firefly-linux-kernel-4.4.55.git
9 years agoath9k: set ATH_OP_INVALID before disabling hardware
Felix Fietkau [Sun, 30 Nov 2014 20:58:31 +0000 (21:58 +0100)]
ath9k: set ATH_OP_INVALID before disabling hardware

Closes another small IRQ handler race

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agoath9k: prevent early IRQs from accessing hardware
Felix Fietkau [Sun, 30 Nov 2014 20:58:30 +0000 (21:58 +0100)]
ath9k: prevent early IRQs from accessing hardware

IRQs are suppressed if ah == NULL and ATH_OP_INVALID being set in
common->op_flags. Close a short time window between those two.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agoath5k: fix hardware queue index assignment
Felix Fietkau [Sun, 30 Nov 2014 20:52:57 +0000 (21:52 +0100)]
ath5k: fix hardware queue index assignment

Like with ath9k, ath5k queues also need to be ordered by priority.
queue_info->tqi_subtype already contains the correct index, so use it
instead of relying on the order of ath5k_hw_setup_tx_queue calls.

Cc: stable@vger.kernel.org
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agoath9k: fix BE/BK queue order
Felix Fietkau [Sun, 30 Nov 2014 19:38:41 +0000 (20:38 +0100)]
ath9k: fix BE/BK queue order

Hardware queues are ordered by priority. Use queue index 0 for BK, which
has lower priority than BE.

Cc: stable@vger.kernel.org
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agoath9k_hw: fix hardware queue allocation
Felix Fietkau [Sun, 30 Nov 2014 19:38:40 +0000 (20:38 +0100)]
ath9k_hw: fix hardware queue allocation

The driver passes the desired hardware queue index for a WMM data queue
in qinfo->tqi_subtype. This was ignored in ath9k_hw_setuptxqueue, which
instead relied on the order in which the function is called.

Cc: stable@vger.kernel.org
Reported-by: Hubert Feurstein <h.feurstein@gmail.com>
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agobrcmsmac: NULL dereferences in brcms_c_detach_mfree()
Dan Carpenter [Fri, 28 Nov 2014 09:43:40 +0000 (12:43 +0300)]
brcmsmac: NULL dereferences in brcms_c_detach_mfree()

The brcms_c_attach_malloc() function can call this with a NULL
"wlc->corestate" or "wlc->hw".

Also I threw in a bonus cleanup by deleting an obvious comment and a
no-op NULL assignment.  :)

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agoRevert "rt2x00: Endless loop on hub port power down"
Stanislaw Gruszka [Wed, 26 Nov 2014 14:29:16 +0000 (15:29 +0100)]
Revert "rt2x00: Endless loop on hub port power down"

This reverts commit 2ad69ac5976191e9bb7dc4044204a504653ad1bb. It
causes wireless device disappear when we get -EPROTO error form USB
request. I encounter such situation occasionally when resume form
suspend with RT3070 adapter:

[  289.619985] ieee80211 phy0: rt2x00usb_vendor_request: Error - Vendor Request 0x06 failed for offset 0x0404 with error -71
[  289.639368] ieee80211 phy0: rt2800_wait_bbp_ready: Error - BBP register access failed, aborting
[  289.639374] ieee80211 phy0: rt2800usb_set_device_state: Error - Device failed to enter state 4 (-5)

Without the patch, except printing error, device works just fine after
resume.

Currently after timeouts and REGISTER_BUSY_COUNT tuning, we should
not have any "endless loop", though we can wait quite long when driver
is trying to communicate with the device through non functioning USB
connection. Generally the problem that commit 2ad69ac597619 solves
is kinda artificial.

