94)
[<c0047c08>] (warn_slowpath_common+0x68/0x8c)
[<c0047c48>] (warn_slowpath_null+0x1c/0x24)
[<bf4d6660>] (ieee80211_start_next_roc+0x164/0x204 [mac80211])
[<bf4d5a74>] (ieee80211_scan_cancel+0xe8/0x190 [mac80211])
[<bf4df970>] (ieee80211_do_stop+0x63c/0x79c [mac80211])
[<bf4dfae0>] (ieee80211_stop+0x10/0x18 [mac80211])
[<c0504d84>] (__dev_close_many+0x84/0xcc)
[<c0504df4>] (__dev_close+0x28/0x3c)
[<c0509708>] (__dev_change_flags+0x78/0x144)
[<c0509854>] (dev_change_flags+0x10/0x48)
[<c055fe3c>] (devinet_ioctl+0x614/0x6d0)
[<c04f22a0>] (sock_ioctl+0x5c/0x2a4)
[<c0124eb4>] (do_vfs_ioctl+0x7c/0x5d8)
[<c012547c>] (SyS_ioctl+0x6c/0x7c)
Signed-off-by: NEliad Peller <eliad@wizery.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

In some cases, determining the completed scan type was
done by testing the SCAN_HW_SCANNING flag.

However, this doesn't take care for the case in which
the hw scan was requested, but hasn't started yet (e.g.
due to active remain_on_channel).

Replace this test by checking whether ops->hw_scan is
defined.

This solves the following warning:

WARNING: CPU: 0 PID: 3552 at net/mac80211/offchannel.c:156 __ieee80211_scan_completed+0x1b4/0x2dc [mac80211]()
[<c001cd38>] (unwind_backtrace+0x0/0xf0)
[<c00181d0>] (show_stack+0x10/0x14)
[<c05c0d8c>] (dump_stack+0x78/0x94)
[<c0047c08>] (warn_slowpath_common+0x68/0x8c)
[<c0047c48>] (warn_slowpath_null+0x1c/0x24)
[<bf4d4504>] (__ieee80211_scan_completed+0x1b4/0x2dc [mac80211])
[<bf4d5a74>] (ieee80211_scan_cancel+0xe8/0x190 [mac80211])
[<bf4df970>] (ieee80211_do_stop+0x63c/0x79c [mac80211])
[<bf4dfae0>] (ieee80211_stop+0x10/0x18 [mac80211])
[<c0504d84>] (__dev_close_many+0x84/0xcc)
[<c0504df4>] (__dev_close+0x28/0x3c)
[<c0509708>] (__dev_change_flags+0x78/0x144)
[<c0509854>] (dev_change_flags+0x10/0x48)
[<c055fe3c>] (devinet_ioctl+0x614/0x6d0)
[<c04f22a0>] (sock_ioctl+0x5c/0x2a4)
[<c0124eb4>] (do_vfs_ioctl+0x7c/0x5d8)
[<c012547c>] (SyS_ioctl+0x6c/0x7c)
Signed-off-by: NEliad Peller <eliad@wizery.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Rename NL80211_TXRATE_MCS to NL80211_TXRATE_HT and also
rename mcs to ht_mcs in struct cfg80211_bitrate_mask.
Signed-off-by: NJanusz Dziedzic <janusz.dziedzic@tieto.com>
[reword commit message]
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

This shouldn't be happening, so warn if it does, since we'd
then leak a station entry.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Remove the unnecessary duplicate test of "if (skb) {"
when !CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS.

Remove now unnecessary __maybe_unused, neaten comment
Remove unnecessary parenthesis around align cast.
Substitute reference to deprecated compare_ether_addr
with ether_addr_equal (in the comment.)
Signed-off-by: NJoe Perches <joe@perches.com>
[edit commit log slightly]
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Add support for vendor-specific commands to nl80211. This is
intended to be used for really vendor-specific functionality
that can't be implemented in a generic fashion for any reason.
It's *NOT* intended to be used for any normal/generic feature
or any optimisations that could be implemented across drivers.

Currently, only vendor commands (with replies) are supported,
no dump operations or vendor-specific notifications.

Also add a function wdev_to_ieee80211_vif() to mac80211 which
is needed for mac80211-based drivers wanting to implement any
vendor commands.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

The same code appears just a few lines below.
Signed-off-by: NEliad Peller <eliad@wizery.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

There's no reason for the variable to not be inside
the #ifdef, move it.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

This reverts commit 488b366a.

