~shefty/rdma-dev.git
8 years agoMerge branch 'for-linville' of git://github.com/kvalo/ath6kl
John W. Linville [Fri, 15 Feb 2013 19:06:32 +0000 (14:06 -0500)]
Merge branch 'for-linville' of git://github.com/kvalo/ath6kl

8 years agoMerge branch 'for-john' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac802...
John W. Linville [Fri, 15 Feb 2013 18:59:53 +0000 (13:59 -0500)]
Merge branch 'for-john' of git://git./linux/kernel/git/jberg/mac80211-next

8 years agomac80211_hwsim: ask mac80211 to reserve space for chanctx.drv_priv
Karl Beldan [Thu, 14 Feb 2013 23:03:14 +0000 (00:03 +0100)]
mac80211_hwsim: ask mac80211 to reserve space for chanctx.drv_priv

Otherwise memory corruption occurs when using channel contexts (ATM when
param 'channel' > 1).

Signed-off-by: Karl Beldan <karl.beldan@rivierawaves.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: prevent spurious HT/VHT downgrade message
Johannes Berg [Thu, 14 Feb 2013 11:13:53 +0000 (12:13 +0100)]
mac80211: prevent spurious HT/VHT downgrade message

Even when connecting to an AP that doesn't support VHT,
and even when the local device doesn't support it either,
the downgrade message gets printed. Suppress the message
if HT and/or VHT is disabled.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agonl80211: renumber NL80211_FEATURE_FULL_AP_CLIENT_STATE
Johannes Berg [Thu, 14 Feb 2013 10:56:13 +0000 (11:56 +0100)]
nl80211: renumber NL80211_FEATURE_FULL_AP_CLIENT_STATE

Adding the flag to mac80211 already without testing was
clearly a mistake, one that we now pay for by having to
reserve bit 13 forever. The problem is cfg80211 doesn't
allow capability/rate changes for station entries that
were added unassociated, so the station entries cannot
be set up properly when marked associated.

Change the NL80211_FEATURE_FULL_AP_CLIENT_STATE value
to make it clear to userspace implementations that all
current kernels don't actually support it, even though
the previous bit is set, and of course also remove the
flag from mac80211 until we test and fix the issues.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agocfg80211: Pass TDLS peer's QoS/HT/VHT information during set_station
Jouni Malinen [Thu, 14 Feb 2013 19:10:54 +0000 (21:10 +0200)]
cfg80211: Pass TDLS peer's QoS/HT/VHT information during set_station

The information of the peer's capabilities is required for the driver
to perform TDLS Peer UAPSD operations. This information of the peer is
passed by the supplicant using NL80211_CMD_SET_STATION command. This
commit enhances the function nl80211_set_station to pass this
information of the peer to the driver in case this command is used
with the TDLS peer STA.

In addition, make the HT/VHT capability configuration handled more
consistently for other STA cases (reject both instead of just HT).

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agocfg80211: Pass station (extended) capability info to kernel
Jouni Malinen [Thu, 14 Feb 2013 19:10:13 +0000 (21:10 +0200)]
cfg80211: Pass station (extended) capability info to kernel

The information of the peer's capabilities and extended capabilities are
required for the driver to perform TDLS Peer UAPSD operations and off
channel operations. This information of the peer is passed from user space
using NL80211_CMD_SET_STATION command. This commit enhances
the function nl80211_set_station to pass the capability information of
the peer to the driver.

Similarly, there may be need for capability information for other modes,
so allow this to be provided with both add_station and change_station.

Signed-off-by: Jouni Malinen <jouni@qca.qualcomm.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: advertise operating mode notification capability
Johannes Berg [Mon, 11 Feb 2013 13:27:08 +0000 (14:27 +0100)]
mac80211: advertise operating mode notification capability

Use the new extended capabilities advertising to advertise
the fact that operating mode notification is supported.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agocfg80211: advertise extended capabilities to userspace
Johannes Berg [Mon, 11 Feb 2013 13:20:05 +0000 (14:20 +0100)]
cfg80211: advertise extended capabilities to userspace

In many cases, userspace may need to know which of the
802.11 extended capabilities ("Extended Capabilities
element") are implemented in the driver or device, to
include them e.g. in beacons, assoc request/response
or other frames. Add a new nl80211 attribute to hold
the extended capabilities bitmap for this.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: stop modifying HT SMPS capability
Johannes Berg [Tue, 12 Feb 2013 13:21:00 +0000 (14:21 +0100)]
mac80211: stop modifying HT SMPS capability

Instead of modifying the HT SMPS capability field
for stations, track the SMPS mode explicitly in a
new field in the station struct and use it in the
drivers that care about it. This simplifies the
code using it.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: generate mesh probe responses
Thomas Pedersen [Thu, 14 Feb 2013 19:20:14 +0000 (11:20 -0800)]
mac80211: generate mesh probe responses

Mesh interfaces will now respond to any broadcast (or
matching directed mesh) probe requests with a probe
response.

Signed-off-by: Thomas Pedersen <thomas@cozybit.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: cache mesh beacon
Thomas Pedersen [Thu, 14 Feb 2013 19:20:13 +0000 (11:20 -0800)]
mac80211: cache mesh beacon

Previously, the entire mesh beacon would be generated each
time the beacon timer fired. Instead generate a beacon
head and tail (so the TIM can easily be inserted when mesh
power save is on) when starting a mesh or the MBSS
parameters change.

Also add a mutex for protecting beacon updates and
preventing leaks.

Signed-off-by: Thomas Pedersen <thomas@cozybit.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: constify IE parsing
Johannes Berg [Tue, 12 Feb 2013 15:43:19 +0000 (16:43 +0100)]
mac80211: constify IE parsing

Make all the parsed IE pointers const, and propagate
the change to all the users etc.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agonl80211: advertise HT/VHT channel limitations
Johannes Berg [Wed, 12 Dec 2012 16:59:39 +0000 (17:59 +0100)]
nl80211: advertise HT/VHT channel limitations

When drivers or regulatory have limitations on
40, 80 or 160 MHz channels, advertise these to
userspace via nl80211. Also add a new feature
flag to let userspace know this is supported.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agocfg80211: allow drivers to selectively disable 80/160 MHz
Johannes Berg [Wed, 12 Dec 2012 16:50:39 +0000 (17:50 +0100)]
cfg80211: allow drivers to selectively disable 80/160 MHz

Some drivers might support 80 or 160 MHz only on some
channels for whatever reason, so allow them to disable
these channel widths. Also maintain the new flags when
regulatory bandwidth limitations would disable these
wide channels.