Cc: Richard Genoud <richard.genoud@gmail.com>
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agort2x00: change REGISTER_TIMEOUT
Stanislaw Gruszka [Wed, 26 Nov 2014 14:29:15 +0000 (15:29 +0100)]
rt2x00: change REGISTER_TIMEOUT

Waiting 500ms for register access is too long, decrease this value
to 100ms.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agort2x00: change REGISTER_BUSY_COUNT for USB
Stanislaw Gruszka [Wed, 26 Nov 2014 14:29:14 +0000 (15:29 +0100)]
rt2x00: change REGISTER_BUSY_COUNT for USB

Because of delays on USB we do not have to iterate so many times on
USB hardware when waiting for H/W register become valid.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agort2x00: use timeout in rt2x00usb_vendor_request
Stanislaw Gruszka [Wed, 26 Nov 2014 14:29:13 +0000 (15:29 +0100)]
rt2x00: use timeout in rt2x00usb_vendor_request

Use provided timeout value in rt2x00usb_vendor_request() instead
of iterating REGISTER_BUSY_COUNT times.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agobrcmsmac: extend hardware info shown in debugfs
Arend van Spriel [Tue, 25 Nov 2014 19:55:54 +0000 (20:55 +0100)]
brcmsmac: extend hardware info shown in debugfs

The hardware info now also include radio and phy information, which
can be helpful in debugging issues.

Reviewed-by: Hante Meuleman <meuleman@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agobrcmutil: add helper function to format board revision
Arend van Spriel [Tue, 25 Nov 2014 19:55:53 +0000 (20:55 +0100)]
brcmutil: add helper function to format board revision

The board revision that is available in hardware can be translated
so it matches the labelling on the board. This is accomplished by
this helper function.

Reviewed-by: Hante Meuleman <meuleman@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agort2800: calculate tx power temperature compensation on selected chips
Stanislaw Gruszka [Tue, 25 Nov 2014 14:17:29 +0000 (15:17 +0100)]
rt2800: calculate tx power temperature compensation on selected chips

Currently implemented temperature compensation is only valid on some of
supported chips. Other chips do not need temperature compensation or
need different way to do this (not yet implemented in the rt2800
driver). Trying to do run rt2800_get_gain_calibration_delta() when this
is not appropriate on particular chip gives bogus result of TX power
and can make connection unstable.

This is follow up to commit 8c8d2017ba25c510ddf093419048460db1109bc4
"rt2800: fix RT5390 & RT3290 TX power settings regression". On that
commit we avoid setting BBP_R1 register, but the real problem is wrong
temperature compensation calculation.

Reported-and-tested-by: Ronald Wahl <ronald.wahl@raritan.com>
Debugged-by: Ronald Wahl <ronald.wahl@raritan.com>
Cc: Mike Romberg <mike-romberg@comcast.net>
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agowil6210: Fix potential memory leaks on error paths
Lino Sanfilippo [Fri, 28 Nov 2014 01:47:19 +0000 (02:47 +0100)]
wil6210: Fix potential memory leaks on error paths

Fix missing memory deallocation on error paths in wil_write_file_wmi()
and wil_write_file_txmgmt().

Reported-by: Ahmed Tamrawi <ahmedtamrawi@gmail.com>
Signed-off-by: Lino Sanfilippo <LinoSanfilippo@gmx.de>
Signed-off-by: Vladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agoMerge branch 'for-linville' of git://github.com/kvalo/ath
John W. Linville [Mon, 1 Dec 2014 20:51:31 +0000 (15:51 -0500)]
Merge branch 'for-linville' of git://github.com/kvalo/ath

9 years agoMerge branch 'for-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/bluetoot...
John W. Linville [Mon, 1 Dec 2014 20:49:58 +0000 (15:49 -0500)]
Merge branch 'for-upstream' of git://git./linux/kernel/git/bluetooth/bluetooth-next

9 years agoath10k: simplify rx ring size/fill calculation
Michal Kazior [Thu, 27 Nov 2014 10:12:43 +0000 (11:12 +0100)]
ath10k: simplify rx ring size/fill calculation