The code isn't used by anyone, and the Intel driver isn't planning
to use it either right now.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Drivers with hardware rate control were given
sta->rx_nss set to 0. This was because rx_nss
calculation procedure was protected by hw/sw rate
control check.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

When external CSA IEs are received (beacons or action messages), a
channel switch is triggered as well. This should only be allowed on
devices which actually support channel switches, otherwise disconnect.
(For the corresponding userspace invocation, the wiphy flag is checked
in nl80211).
Signed-off-by: NSimon Wunderlich <sw@simonwunderlich.de>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Signed-off-by: NSimon Wunderlich <sw@simonwunderlich.de>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

The current channel switch code has a potential deadlock:
1) * cfg80211_stop_ap acquires wdev-lock
   * ieee80211_stop_ap calls cancel_work_sync for the csa_finalize_work,
     which acquires the associated worker-lock
2) * ieee80211_csa_finalize_work holds the worker-lock when run
   * it calls cfg80211_ch_switch_notify which will claim the wdev-lock,
     and also needs to claim the sdata-lock (which is the same as the
     wdev-lock) to modify the beacons.

It is sufficient to just set the channel switch active to false. If the
worker is running later, it will find the channel switch to not be
active anymore and returns immediately without changing anything.

Canceling the worker is done anyway when the interface goes down
(ieee80211_do_stop).
Reported-by: NJohannes Berg <johannes@sipsolutions.net>
Signed-off-by: NSimon Wunderlich <sw@simonwunderlich.de>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

The channel switch notification should be sent under the
wdev/sdata-lock, preferably in the same moment as the channel change
happens, to avoid races by other callers (e.g. start/stop_ap).
This also adds the previously missing sdata_lock protection in
csa_finalize_work.
Reported-by: NJohannes Berg <johannes@sipsolutions.net>
Signed-off-by: NSimon Wunderlich <sw@simonwunderlich.de>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

The csa finalize worker needs to change the beacon information (for
different modes). These are normally protected under rtnl lock, but the
csa finalize worker is called by drivers and should not acquire the RTNL
lock. Therefore change access protection for beacons to sdata/wdev lock.
Reported-by: NJohannes Berg <johannes@sipsolutions.net>
Signed-off-by: NSimon Wunderlich <sw@simonwunderlich.de>
[fix sdata_dereference]
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

The local TSF timer is used to compute the timing offset between
mesh peers on beacon reception.  However, asking the device for
the TSF is not very accurate, so we prefer to use rx->mactime
if available.  In the latter case, calling drv_get_tsf() just
adds more delay into the RX path, so skip it if we can.
Signed-off-by: NBob Copeland <bob@cozybit.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Change cfg80211 and mac80211 to use cfg80211_mgmt_tx_params
struct to aggregate parameters for mgmt_tx functions.
This makes the functions' signatures less clumsy and allows
less painful parameters extension.
Signed-off-by: NAndrei Otcheretianski <andrei.otcheretianski@intel.com>
[fix all other drivers]
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

There's a bug in tracking HT opmode changes in mac80211, it
fails to update the driver when the channel parameters don't
change.

Move the code to do the HT opmode checking independently of
the channel/bandwidth tracking.
Signed-off-by: NAvri Altman <avri.altman@intel.com>
[edit commit message]
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

All interface types now properly clean up their stations
using some form of sta_info_flush() themselves, so there's
no need to try it again at teardown. Remove the call to
get rid of the extra delay from the synchronize_net() and
rcu_barrier() calls.
Reported-by: NMoshe Benji <moshe.benji@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Measure TX latency and jitter statistics per station per TID.
These Measurements are disabled by default and can be enabled
via debugfs.

Features included for each station's TID:

1. Keep count of the maximum and average latency of Tx frames.
2. Keep track of many frames arrived in a specific time range
   (need to enable through debugfs and configure the bins ranges)
Signed-off-by: NMatti Gottlieb <matti.gottlieb@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

ATM minstrel_ht does not check whether a sampling rate is supported.
Unsupported rates attempts can trigger when there are holes in bitfields
of supported MCSes belonging to the same group (e.g many devices are
MCS32 capable without MCS33->39 capable, also we systematically have a
hole for CCK rates).
Drop any attempts to sample unsupported rates, as suggested by Felix.