Reviewed-by: Luis R. Rodriguez <mcgrof@do-not-panic.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: properly track HT/VHT operation changes
Johannes Berg [Fri, 8 Feb 2013 14:12:14 +0000 (15:12 +0100)]
mac80211: properly track HT/VHT operation changes

A while ago, I made the mac80211 station code never change
the channel type after association. This solved a number of
issues but is ultimately wrong, we should react if the AP
changes the HT operation IE and switches bandwidth. One of
the issues is that we associate as HT40 capable, but if the
AP ever switches to 40 MHz we won't be able to receive such
frames because we never set our channel to 40 MHz.

This addresses this and VHT operation changes. If there's a
change that is incompatible with our setup, e.g. if the AP
decides to change the channel entirely (and for some reason
we still hear the beacon) we'll just disconnect.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: move ieee80211_determine_chantype function
Johannes Berg [Fri, 8 Feb 2013 13:52:32 +0000 (14:52 +0100)]
mac80211: move ieee80211_determine_chantype function

The next patch will need it further up in the file, so
move it unchanged now.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: add ieee80211_vif_change_bandwidth
Johannes Berg [Thu, 7 Feb 2013 20:37:29 +0000 (21:37 +0100)]
mac80211: add ieee80211_vif_change_bandwidth

For HT and VHT the current bandwidth can change,
add the function ieee80211_vif_change_bandwidth()
to take care of this. It returns a failure if the
new bandwidth isn't compatible with the existing
channel context, the caller has to handle that.
When it happens, also inform the driver that the
bandwidth changed for this virtual interface (no
drivers would actually care today though.)

Changing to/from HT/VHT isn't allowed though.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: clean up channel use in ieee80211_config_ht_tx
Johannes Berg [Fri, 8 Feb 2013 12:15:55 +0000 (13:15 +0100)]
mac80211: clean up channel use in ieee80211_config_ht_tx

The channel use is confusing, some uses the channel
context and some the bss_conf.chandef. The latter is
fine, so get rid of the channel context part.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: disable HT/VHT if AP has no HT/VHT capability
Johannes Berg [Thu, 7 Feb 2013 22:33:32 +0000 (23:33 +0100)]
mac80211: disable HT/VHT if AP has no HT/VHT capability

Having HT/VHT operation IEs but not capability IEs
leads to a strange situation where we configure the
channel to an HT or VHT bandwidth and then can't
actually use it. Prevent this by checking that the
HT and VHT capability IEs are present as well as
the operation IEs; if not, disable HT and/or VHT.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: handle operating mode notif in beacon/assoc response
Johannes Berg [Thu, 7 Feb 2013 21:24:55 +0000 (22:24 +0100)]
mac80211: handle operating mode notif in beacon/assoc response

In beacons and association response frames an AP may include an
operating mode notification element to advertise changes in the
number of spatial streams it can receive. Handle this using the
existing function that handles the action frame, but only handle
NSS changes, not bandwidth changes which aren't allowed here.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: fix ieee80211_change_chandef name
Johannes Berg [Thu, 7 Feb 2013 20:30:37 +0000 (21:30 +0100)]
mac80211: fix ieee80211_change_chandef name

This should be called ieee80211_change_chanctx() since
it changes the channel context, not a chandef.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: fix HT/VHT disable flags
Johannes Berg [Thu, 7 Feb 2013 19:41:50 +0000 (20:41 +0100)]
mac80211: fix HT/VHT disable flags

The code to disable HT and VHT if VHT was advertised
without VHT is wrong -- it accidentally uses the wrong
flags. Fix that.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: init HT TX data before rate control
Johannes Berg [Thu, 7 Feb 2013 16:36:12 +0000 (17:36 +0100)]
mac80211: init HT TX data before rate control

In case of connection, the station data is initialised from
the beacon/probe response first and then updated from the
association response. If the latter is different we update
the rate control algorithm and driver. Instead of doing it
this way, set the station data properly with data from the
association response before initializing rate control.

Also simplify the code by passing the station pointer.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: handle VHT operating mode notification
Johannes Berg [Thu, 27 Dec 2012 17:55:36 +0000 (18:55 +0100)]
mac80211: handle VHT operating mode notification

Handle the operating mode notification action frame.
When the supported streams or the bandwidth change
let the driver and rate control algorithm know.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: track number of spatial streams
Johannes Berg [Thu, 27 Dec 2012 17:26:42 +0000 (18:26 +0100)]
mac80211: track number of spatial streams

With VHT, a station can change the number of spatial
streams it can receive on the fly, not unlike spatial
multiplexing in HT. Prepare for that by tracking the
maximum number of spatial streams it can receive when
the connection is established.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agowireless: define operating mode action frame
Johannes Berg [Thu, 27 Dec 2012 17:45:41 +0000 (18:45 +0100)]
wireless: define operating mode action frame

Define the action frame format, the VHT category
and its action types and the field format and EID
for operating mode notifications. The frame may
be used outside of VHT context as well, so don't
include "VHT" in the names.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: stop toggling IEEE80211_HT_CAP_SUP_WIDTH_20_40
Johannes Berg [Thu, 7 Feb 2013 10:47:44 +0000 (11:47 +0100)]
mac80211: stop toggling IEEE80211_HT_CAP_SUP_WIDTH_20_40

For VHT, many more bandwidth changes are possible. As a first
step, stop toggling the IEEE80211_HT_CAP_SUP_WIDTH_20_40 flag
in the HT capabilities and instead introduce a bandwidth field
indicating the currently usable bandwidth to transmit to the
station. Of course, make all drivers use it.

To achieve this, make ieee80211_ht_cap_ie_to_sta_ht_cap() get
the station as an argument, rather than the new capabilities,
so it can set up the new bandwidth field.

If the station is a VHT station and VHT bandwidth is in use,
also set the bandwidth accordingly.

Doing this allows us to get rid of the supports_40mhz flag as
the HT capabilities now reflect the true capability instead of
the current setting.

While at it, also fix ieee80211_ht_cap_ie_to_sta_ht_cap() to not
ignore HT cap overrides when MCS TX isn't supported (not that it
really happens...)

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: pass station to ieee80211_vht_cap_ie_to_sta_vht_cap
Johannes Berg [Thu, 7 Feb 2013 10:58:58 +0000 (11:58 +0100)]
mac80211: pass station to ieee80211_vht_cap_ie_to_sta_vht_cap

Like with HT, make things a bit simpler in future patches by
passing the station to ieee80211_vht_cap_ie_to_sta_vht_cap()
instead of the vht_cap pointer. Also disable VHT here if HT
isn't supported.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agoMerge remote-tracking branch 'mac80211/master' into HEAD
Johannes Berg [Fri, 15 Feb 2013 08:41:21 +0000 (09:41 +0100)]
Merge remote-tracking branch 'mac80211/master' into HEAD

8 years agomac80211: use spin_lock_bh() for TKIP lock
Johannes Berg [Wed, 13 Feb 2013 17:03:00 +0000 (18:03 +0100)]
mac80211: use spin_lock_bh() for TKIP lock

Since no driver calls the TKIP functions from interrupt
context, there's no need to use spin_lock_irqsave().
Just use spin_lock_bh() (and spin_lock() in the TX path
where we're in a BH or they're already disabled.)