Don't bother with fancy arithmetic and just
hardcode the final values.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: prevent pci tx/rx starvation
Michal Kazior [Thu, 27 Nov 2014 10:09:38 +0000 (11:09 +0100)]
ath10k: prevent pci tx/rx starvation

In theory it was possible to starve the system if
a tx/rx handler could implicitly trigger more
tx/rx pci events.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: remove transfer_id from ath10k_hif_cb::tx_completion
Michal Kazior [Thu, 27 Nov 2014 10:09:37 +0000 (11:09 +0100)]
ath10k: remove transfer_id from ath10k_hif_cb::tx_completion

Pass the eid argument via skbuff control buffer.
This will make it possible to work with queues of
HTC event buffers.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: remove unused callback argument from struct ath10k_hif_cb::rx_completion
Michal Kazior [Thu, 27 Nov 2014 10:09:36 +0000 (11:09 +0100)]
ath10k: remove unused callback argument from struct ath10k_hif_cb::rx_completion

This wasn't used since forever and there are no
plans on using it.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: make wmi service bitmap non-debug
Michal Kazior [Thu, 27 Nov 2014 09:11:17 +0000 (10:11 +0100)]
ath10k: make wmi service bitmap non-debug

The service bitmap can be used to determine
firmware capabilities.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: add sanity checks for service bmap parsing
Michal Kazior [Thu, 27 Nov 2014 09:11:16 +0000 (10:11 +0100)]
ath10k: add sanity checks for service bmap parsing

This shouldn't really happen but take into account
the original service bitmap length when mapping
service ids.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: fix wmi svc bitmap dbg print
Michal Kazior [Thu, 27 Nov 2014 09:11:15 +0000 (10:11 +0100)]
ath10k: fix wmi svc bitmap dbg print

The dump print uses sizeof() but since service_map
was a mere pointer the dump was too short.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: add pointer constness to traces
Michal Kazior [Thu, 27 Nov 2014 07:42:11 +0000 (08:42 +0100)]
ath10k: add pointer constness to traces

Traces shouldn't modified passed data. This will
make it possible to pass const arguments to
traces.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: do not limit RTS threshold value to 2347
Vivek Natarajan [Wed, 26 Nov 2014 07:06:12 +0000 (09:06 +0200)]
ath10k: do not limit RTS threshold value to 2347

Increase the rts threshold from the legacy value of 2347 to support higher
threshold limit.

Signed-off-by: Vivek Natarajan <nataraja@qti.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoBluetooth: Automatically flushable packets aren't allowed on LE links
Steven Walter [Wed, 19 Nov 2014 14:41:17 +0000 (09:41 -0500)]
Bluetooth: Automatically flushable packets aren't allowed on LE links

The Bluetooth spec states that automatically flushable packets may not
be sent over a LE-U link.

Signed-off-by: Steven Walter <stevenrwalter@gmail.com>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
9 years agoBluetooth: ath3k: Add support of MCI 13d3:3408 bt device
Dmitry Tunin [Tue, 25 Nov 2014 17:19:52 +0000 (20:19 +0300)]
Bluetooth: ath3k: Add support of MCI 13d3:3408 bt device

Add support for Bluetooth MCI WB335 (AR9565) Wi-Fi+bt module. This
Bluetooth module requires loading patch and sysconfig by ath3k driver.