This is not a problem in minstrel which fills a per STA sample table
with only supported rates (though only at init).
Signed-off-by: NKarl Beldan <karl.beldan@rivierawaves.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

ATM, only the first array value returned by get_random_bytes is used.
This change moves the call to get_random_bytes from the nested loop it
is in to its parent.
While at it, replace get_random_bytes with prandom_bytes since PRNs are
way enough for the selection process.
After this, minstrel_ht reclaims 80 PR-bytes instead of 640 R-bytes.

minstrels use sample tables to probe different rates in a randomized
manner.
minstrel_ht inits one single sample table upon registration (during
subsys_initcalls) and minstrel uses one per STA addition in minstrel.
Signed-off-by: NKarl Beldan <karl.beldan@rivierawaves.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Consecutive MCSes in [8*(NSS-1)->8*NSS[ have the same number NSS of
streams (except for MCS32 which is mishandled ATM).
ATM minstrel_ht uses MCS_GROUP_RATES in place of this 8 modulus.
This change replaces such occurences and by doing so allows for different
values of MCS_GROUP_RATES (e.g to cope with VHT MCS8,9).
Signed-off-by: NKarl Beldan <karl.beldan@rivierawaves.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Add a new field to ieee80211_chanctx_conf to indicate
the min required channel configuration.

Tuning to a narrower channel might help reducing
the noise level and saving some power.

The min required channel definition is the max of
all min required channel definitions of the interfaces
bound to this channel context.

In AP mode, use 20MHz when there are no connected station.
When a new station is added/removed, calculate the new max
bandwidth supported by any of the stations (e.g. 80MHz when
80MHz and 40MHz stations are connected).

In other cases, simply use bss_conf.chandef as the
min required chandef.

Notify drivers about changes to this field by calling
drv_change_chanctx with a new CHANGE_MIN_WIDTH notification.
Signed-off-by: NEliad Peller <eliad@wizery.com>
Reviewed-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Introduce shift and mask defines for beamformee STS cap and number
of sounding dimensions cap as these can take any 3 bit value.
While at it also cleanup an unrequired parenthesis.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Reviewed-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

There is no reason why we should have only one channel switch
announcement at a time for a single phy.  When support for channel
switch with multiple contexts and multiple vifs per context is
implemented, we will need the chandef data for each vif.  Move the
csa_chandef structure to sdata to prepare for this.
Reviewed-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NLuciano Coelho <luciano.coelho@intel.com>
[Fixed compilation with mesh]
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Use put_unaligned_le16 and put_unaligned_le32 for
mesh_path_error_tx and mesh_path_sel_frame_tx.
Signed-off-by: NChun-Yeow Yeoh <yeohchunyeow@gmail.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Use put_unaligned_le16 in mesh_plink_frame_tx.
Signed-off-by: NChun-Yeow Yeoh <yeohchunyeow@gmail.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

use put_unaligned_le16 for precedence value in mesh
channel switch support
Signed-off-by: NChun-Yeow Yeoh <yeohchunyeow@gmail.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

This adds generic cipher scheme support to mac80211, such schemes
are fully under control by the driver. On hw registration drivers
may specify additional HW ciphers with a scheme how these ciphers
have to be handled by mac80211 TX/RR. A cipher scheme specifies a
cipher suite value, a size of the security header to be added to
or stripped from frames and how the PN is to be verified on RX.
Signed-off-by: NMax Stepanov <Max.Stepanov@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

To report channel width correctly we have
to send correct channel parameters from
mac80211 when calling cfg80211_cac_event().

This is required in case of using channel width
higher than 20MHz and we have to set correct
dfs channel state after CAC (NL80211_DFS_AVAILABLE).
Signed-off-by: NJanusz Dziedzic <janusz.dziedzic@tieto.com>
Reviewed-by: NLuis R. Rodriguez <mcgrof@do-not-panic.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

There's no code calling ieee80211_key_replace() with both
arguments NULL and it wouldn't make sense, but in the
interest of maintainability add a warning for it. As a
side effect, this also shuts up a smatch warning.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Coverity points out that checking assoc_data->ie is
completely useless since it's an array in the struct
and can't be NULL - remove the useless checks.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

802.11-2012 13.3.1 implicitly limits the mesh local link
ID range to that of AID, since for mesh PS the local link
ID must be indicated in the TIM IE, which only holds
IEEE80211_MAX_AID bits.

Also the code was allowing a local link ID of 0, but this
is not correct since that TIM bit is used for indicating
buffered mcast frames.

Generate a random, unique, link ID from 1 - 2007, and drop
a modulo conversion for the local link ID, but keep it for
the peer link ID in case he chose something > MAX_AID.
Signed-off-by: NThomas Pedersen <thomas@cozybit.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>