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: use spin_lock_bh() for tim_lock
Johannes Berg [Wed, 13 Feb 2013 16:39:53 +0000 (17:39 +0100)]
mac80211: use spin_lock_bh() for tim_lock

There's no need to use _irqsave() as the lock
is never used in interrupt context.

This also fixes a problem in the iwlwifi MVM
driver that calls spin_unlock_bh() within its
set_tim() callback.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: don't pick up WPA vendor IE
Johannes Berg [Tue, 12 Feb 2013 15:27:04 +0000 (16:27 +0100)]
mac80211: don't pick up WPA vendor IE

There's no use for it, WPA is entirely handled in
wpa_supplicant in userspace, so don't pick the IE.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: always unblock CSA queue stop when disconnecting
Johannes Berg [Wed, 13 Feb 2013 13:21:45 +0000 (14:21 +0100)]
mac80211: always unblock CSA queue stop when disconnecting

In some cases when disconnecting after (or during?) CSA
the queues might not recover, and then the only way to
recover is reloading the module.

Fix this by always unblocking the queue CSA reason when
disconnecting.

Cc: stable@vger.kernel.org
Reported-by: Jan-Michael Brummer <jan.brummer@tabos.org>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: don't call bss_info_changed on p2p-device/monitor
Johannes Berg [Wed, 13 Feb 2013 12:50:51 +0000 (13:50 +0100)]
mac80211: don't call bss_info_changed on p2p-device/monitor

Since the idle decision rework, mac80211 started calling
bss_info_changed() for the driver's monitor interface,
which causes a crash for iwlwifi, but drivers generally
don't expect this to happen. Therefore, avoid it.

While at it, also prevent calling it in such cases and
only print a warning. For the P2P Device interface the
idle will no longer be called (no channel context), so
also prevent that and warn on it.

Reported-by: Chaitanya <chaitanya.mgit@gmail.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: fix auth/assoc timeout handling
Johannes Berg [Wed, 13 Feb 2013 14:39:57 +0000 (15:39 +0100)]
mac80211: fix auth/assoc timeout handling

In my commit 1672c0e31917f49d31d30d79067103432bc20cc7
("mac80211: start auth/assoc timeout on frame status")
I broke auth/assoc timeout handling: in case we wait
for the TX status, it now leaves the timeout field set
to 0, which is a valid time and can compare as being
before now ("jiffies"). Thus, if the work struct runs
for some other reason, the auth/assoc is treated as
having timed out.

Fix this by introducing a separate "timeout_started"
variable that tracks whether the timeout has started
and is checked before timing out.

Additionally, for proper TX status handling the change
requires that the skb->dev pointer is set up for all
the frames, so set it up for all frames in mac80211.

Reported-by: Wojciech Dubowik <Wojciech.Dubowik@neratec.com>
Tested-by: Wojciech Dubowik <Wojciech.Dubowik@neratec.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: fix ieee80211_sta_tx_notify for nullfunc
Wojciech Dubowik [Thu, 14 Feb 2013 13:08:37 +0000 (14:08 +0100)]
mac80211: fix ieee80211_sta_tx_notify for nullfunc

Function ieee80211_sta_reset_conn_monitor has been
resetting probe_send_count too early and nullfunc
check was never called after succesfull ack.

Reported-by: Magnus Cederlöf <mcider@gmail.com>
Tested-by: Magnus Cederlöf <mcider@gmail.com>
Signed-off-by: Wojciech Dubowik <Wojciech.Dubowik@neratec.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: consolidate MBSS change notification
Thomas Pedersen [Wed, 13 Feb 2013 20:14:19 +0000 (12:14 -0800)]
mac80211: consolidate MBSS change notification

A few mesh utility functions will call
ieee80211_bss_info_change_notify(), and then the caller
might notify the driver of the same change again. Avoid
this redundancy by propagating the BSS changes and
generally calling bss_info_change_notify() once per
change.

Signed-off-by: Thomas Pedersen <thomas@cozybit.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: fix WPA with VLAN on AP side with ps-sta
Michael Braun [Wed, 13 Feb 2013 15:49:16 +0000 (16:49 +0100)]
mac80211: fix WPA with VLAN on AP side with ps-sta

When sending a broadcast while at least on of the connected stations is
sleeping, it gets queued and send after a DTIM beacon is sent.
If the packet was to be sent on a vlan interface, the vif used for dequeing
from the per-bss queue does not hold the per-vlan sdata. The correct sdata is
required to use the correct per-vlan broadcast/multicast key.

This patch fixes this by restoring the per-vlan sdata using the skb->dev entry.

Signed-off-by: Michael Braun <michael-dev@fami-braun.de>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: free ps->bc_buf skbs on vlan device stop
Michael Braun [Wed, 13 Feb 2013 15:49:15 +0000 (16:49 +0100)]
mac80211: free ps->bc_buf skbs on vlan device stop

When the vlan device is removed, ps->bc_buf processing can no longer
send its frames.

Signed-off-by: Michael Braun <michael-dev@fami-braun.de>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: add radar detection command/event
Simon Wunderlich [Fri, 8 Feb 2013 17:16:20 +0000 (18:16 +0100)]
mac80211: add radar detection command/event

Add command to trigger radar detection in the driver/FW.
Once radar detection is started it should continuously
monitor for radars as long as the channel active.
If radar is detected usermode notified with 'radar
detected' event.

Scanning and remain on channel functionality must be disabled
while doing radar detection/scanning, and vice versa.

Based on original patch by Victor Goldenshtein <victorg@ti.com>

Signed-off-by: Simon Wunderlich <siwu@hrz.tu-chemnitz.de>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agonl80211/cfg80211: add radar detection command/event
Simon Wunderlich [Fri, 8 Feb 2013 17:16:19 +0000 (18:16 +0100)]
nl80211/cfg80211: add radar detection command/event

Add new NL80211_CMD_RADAR_DETECT, which starts the Channel
Availability Check (CAC). This command will also notify the
usermode about events (CAC finished, CAC aborted, radar
detected, NOP finished).
Once radar detection has started it should continuously
monitor for radars as long as the channel is active.