T:  Bus=01 Lev=02 Prnt=03 Port=00 Cnt=01 Dev#= 20 Spd=12   MxCh= 0
D:  Ver= 1.10 Cls=e0(wlcon) Sub=01 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=13d3 ProdID=3408 Rev= 0.02
C:* #Ifs= 2 Cfg#= 1 Atr=e0 MxPwr=100mA
A:  FirstIf#= 0 IfCount= 2 Cls=e0(wlcon) Sub=01 Prot=01
I:* If#= 0 Alt= 0 #EPs= 3 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=81(I) Atr=03(Int.) MxPS=  16 Ivl=1ms
E:  Ad=82(I) Atr=02(Bulk) MxPS=  64 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS=  64 Ivl=0ms
I:* If#= 1 Alt= 0 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=83(I) Atr=01(Isoc) MxPS=   0 Ivl=1ms
E:  Ad=03(O) Atr=01(Isoc) MxPS=   0 Ivl=1ms
I:  If#= 1 Alt= 1 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=83(I) Atr=01(Isoc) MxPS=   9 Ivl=1ms
E:  Ad=03(O) Atr=01(Isoc) MxPS=   9 Ivl=1ms
I:  If#= 1 Alt= 2 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=83(I) Atr=01(Isoc) MxPS=  17 Ivl=1ms
E:  Ad=03(O) Atr=01(Isoc) MxPS=  17 Ivl=1ms
I:  If#= 1 Alt= 3 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=83(I) Atr=01(Isoc) MxPS=  25 Ivl=1ms
E:  Ad=03(O) Atr=01(Isoc) MxPS=  25 Ivl=1ms
I:  If#= 1 Alt= 4 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=83(I) Atr=01(Isoc) MxPS=  33 Ivl=1ms
E:  Ad=03(O) Atr=01(Isoc) MxPS=  33 Ivl=1ms
I:  If#= 1 Alt= 5 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=83(I) Atr=01(Isoc) MxPS=  49 Ivl=1ms
E:  Ad=03(O) Atr=01(Isoc) MxPS=  49 Ivl=1ms

Signed-off-by: Dmitry Tunin <hanipouspilot@gmail.com>
Signed-off-by: Johan Hedberg <johan.hedberg@intel.com>
Cc: stable@vger.kernel.org
9 years agoath10k: fix station count enforcement
Michal Kazior [Tue, 25 Nov 2014 14:16:05 +0000 (15:16 +0100)]
ath10k: fix station count enforcement

The number of peers isn't directly translatable to
the number of stations because ath10k needs to
reserve a few extra peers for special cases like
multi-vif concurrency.

The previous limit was 126 and 15 stations in AP
mode for 10.x and main firmware branches
respectively. The limit is now 128 and 16 which
was the original intention.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: clean up num_peers locking
Michal Kazior [Tue, 25 Nov 2014 14:16:04 +0000 (15:16 +0100)]
ath10k: clean up num_peers locking

The var was supposed to be protected by data_lock
but it wasn't so in all instances. It's actually
not necessary to have a spinlock protected
num_peers so drop it.

All instances of num_peers are already within
conf_mutex sections so use that.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: add missing goto
Michal Kazior [Tue, 25 Nov 2014 14:16:03 +0000 (15:16 +0100)]
ath10k: add missing goto

This prevents warning spamming if peer creation
fails during sta_state in some cases.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: add memory dump debugfs interface
Yanbo Li [Tue, 25 Nov 2014 10:24:48 +0000 (12:24 +0200)]
ath10k: add memory dump debugfs interface

Add mem_val debugfs file for dumping the firmware (target) memory and also for
writing to the memory. The firmware memory is accessed through one file which
uses position of the file as the firmware memory address. For example, with dd
use skip parameter for the address.

Beucase target memory width is 32 bits it's strongly recommended to use
blocksize divisable with 4 when using this interface. For example, when using
dd use bs=4 to set the block size to 4 and remember to divide both count and
skip values with four.

To read 4 kB chunk from address 0x400000:

dd if=mem_value bs=4 count=1024 skip=1048576 | xxd -g1

To write value 0x01020304 to address 0x400400:

echo 0x01020304 | xxd -r | dd of=mem_value bs=4 seek=1048832

To read 4 KB chunk of memory and then write back after edit:

dd if=mem_value of=tmp.bin bs=4 count=1024 skip=1048576
emacs tmp.bin
dd if=tmp.bin of=mem_value bs=4 count=1024 seek=1048576

Signed-off-by: Yanbo Li <yanbol@qti.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: add register access debugfs interface
Yanbo Li [Tue, 25 Nov 2014 10:24:33 +0000 (12:24 +0200)]
ath10k: add register access debugfs interface

Debugfs files reg_addr and reg_val are used for reading and writing to the
firmware (target) registers. reg_addr contains the address to be accessed,
which also needs to be set first, and reg_value is when used for reading and
writing the actual value in ASCII.