This patch enables DFS for AP mode in nl80211/cfg80211.

Based on original patch by Victor Goldenshtein <victorg@ti.com>

Signed-off-by: Simon Wunderlich <siwu@hrz.tu-chemnitz.de>
[remove WIPHY_FLAG_HAS_RADAR_DETECT again -- my mistake]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agobrcmsmac: Remove unnecessary memset casts
Joe Perches [Thu, 14 Feb 2013 01:33:21 +0000 (17:33 -0800)]
brcmsmac: Remove unnecessary memset casts

Remove the casts of the first argument of memset.

Neaten the style by using the sizeof the actual variable
being memset not the sizeof the type of variable.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agobrcmsmac: Downgrade d11hdrs_mac80211 error messages to warnings.
Joe Perches [Thu, 14 Feb 2013 01:33:20 +0000 (17:33 -0800)]
brcmsmac: Downgrade d11hdrs_mac80211 error messages to warnings.

These messages don't seem to be errors but notifications
that some attribute isn't quite right.

Don't mark them as errors.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agoiwlegacy: more checks for dma mapping errors
Stanislaw Gruszka [Wed, 13 Feb 2013 14:49:08 +0000 (15:49 +0100)]
iwlegacy: more checks for dma mapping errors

This patch check output of pci_map_single() calls. I missed them on
my previous patch "iwlegacy: check for dma mapping errors", which
fixed only pci_map_page() calls.

To handle remaining possible dma mappings errors, we need to rearrange
ilXXXX_tx_skb() and il_enqueue_hcmd() functions.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agort2x00: check for dma mappings errors
Stanislaw Gruszka [Wed, 13 Feb 2013 13:27:05 +0000 (14:27 +0100)]
rt2x00: check for dma mappings errors

Check output of dma_map_single functions which nowadays can fail (when
IOMMU is used). On write_beacon callbacks just print error, similar
like padding error is handled by rt2800_write_beacon.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agomwifiex: device specific sleep cookie handling for PCIe
Avinash Patil [Tue, 12 Feb 2013 22:38:32 +0000 (14:38 -0800)]
mwifiex: device specific sleep cookie handling for PCIe

This patch adds support for handling of PCIe sleep cookie depending
upon device properties. Some PCIe devices need sleep cookie probing
before accessing HW while some others don't. A new sleep_cookie
variable is defined as part of mwifiex_pcie_card_reg strcture and
set/reset as per device capability.

Sleep cookie is allocated/accessed/freed only when flag sleep_cookie
for this particular device is enabled.

Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agortlwifi: Rework Kconfig
Larry Finger [Tue, 12 Feb 2013 22:09:02 +0000 (16:09 -0600)]
rtlwifi: Rework Kconfig

As the number of drivers in the rtlwifi family has grown, the Kconfig
section for them has grown unwieldy. This change has two effects: (1)
Variable RTLWIFI_DEBUG is documented, and (2) the entries for the
drivers that depend on RTLWIFI are indented.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agobrcmsmac: fix brcms_c_country_valid()
Dan Carpenter [Tue, 12 Feb 2013 12:23:14 +0000 (15:23 +0300)]
brcmsmac: fix brcms_c_country_valid()

ccode is not NUL terminated.  Presumably insisting on a terminator makes
brcms_c_country_valid() return false when it's not intended.  ccode[2]
is sprom->leddc_on_time.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless
John W. Linville [Thu, 14 Feb 2013 19:23:33 +0000 (14:23 -0500)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless

Conflicts:
drivers/net/wireless/iwlwifi/dvm/tx.c

8 years agocfg80211: configuration for WoWLAN over TCP
Johannes Berg [Wed, 23 Jan 2013 21:57:40 +0000 (22:57 +0100)]
cfg80211: configuration for WoWLAN over TCP

Intel Wireless devices are able to make a TCP connection
after suspending, sending some data and waking up when
the connection receives wakeup data (or breaks). Add the
WoWLAN configuration and feature advertising API for it.

Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211/minstrel_ht: add support for using CCK rates
Felix Fietkau [Wed, 13 Feb 2013 09:51:08 +0000 (10:51 +0100)]
mac80211/minstrel_ht: add support for using CCK rates

When MCS rates start to get bad in 2.4 GHz because of long range or
strong interference, CCK rates can be a lot more robust.

This patch adds a pseudo MCS group containing CCK rates (long preamble
in the lower 4 slots, short preamble in the upper slots).

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
[make minstrel_ht_get_stats static]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agocfg80211: check vendor IE length to avoid overrun
Luciano Coelho [Tue, 12 Feb 2013 18:11:38 +0000 (20:11 +0200)]
cfg80211: check vendor IE length to avoid overrun

cfg80211_find_vendor_ie() was checking only that the vendor IE would
fit in the remaining IEs buffer.  If a corrupt includes a vendor IE
that is too small, we could potentially overrun the IEs buffer.

Fix this by checking that the vendor IE fits in the reported IE length
field and skip it otherwise.

Reported-by: Jouni Malinen <j@w1.fi>
Signed-off-by: Luciano Coelho <coelho@ti.com>
[change BUILD_BUG_ON to != 1 (from >= 2)]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agonl80211: add packet offset information for wowlan pattern
Amitkumar Karwar [Tue, 12 Feb 2013 20:16:26 +0000 (12:16 -0800)]
nl80211: add packet offset information for wowlan pattern

If user knows the location of a wowlan pattern to be matched in
Rx packet, he can provide an offset with the pattern. This will
help drivers to ignore initial bytes and match the pattern
efficiently.

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
[refactor pattern sending]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agoMerge branch 'for-john' of git://git.kernel.org/pub/scm/linux/kernel/git/iwlwifi...
John W. Linville [Tue, 12 Feb 2013 17:51:08 +0000 (12:51 -0500)]
Merge branch 'for-john' of git://git./linux/kernel/git/iwlwifi/iwlwifi-next

Conflicts:
drivers/net/wireless/iwlwifi/mvm/mac80211.c

8 years agortlwifi: Initialize rate_init member of struct rate_control_ops
Catalin Iacob [Mon, 11 Feb 2013 21:18:05 +0000 (22:18 +0100)]
rtlwifi: Initialize rate_init member of struct rate_control_ops

This partially reverts commit 44ba973699b831414c3f8eef68ee5a7fe1208a05.

rate_control_rate_init assumes the rate_init member of
struct rate_control_ops is not NULL therefore not initializing it leads to
an oops as soon the driver succesfully associates to an AP.

The removal of rate_update from 44ba973699b831414c3f8eef68ee5a7fe1208a05
is ok because rate_update is checked for NULL before being
called.

Signed-off-by: Catalin Iacob <iacobcatalin@gmail.com>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agoMerge branch 'for-john' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac802...
John W. Linville [Tue, 12 Feb 2013 16:06:52 +0000 (11:06 -0500)]
Merge branch 'for-john' of git://git./linux/kernel/git/jberg/mac80211-next

8 years agoiwlwifi: mvm: beautify code in rx_handlers
Emmanuel Grumbach [Sun, 10 Feb 2013 11:25:25 +0000 (13:25 +0200)]
iwlwifi: mvm: beautify code in rx_handlers

Make the code more readable, and while at it also
add a missing "break" to avoid checking handlers
that cannot be used.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agoiwlwifi: mvm: update station when marked associated
Johannes Berg [Tue, 12 Feb 2013 12:10:44 +0000 (13:10 +0100)]
iwlwifi: mvm: update station when marked associated

In managed mode, the HT/VHT capabilities aren't set when
the station is initially added, so update the station
when it is marked associated. In AP/GO mode, the station
will typically be added with full capabilities today,
but an upcoming change in hostapd may mean a similar
scenario as for managed mode, therefore do the update
unconditionally.

Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agoiwlwifi: mvm: don't wait for session protection to start
Johannes Berg [Tue, 29 Jan 2013 14:28:17 +0000 (15:28 +0100)]
iwlwifi: mvm: don't wait for session protection to start

Now that mac80211 no longer starts the auth/assoc
timeouts when it transmits the frame, but only when
the frame status arrives, we no longer need to wait
for the session protection time event to start, we
can schedule it and enqueue the auth/assoc frame
right away. This reduces the amount of time we block
mac80211's workqueue.

Also, since now we no longer need different behavior
for session protection and P2P time events, refactor
the code to have just a common implementation.

Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agoiwlwifi: mvm: don't delay the association until after beacon
Johannes Berg [Thu, 24 Jan 2013 22:48:23 +0000 (23:48 +0100)]
iwlwifi: mvm: don't delay the association until after beacon

If we haven't heard a beacon before we associate we can
still start the association process and set the MAC in
the firmware to associated only after having received a
beacon with DTIM period by reacting to the new change
flag (BSS_CHANGED_DTIM_PERIOD) from mac80211.

This reduces the association time in these cases.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agoiwlwifi: dvm: apply beacon changes immediately
Johannes Berg [Fri, 8 Feb 2013 22:13:39 +0000 (23:13 +0100)]
iwlwifi: dvm: apply beacon changes immediately

If the AP/GO beacon changes, apply such a change
immediately, otherwise the AP/GO beacon can be
stale for a long time.

Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agoiwlwifi: mvm: report wakeup reasons
Johannes Berg [Tue, 22 Jan 2013 12:02:09 +0000 (13:02 +0100)]
iwlwifi: mvm: report wakeup reasons

Query the wakeup reasons properly and then
report them to mac80211.

Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agoiwlwifi: dvm: query and report WoWLAN wakeup reason
Johannes Berg [Tue, 8 Jan 2013 10:29:12 +0000 (11:29 +0100)]
iwlwifi: dvm: query and report WoWLAN wakeup reason

Implement proper WoWLAN wakeup and query the wakeup
reasons, then report them to userspace.

Note that this is tricky: a firmware bug (that has
been fixed in later versions) means that the status
command response isn't properly closed in hardware
and thus won't arrive at the host. Sending another
command after it closes the status response but the
next command gets stuck, etc. We reset the device
after querying though, so this is not a big issue,
just makes for strange code.

Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agoMerge remote-tracking branch 'wireless-next/master' into iwlwifi-next
Johannes Berg [Tue, 12 Feb 2013 15:52:04 +0000 (16:52 +0100)]
Merge remote-tracking branch 'wireless-next/master' into iwlwifi-next

8 years agoiwlwifi: mvm: use atomic interface iteration to avoid deadlock
Emmanuel Grumbach [Mon, 11 Feb 2013 13:56:01 +0000 (15:56 +0200)]
iwlwifi: mvm: use atomic interface iteration to avoid deadlock

Using the non-atomic version creates a dependency between
mac80211's iflist_mtx and mvm->mutex. Use the atomic version
instead which doesn't take iflist_mtx but can't sleep, so
send the HCMD in ASYNC.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: allow driver to be stateless wrt. SMSP requests
Emmanuel Grumbach [Sun, 10 Feb 2013 13:55:20 +0000 (15:55 +0200)]
mac80211: allow driver to be stateless wrt. SMSP requests

Bail out if no update is made to the SMPS state. This
allows the driver to avoid duplicating the state.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agonl80211: minor correction in sample wowlan mask calculation
Amitkumar Karwar [Tue, 12 Feb 2013 07:56:40 +0000 (23:56 -0800)]
nl80211: minor correction in sample wowlan mask calculation

The match 00:xx:00:00:xx:00:00:00:00:xx:xx:xx (where xx indicates
"don't care") should be represented by a pattern of twelve zero
bytes, and a mask of "0xed,0x01", not "0xed,0x07".

mask_len = (pat_len + 7) / 8 = (12 + 7) / 8 = 2
Hence the mask will be of 2 bytes.
Replace each valid byte in pattern by 1 and don't care byte by 0:
10110111 1000 (0000)

1st byte of pattern corresponds to lower order bit in first byte
of mask. And 9th byte of pattern corresponds to lower order bit
in second byte of mask. With this logic the mask will be
11101101 00000001 = 0xed 0x01

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: Add flushes before going off-channel
Seth Forshee [Mon, 11 Feb 2013 17:21:08 +0000 (11:21 -0600)]
mac80211: Add flushes before going off-channel

We've got a couple of races when enabling powersave with an AP for
off-channel operation. The first is fairly simple. If we go off-channel
before the nullfunc frame to enable PS is transmitted then it may not be
received by the AP. Add a flush after enabling off-channel PS to prevent
this from happening.

The second race is a bit more subtle. If the driver supports QoS and has
frames queued when the nullfunc frame is queued, those frames may get
transmitted after the nullfunc frame. If PM is not set then the AP is
being told that we've exited PS before we go off-channel and may try to
deliver frames. To prevent this, add a flush after stopping the queues
but before passing the nullfunc frame to the driver.

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: Fix tx queue handling during scans
Seth Forshee [Mon, 11 Feb 2013 17:21:07 +0000 (11:21 -0600)]
mac80211: Fix tx queue handling during scans

Scans currently work by stopping the netdev tx queues but leaving the
mac80211 queues active. This stops the flow of incoming packets while
still allowing mac80211 to transmit nullfunc and probe request frames to
facilitate scanning. However, the driver may try to wake the mac80211
queues while in this state, which will also wake the netdev queues.