To read a value from the firmware register 0x100000:

# echo 0x100000 > reg_addr
# cat reg_value
0x00100000:0x000002d3

To write value 0x2400 to address 0x100000:

# echo 0x100000 > reg_addr
# echo  0x2400 > reg_value
#

Signed-off-by: Yanbo Li <yanbol@qti.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: fix bug reported by lockdep
Sujith Manoharan [Tue, 25 Nov 2014 06:17:00 +0000 (11:47 +0530)]
ath10k: fix bug reported by lockdep

ath10k_tx_wep_key_work() acquires conf_mutex, so
cancelling it when conf_mutex is already taken
in ath10k_remove_interface() is incorrect, so
move it outside the lock.

Snippet from the lockdep report:

kernel: ======================================================
kernel: [ INFO: possible circular locking dependency detected ]
kernel: 3.18.0-rc5-wl-debug #34 Tainted: G           O
kernel: -------------------------------------------------------
kernel: hostapd/451 is trying to acquire lock:
kernel:  ((&arvif->wep_key_work)){+.+...}, at: [<ffffffff810872d5>] flush_work+0x5/0x290
kernel: but task is already holding lock:
kernel:  (&ar->conf_mutex){+.+.+.}, at: [<ffffffffa0b99f00>] ath10k_remove_interface+0x40/0x290 [ath10k_core]
kernel: which lock already depends on the new lock.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: fix locking for WEP keys
Sujith Manoharan [Tue, 25 Nov 2014 06:16:59 +0000 (11:46 +0530)]
ath10k: fix locking for WEP keys

peer->keys needs to be protected by data_lock
since it is also accessed from the WMI path.

Both install() and clear() routines for peer
keys modify the key contents, so use the data_lock
to avoid races.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: fix shared WEP
Sujith Manoharan [Tue, 25 Nov 2014 06:16:58 +0000 (11:46 +0530)]
ath10k: fix shared WEP

When static keys are used in shared WEP, when a
station is associated, message 3 is sent with an
encrypted payload. But, for subsequent
authentications that are triggered without a
deauth, the auth frame is decrypted by the HW.

To handle this, check if the WEP keys have already
been set for the peer and if so, mark the
frame as decrypted. This scenario can happen
when a station changes its default TX key and initiates
a new authentication sequence.

Signed-off-by: Sujith Manoharan <c_manoha@qca.qualcomm.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: don't drop corrupted mgmt frames
Michal Kazior [Mon, 24 Nov 2014 14:34:08 +0000 (15:34 +0100)]
ath10k: don't drop corrupted mgmt frames

Some firmware revisions don't seem to deilver
management frames with FCS error via WMI so narrow
down the HTT rule to not drop corrupted management
frames.

This basically increases number of frames ath10k
reports while sniffing.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: fix offchannel cancel failures
Michal Kazior [Mon, 24 Nov 2014 13:58:33 +0000 (14:58 +0100)]
ath10k: fix offchannel cancel failures

When mac80211 requests driver to cancel a hw roc
the driver must not call the expired() callback or
else roc will fail in some cases depending on how
things get scheduled.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: make hw roc more reliable
Michal Kazior [Mon, 24 Nov 2014 13:58:32 +0000 (14:58 +0100)]
ath10k: make hw roc more reliable

With P2P concurrency requested hw roc duration
time can be very small. Some firmware revisions
refuse scan requests with too small channel dwell
time.