To prevent this, add a new queue stop reason,
IEEE80211_QUEUE_STOP_REASON_OFFCHANNEL, to be used when stopping the tx
queues for off-channel operation. This prevents the netdev queues from
waking when a driver wakes the mac80211 queues.

This also stops all frames from being transmitted, even those meant to
be sent off-channel. Add a new tx control flag,
IEEE80211_TX_CTL_OFFCHAN_TX_OK, which allows frames to be transmitted
when the queues are stopped only for the off-channel stop reason. Update
all locations transmitting off-channel frames to use this flag.

Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211/minstrel_ht: remove the sampling bypass check for the lowest rate
Felix Fietkau [Mon, 11 Feb 2013 13:36:04 +0000 (14:36 +0100)]
mac80211/minstrel_ht: remove the sampling bypass check for the lowest rate

It's more of an unnecessary micro-optimization and it prevents switching
from long-GI to short-GI in HT20/single-stream for the lowest rate

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211/minstrel_ht: show the number of retries for each rate in debugfs
Felix Fietkau [Mon, 11 Feb 2013 13:36:03 +0000 (14:36 +0100)]
mac80211/minstrel_ht: show the number of retries for each rate in debugfs

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agoath6kl: provide 64-bit per-station byte counters
Vladimir Kondratiev [Tue, 5 Feb 2013 09:44:48 +0000 (11:44 +0200)]
ath6kl: provide 64-bit per-station byte counters

Internally, 64-bit byte counters maintained for per-station
statistics. Tell to the netlink that full 64-bit value provided

Signed-off-by: Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agowlcore: remove newly introduced alloc/OOM messages
Luciano Coelho [Mon, 11 Feb 2013 08:21:18 +0000 (10:21 +0200)]
wlcore: remove newly introduced alloc/OOM messages

In commit 0d2e7a5c (wireless: Remove unnecessary alloc/OOM messages,
alloc cleanups) OOM messages after alloc were removed from the wlcore
modules.

Commit afb43e6d (wlcore: remove if_ops from platform_data)
reintroduced a couple of those.  This patch removes them.

Signed-off-by: Luciano Coelho <coelho@ti.com>
Reviewed-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agowireless: rt2x00: rt2800usb add "unknown" devices
Xose Vazquez Perez [Sat, 9 Feb 2013 03:14:35 +0000 (04:14 +0100)]
wireless: rt2x00: rt2800usb add "unknown" devices

0x06f8, 0xe036 Hercules  Wireless Dual Band 600 USB Key HWNUm-600
0x0b05, 0x17a7 ASUS      USB-N10H 150Mbps 11n Wireless USB dongle
0x0df6, 0x0069 Sitecom   Wireless Dualband Network Adapter 300N X5 WLA-5000
0x0df6, 0x006f Sitecom   WiFi USB adapter N600 WLA-5100
0x13d3, 0x3340 AzureWave 802.11n USB Wireless LAN Card
0x13d3, 0x3399 AzureWave 802.11n USB Wireless LAN Card
0x13d3, 0x3400 AzureWave 802.11n USB Wireless LAN Card
0x13d3, 0x3401 AzureWave 802.11n USB Wireless LAN Card
0x1740, 0x0600 EnGenius  Dual-Band Wireless Media Adapter
0x1740, 0x0602 EnGenius  802.11 a/b/g/n Wireless USB Adapter
0x177f, 0x0254 Sweex     LW054 Wireless 54G Adapter USB
0xf201, 0x5370 TP-LINK   54Mbps Wireless USB Adapter

no RF3053, and I believe no RT5572.

Cc: Ivo van Doorn <IvDoorn@gmail.com>
Cc: Gertjan van Wingerde <gwingerde@gmail.com>
Cc: Helmut Schaa <helmut.schaa@googlemail.com>
Cc: John W. Linville <linville@tuxdriver.com>
Cc: users@rt2x00.serialmonkey.com
Cc: linux-wireless@vger.kernel.org
Signed-off-by: Xose Vazquez Perez <xose.vazquez@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agomwifiex: add PCIe8897 support
Avinash Patil [Sat, 9 Feb 2013 02:18:09 +0000 (18:18 -0800)]
mwifiex: add PCIe8897 support

This patch adds PCIe8897 support to mwifiex.
In PCIe8897 PFU (pre-fetch unit) is enabled by default.
This patch adds support to accommodate this feaure as well.

Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Yogesh Ashok Powar <yogeshp@marvell.com>
Signed-off-by: Nishant Sarmukadam <nishants@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: Frank Huang <frankh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agomwifiex: define generic data type for PCIe ring buffers
Avinash Patil [Sat, 9 Feb 2013 02:18:08 +0000 (18:18 -0800)]
mwifiex: define generic data type for PCIe ring buffers

This patch defines PCIe ring buffer array pointer as void instead
of mwifiex_pcie_buf_desc. This will enable us to use same pointers
for ring operations instead of new structures if buffer descriptor
structure changes.

Also split out event buffer descriptor structure from struct
mwifiex_pcie_buf_desc. For PCIe8766 TX/RX buffer descriptor is
same as evevt buffer descriptor. Newer chips could use different
TX/RX buffer descriptor while event descriptor remains the same.

Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agomwifiex: separate ring initialization and ring creation routines
Avinash Patil [Sat, 9 Feb 2013 02:18:07 +0000 (18:18 -0800)]
mwifiex: separate ring initialization and ring creation routines

This patch separates PCIe ring initialization from ring creation
routines. This modularizes ring creation(TXBD, RXBD and event
rings) functions.

Readability has been improved while moving the code around.

Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agomwifiex: store card specific data in PCI device table entry
Avinash Patil [Sat, 9 Feb 2013 02:18:06 +0000 (18:18 -0800)]
mwifiex: store card specific data in PCI device table entry

This patch adds support for storing PCIe device specific data
into driver_data structure of pci_device_id. When a device with
known device_id is probed, we use this driver_data to populate
card specific structres in driver.
This enables to remove device specific defines for scratch
registers, firmware name, FW download block size, etc. from
source code.

This will make addition of support for new chipsets a lot easier.

Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agoath9k: Fix IBSS joiner mode
Sujith Manoharan [Mon, 4 Feb 2013 10:08:24 +0000 (15:38 +0530)]
ath9k: Fix IBSS joiner mode

On joining an existing IBSS network, beaconing has to start
only after a TSF sync has happened by receiving a beacon from
the BSS. In creator mode, beaconing can start immediately after
a HW reset has been done.