This prevents messages like, e.g. with connected
STA vif and performing P2P Find:

 ath10k_pci 0000:00:05.0: failed to switch to channel for roc scan
 ieee80211 phy3: failed to start next HW ROC (-110)

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: fix offchan reliability
Michal Kazior [Mon, 24 Nov 2014 13:58:31 +0000 (14:58 +0100)]
ath10k: fix offchan reliability

New firmware revisions don't need peer creation
when doing offchannel tx. Earlier revisions would
queue and never release frames without a peer.

This prevent new firmware revisions from stopping
replenishing wmi-htc tx credits and improves
reliability of offchannel tx which would sometimes
silently fail.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: remove extra_tx_headroom
Michal Kazior [Mon, 24 Nov 2014 14:04:46 +0000 (15:04 +0100)]
ath10k: remove extra_tx_headroom

Comment was out-of-date. The headroom is no longer
necessary because HTT Tx fragment list is stored
in dma pool item associated with each sk_buff.

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: use configured nss instead of max nss
Ben Greear [Mon, 24 Nov 2014 14:22:10 +0000 (16:22 +0200)]
ath10k: use configured nss instead of max nss

When re-associating a station, the nss was set back to
maximum value even if user had configured small number
of tx chains.  So, pay attention to user's config in
this case as well.

Signed-off-by: Ben Greear <greearb@candelatech.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: apply chainmask settings to vdev on creation
Ben Greear [Mon, 24 Nov 2014 14:22:10 +0000 (16:22 +0200)]
ath10k: apply chainmask settings to vdev on creation

It appears it takes more than just setting the
hardware's chainmask to make things work well.  Without
this patch, a vdev would only use 1x1 rates when chainmask
was set to 0x3.

Setting the 'nss' (number of spatial streams) on the vdev
helps the firmware's rate-control algorithm work properly.

Signed-off-by: Ben Greear <greearb@candelatech.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agomac802154: remove unnecessary if statement
Varka Bhadram [Tue, 25 Nov 2014 11:04:43 +0000 (16:34 +0530)]
mac802154: remove unnecessary if statement

Removes unnecessary if statement check for net device. Error check performed
after alloc_netdev().

ndev = alloc_netdev(sizeof(*sdata) + local->hw.vif_data_size, name,
    NET_NAME_UNKNOWN, ieee802154_if_setup);
        if (!ndev)
                return ERR_PTR(-ENOMEM);
..

Signed-off-by: Varka Bhadram <varkab@cdac.in>
Acked-by: Alexander Aring <alex.aring@gmail.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
9 years agoieee802154: fix spelling mistakes
Varka Bhadram [Tue, 25 Nov 2014 04:34:57 +0000 (10:04 +0530)]
ieee802154: fix spelling mistakes

Signed-off-by: Varka Bhadram <varkab@cdac.in>
Acked-by: Alexander Aring <alex.aring@gmail.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
9 years agomwifiex: add Tx status support for ACTION frames
Amitkumar Karwar [Tue, 25 Nov 2014 14:43:06 +0000 (06:43 -0800)]
mwifiex: add Tx status support for ACTION frames

ACK status (0/1) for ACTION frames is informed to cfg80211. We
will extend existing logic used for EAPOL frames. The cfg80211
API is different here. Also, we need to explicitly free cloned
skb.

Signed-off-by: Cathy Luo <cluo@marvell.com>
Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agomwifiex: add Tx status support for EAPOL packets
Amitkumar Karwar [Tue, 25 Nov 2014 14:43:05 +0000 (06:43 -0800)]
mwifiex: add Tx status support for EAPOL packets

Firmware notifies the driver through event if EAPOL data packet
has been acked or not. We will inform this status to userspace
listening on a socket.

Signed-off-by: Cathy Luo <cluo@marvell.com>
Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agomwifiex: skip delay main work logic for USB interface.
Amitkumar Karwar [Tue, 25 Nov 2014 14:43:04 +0000 (06:43 -0800)]
mwifiex: skip delay main work logic for USB interface.