Now that mac80211 notifies the driver of the mode type (creator/joiner)
via ieee80211_bss_conf->ibss_creator, make use of it to properly setup
the HW beacon timers.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agoath9k: Fix ATH9K_HW_CAP_HT usage
Sujith Manoharan [Mon, 4 Feb 2013 10:08:23 +0000 (15:38 +0530)]
ath9k: Fix ATH9K_HW_CAP_HT usage

There are a few places where the station's HT capabilities
should be checked instead of ATH9K_HW_CAP_HT, which is a global
feature for the driver. Fix this.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agoMerge tag 'nfc-next-3.9-3' of git://git.kernel.org/pub/scm/linux/kernel/git/sameo...
John W. Linville [Mon, 11 Feb 2013 20:30:47 +0000 (15:30 -0500)]
Merge tag 'nfc-next-3.9-3' of git://git./linux/kernel/git/sameo/nfc-next

Samuel Ortiz <sameo@linux.intel.com> says:

"This 3rd NFC pull request for 3.9 contains a fix for the microread MEI
physical layer support, as the MEI bus API changed.

From the MEI code, we now pass the MEI id back to the driver probe routine,
and we also pass a name and a MEI id table through the mei_bus_driver
structure. A few renames as well like e.g. mei_bus_driver to mei_driver or
mei_bus_client to mei_device in order to be closer to the driver model
practices."

Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agomwl8k: fix band for supported channels
Jonas Gorski [Fri, 8 Feb 2013 15:07:25 +0000 (16:07 +0100)]
mwl8k: fix band for supported channels

The band field for the supported channels were left unpopulated, making
them default to 0 == IEEE80211_BAND_2GHZ, even for the 5GHz channels.

This resulted in null pointer accesses if anything tries to access
wiphy->bands[channel->band] of a 5GHz channel on 5GHz only cards, since
wiphy->bands[2GHZ] is NULL for them (e.g. cfg80211_chandef_usable does).

Example kernel OOPS:

[  665.669993] Unable to handle kernel NULL pointer dereference at virtual address 00000016
[  665.678194] pgd = c6d58000
[  665.680941] [00000016] *pgd=06f8a831, *pte=00000000, *ppte=00000000
[  665.687303] Internal error: Oops: 17 [#1]
(...)
[  666.116373] Backtrace:
[  666.118866] [<bf0368dc>] (cfg80211_chandef_usable+0x0/0x1bc [cfg80211]) from [<bf025e64>] (nl80211_leave_mesh+0x244/0x264 [cfg80211])
[  666.130919]  r7:c6d12100 r6:0000143c r5:c0611c48 r4:c0611b98
[  666.136668] [<bf025d84>] (nl80211_leave_mesh+0x164/0x264 [cfg80211]) from [<bf02634c>] (nl80211_remain_on_channel+0x2a0/0x358 [cfg80211])
[  666.149074]  r7:c6d12000 r6:c6d12000 r5:c6f4f368 r4:00000003
[  666.154814] [<bf0262ec>] (nl80211_remain_on_channel+0x240/0x358 [cfg80211]) from [<bf02ddb0>] (nl80211_set_wiphy+0x264/0x560 [cfg80211])
[  666.167150] [<bf02db4c>] (nl80211_set_wiphy+0x0/0x560 [cfg80211]) from [<c01f94e0>] (genl_rcv_msg+0x1b8/0x1f8)
[  666.177205] [<c01f9328>] (genl_rcv_msg+0x0/0x1f8) from [<c01f89a0>] (netlink_rcv_skb+0x58/0xb4)
[  666.185949] [<c01f8948>] (netlink_rcv_skb+0x0/0xb4) from [<c01f931c>] (genl_rcv+0x20/0x2c)
[  666.194251]  r6:c6f70780 r5:0000002c r4:c6f70780 r3:00000001
[  666.199973] [<c01f92fc>] (genl_rcv+0x0/0x2c) from [<c01f8418>] (netlink_unicast+0x154/0x1f4)
[  666.208449]  r4:c785ea00 r3:c01f92fc
[  666.212057] [<c01f82c4>] (netlink_unicast+0x0/0x1f4) from [<c01f8790>] (netlink_sendmsg+0x230/0x2b0)
[  666.221240] [<c01f8560>] (netlink_sendmsg+0x0/0x2b0) from [<c01cccf8>] (sock_sendmsg+0x90/0xa4)
[  666.229986] [<c01ccc68>] (sock_sendmsg+0x0/0xa4) from [<c01cdcb0>] (__sys_sendmsg+0x290/0x298)
[  666.238637]  r9:00000000 r8:c0611ec8 r6:0000002c r5:c0610000 r4:c0611f64
[  666.245411] [<c01cda20>] (__sys_sendmsg+0x0/0x298) from [<c01cf52c>] (sys_sendmsg+0x44/0x6c)
[  666.253897] [<c01cf4e8>] (sys_sendmsg+0x0/0x6c) from [<c00090a0>] (ret_fast_syscall+0x0/0x2c)
[  666.262460]  r6:00000000 r5:beeff96c r4:00000005

Signed-off-by: Jonas Gorski <jogo@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
8 years agoMerge branch 'for-john' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac80211
John W. Linville [Mon, 11 Feb 2013 19:25:28 +0000 (14:25 -0500)]
Merge branch 'for-john' of git://git./linux/kernel/git/jberg/mac80211

8 years agomac80211: remove unused code to mark AP station authenticated
Johannes Berg [Thu, 7 Feb 2013 16:19:08 +0000 (17:19 +0100)]
mac80211: remove unused code to mark AP station authenticated

When we get to association, the AP station already exists and
is marked authenticated, so moving it into IEEE80211_STA_AUTH
again is a NOP, remove it.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: simplify idle handling
Johannes Berg [Wed, 6 Feb 2013 23:14:51 +0000 (00:14 +0100)]
mac80211: simplify idle handling

Now that we have channel contexts, idle is (pretty
much) equivalent to not having a channel context.
Change the code to use this relation so that there
no longer is a need for a lot of idle recalculate
calls everywhere.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: remove IEEE80211_HW_SCAN_WHILE_IDLE
Johannes Berg [Wed, 6 Feb 2013 22:57:57 +0000 (23:57 +0100)]
mac80211: remove IEEE80211_HW_SCAN_WHILE_IDLE

There are only a few drivers that use HW scan, and
all of those don't need a non-idle transition before
starting the scan -- some don't even care about idle
at all. Remove the flag and code associated with it.

The only driver that really actually needed this is
wl1251 and it can just do it itself in the hw_scan
callback -- implement that.

Acked-by: Luciano Coelho <coelho@ti.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: remove dynamic PS driver interface
Johannes Berg [Wed, 6 Feb 2013 22:07:41 +0000 (23:07 +0100)]
mac80211: remove dynamic PS driver interface

The functions were added for some sort of Bluetooth
coexistence, but aren't used, so remove them again.