We had introduced delay main work logic to avoid processing
interrupts when Rx pending packet count reaches high threshold.
interrupt processing is restarted later when packet count
reduces lower threashold. This helped to reduce unnecessary
overhead and improve throughput for SD and PCIe chipsets.

As there are no interrupts for USB, we will skip this logic for
USB chipsets.

Signed-off-by: Cathy Luo <cluo@marvell.com>
Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agomwifiex: fix scan problem on big endian platforms
Amitkumar Karwar [Tue, 25 Nov 2014 14:43:03 +0000 (06:43 -0800)]
mwifiex: fix scan problem on big endian platforms

This patch adds missing endian conversion for beacon size while
processing scan response.

Reported-by: Daniel Mosquera <daniel.mosquera@ctag.com>
Tested-by: Daniel Mosquera <daniel.mosquera@ctag.com>
Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agomwifiex: fix sparse warning
Amitkumar Karwar [Tue, 25 Nov 2014 14:43:02 +0000 (06:43 -0800)]
mwifiex: fix sparse warning

This patch fixes following sparse warnings:

drivers/net/wireless/mwifiex/util.c:152:19: warning: cast from restricted __le16
drivers/net/wireless/mwifiex/util.c:152:19: warning: restricted __le16 degrades to integer

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agoath9k: add TPC capability to TX descriptor path
Lorenzo Bianconi [Mon, 24 Nov 2014 23:21:41 +0000 (00:21 +0100)]
ath9k: add TPC capability to TX descriptor path

Add TPC capability to TX descriptor path. Cap per-packet TX power according to
TX power per-rate tables. Currently TPC is supported just by AR9003 based chips

Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agoath9k: add TX power per-rate tables
Lorenzo Bianconi [Mon, 24 Nov 2014 23:21:40 +0000 (00:21 +0100)]
ath9k: add TX power per-rate tables

Add TX power per-rate tables for different MIMO modes (e.g STBC) in order to
cap the maximum TX power value per-rate in the TX descriptor path.
Cap TX power for self generated frames (ACK, RTS/CTS).
Currently TPC is supported just by AR9003 based chips

Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agobrcmfmac: correct .disconnect() callback while connecting
Arend van Spriel [Thu, 20 Nov 2014 21:27:02 +0000 (22:27 +0100)]
brcmfmac: correct .disconnect() callback while connecting

When the driver has sent a join iovar to the firmware it waits
for the events to report result of the connection. However, the
wpa_supplicant will request a .disconnect() after a timeout. So
upon calling .disconnect() the interface state may still be
CONNECTING. Clear the CONNECTING bit as well.

Reviewed-by: Hante Meuleman <meuleman@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agonet: wireless: rtlwifi: rtl8192ee: Fix compilation of the driver
Andreas Ruprecht [Mon, 24 Nov 2014 18:34:32 +0000 (19:34 +0100)]
net: wireless: rtlwifi: rtl8192ee: Fix compilation of the driver

In the Makefile for this driver, the wrong Kconfig option is used
to trigger the compilation of the object file. This leads to the
driver only being included into the kernel when both CONFIG_RTL8821AE
and CONFIG_RTL8192AE are set to "y".

Signed-off-by: Andreas Ruprecht <rupran@einserver.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agonet: wireless: rtlwifi: Do not always include drivers in obj-m
Andreas Ruprecht [Mon, 24 Nov 2014 18:34:31 +0000 (19:34 +0100)]
net: wireless: rtlwifi: Do not always include drivers in obj-m

In four of the rtlwifi drivers, the Makefile contains superfluous
statements indicating the compilation of the driver as an LKM
regardless of the corresponding Kconfig option.

If the corresponding option is set to 'y', the build system will then
see the object file in obj-m and obj-y, which leads to a compilation
as a built-in only. Even though this leads to the desired behavior,
the unconditional appearance in obj-m is confusing for someone reading
the Makefile.

This patch removes the superfluous Makefile statements.