Reviewed-by: Luciano Coelho <coelho@ti.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: introduce beacon-only timing data
Johannes Berg [Tue, 5 Feb 2013 16:48:40 +0000 (17:48 +0100)]
mac80211: introduce beacon-only timing data

In order to be able to predict the next DTIM TBTT
in the driver, add the ability to use timing data
from beacons only with the new hardware flag
IEEE80211_HW_TIMING_BEACON_ONLY and the BSS info
value sync_dtim_count which is only valid if the
timing data came from a beacon. The data can only
come from a beacon, and if no beacon was received
before association it is updated later together
with the DTIM count notification.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agocfg80211: move TSF into IEs
Johannes Berg [Tue, 5 Feb 2013 15:54:31 +0000 (16:54 +0100)]
cfg80211: move TSF into IEs

While technically the TSF isn't an IE, it can be
necessary to distinguish between the TSF from a
beacon and a probe response, in particular in
order to know the next DTIM TBTT, as not all APs
are spec compliant wrt. TSF==0 being a DTIM TBTT
and thus the DTIM count needs to be taken into
account as well.

To allow this, move the TSF into the IE struct
so it can be known whence it came.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agocfg80211: remove scan ies NULL check
Johannes Berg [Tue, 5 Feb 2013 15:51:29 +0000 (16:51 +0100)]
cfg80211: remove scan ies NULL check

There's no way scan BSS IEs can be NULL as even
if the allocation fails the frame is discarded.
Remove some code checking for this and document
that it is always non-NULL.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: add vif debugfs driver callbacks
Alexander Bondar [Sat, 22 Dec 2012 08:43:33 +0000 (10:43 +0200)]
mac80211: add vif debugfs driver callbacks

Add debugfs driver callbacks so drivers can add
debugfs entries for interfaces. Note that they
_must_ remove the entries again as add/remove in
the driver doesn't correspond to add/remove in
debugfs; the former is up/down while the latter
is netdev create/destroy.

Signed-off-by: Alexander Bondar <alexander.bondar@intel.com>
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agocfg80211: track hidden SSID networks properly
Johannes Berg [Fri, 1 Feb 2013 01:06:18 +0000 (02:06 +0100)]
cfg80211: track hidden SSID networks properly

Currently, cfg80211 will copy beacon IEs from a previously
received hidden SSID beacon to a probe response entry, if
that entry is created after the beacon entry. However, if
it is the other way around, or if the beacon is updated,
such changes aren't propagated.

Fix this by tracking the relation between the probe
response and beacon BSS structs in this case.

In case drivers have private data stored in a BSS struct
and need access to such data from a beacon entry, cfg80211
now provides the hidden_beacon_bss pointer from the probe
response entry to the beacon entry.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: explicitly copy channels to VLANs where needed
Johannes Berg [Fri, 8 Feb 2013 11:07:44 +0000 (12:07 +0100)]
mac80211: explicitly copy channels to VLANs where needed

Currently the code assigns channel contexts to VLANs
(for use by the TX/RX code) when the AP master gets
its channel context assigned. This works fine, but
in the upcoming radar detection work the VLANs don't
require a channel context (during radar detection)
and assigning one to them anyway causes issues with
locking and also inconsistencies -- a VLAN interface
that is added before radar detection would get the
channel context, while one added during it wouldn't.

Fix these issues moving the channel context copying
to a new explicit operation that will not be used
in the radar detection code.

Acked-by: Simon Wunderlich <siwu@hrz.tu-chemnitz.de>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: fix chandef tracing bug
Johannes Berg [Fri, 8 Feb 2013 20:29:59 +0000 (21:29 +0100)]
mac80211: fix chandef tracing bug

The chandef tracing writes center_freq1 twice, so
that it is always 0 (no driver supports 80+80 yet)
and leaves center_freq2 unset. Fix this mistake.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: fix AP beacon loss messages
Johannes Berg [Fri, 8 Feb 2013 21:59:00 +0000 (22:59 +0100)]
mac80211: fix AP beacon loss messages

The messages currently refer to probe request probes,
but on some devices null data packets will be used
instead. Make the messages more generic.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: protect rx-path with spinlock
Christian Lamparter [Mon, 4 Feb 2013 17:44:44 +0000 (17:44 +0000)]
mac80211: protect rx-path with spinlock

This patch fixes the problem which was discussed in
"mac80211: Fix PN corruption in case of multiple
virtual interface" [1].

Amit Shakya reported a serious issue with my patch:
mac80211: serialize rx path workers" [2]:

In case, ieee80211_rx_handlers processing is going on
for skbs received on one vif and at the same time, rx
aggregation reorder timer expires on another vif then
sta_rx_agg_reorder_timer_expired is invoked and it will
push skbs into the single queue (local->rx_skb_queue).

ieee80211_rx_handlers in the while loop assumes that
the skbs are for the same sdata and sta. This assumption
doesn't hold good in this scenario and the PN gets
corrupted by PN received in other vif's skb, causing
traffic to stop due to PN mismatch."

[1] Message-Id: http://mid.gmane.org/201302041844.44436.chunkeey@googlemail.com
[2] Commit-Id: 24a8fdad35835e8d71f7

Reported-by: Amit Shakya <amit.shakya@stericsson.com>
Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: Add the DS Params for mesh to every band
Emanuel Taube [Wed, 6 Feb 2013 13:17:17 +0000 (14:17 +0100)]
mac80211: Add the DS Params for mesh to every band

There seems to be no reason, why it has to be limited to 2.4 GHz.

Signed-off-by: Emanuel Taube <emanuel.taube@gmail.com>
[remove 'local' variable]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agomac80211: fix mesh sta teardown
Thomas Pedersen [Wed, 6 Feb 2013 18:17:21 +0000 (10:17 -0800)]
mac80211: fix mesh sta teardown

The patch "mac80211: clean up mesh sta allocation warning"
moved some mesh initialization into a path which is only
called when the kernel handles peering. This causes a hang
when mac80211 tries to clean up a userspace-allocated
station entry and delete a timer which has never been
initialized.

To avoid this, only do any mesh sta peering teardown if
the kernel is actually handling it.

The same is true when quiescing before suspend.

Signed-off-by: Thomas Pedersen <thomas@cozybit.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
8 years agowireless: fix kernel-doc
Johannes Berg [Fri, 8 Feb 2013 08:06:36 +0000 (09:06 +0100)]
wireless: fix kernel-doc

Fix most kernel-doc warnings, for some reason it
seems to have issues with __aligned, don't remove
the documentation entries it considers to be in
excess due to that.

Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>