Signed-off-by: Andreas Ruprecht <rupran@einserver.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agoath: Fix a false radar detection pattern
Vivek Natarajan [Mon, 24 Nov 2014 11:56:22 +0000 (17:26 +0530)]
ath: Fix a false radar detection pattern

For FCC and JP, in one of the radar patterns, PPB and PRF seems to be
interchanged leading to frequent incorrect radar detections.

Signed-off-by: Vivek Natarajan <nataraja@qti.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agowireless/p54: Remove duplicated net2280 header
Ricardo Ribalda Delgado [Mon, 24 Nov 2014 10:19:51 +0000 (11:19 +0100)]
wireless/p54: Remove duplicated net2280 header

The usb gadget driver net2280 has exported a header file with the
register definition of the net2280 chip.

Remove the custom/duplicated header file in favor of that header file
in include/linux

Signed-off-by: Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agortlwifi: update RCR register in rtl_op_configure_filter()
Lorenzo Bianconi [Sat, 22 Nov 2014 11:18:26 +0000 (12:18 +0100)]
rtlwifi: update RCR register in rtl_op_configure_filter()

Reconfigure RCR register in rtl_op_configure_filter() in order to apply RX
filter configuration

Signed-off-by: Lorenzo Bianconi <lorenzo.bianconi83@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agobrcmfmac: fix static checker warning in pmklist handling
Arend van Spriel [Thu, 20 Nov 2014 21:26:59 +0000 (22:26 +0100)]
brcmfmac: fix static checker warning in pmklist handling

The patch fixes a static checker warning:

   drivers/net/wireless/brcm80211/brcmfmac/cfg80211.c:2965
      brcmf_cfg80211_set_pmksa()
warn: can 'pmkid_len' be negative?

The answer to the question above is likely no so changing its
type to unsigned is sufficient.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Hante Meuleman <meuleman@broadcom.com>
Reviewed-by: Franky (Zhenhui) Lin <frankyl@broadcom.com>
Reviewed-by: Daniel (Deognyoun) Kim <dekim@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agoMerge tag 'iwlwifi-next-for-john-2014-11-24' of git://git.kernel.org/pub/scm/linux...
John W. Linville [Mon, 24 Nov 2014 18:59:37 +0000 (13:59 -0500)]
Merge tag 'iwlwifi-next-for-john-2014-11-24' of git://git./linux/kernel/git/iwlwifi/iwlwifi-next

Emmanuel Grumbach <egrumbach@gmail.com> says:

"Major works are CSA and TDLS. On top of that I have a new
firmware API for scan and a few rate control improvements.
Johannes find a few tricks to improve our CPU utilization
and adds support for a new spin of 7265 called 7265D.
Along with this a few random things that don't stand out."

Signed-off-by: John W. Linville <linville@tuxdriver.com>
9 years agoath10k: add ATH10K_DBG_WMI_PRINT debug level
Ben Greear [Tue, 23 Sep 2014 21:17:19 +0000 (14:17 -0700)]
ath10k: add ATH10K_DBG_WMI_PRINT debug level

There are not many of these messages producted by the
firmware, but they are generally fairly useful, so make it easy to print them
with a separate debug level.

kvalo: fix commit log, rename debug level

Signed-off-by: Ben Greear <greearb@candelatech.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoath10k: don't rebuild all the time
Johannes Berg [Fri, 21 Nov 2014 16:58:49 +0000 (18:58 +0200)]
ath10k: don't rebuild all the time

There are better ways to get the kernel information, use the
utsname and omit the version code entirely since it's duplicate.
The version magic is rather useless anyway

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoiwlwifi: update the secure mem space and for the CPUs
Emmanuel Grumbach [Sun, 23 Nov 2014 19:39:50 +0000 (21:39 +0200)]
iwlwifi: update the secure mem space and for the CPUs

Also update the timeout for the data verification.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: sdio: new SDIO card id for 4165 series
Oren Givon [Tue, 14 Oct 2014 08:26:52 +0000 (11:26 +0300)]
iwlwifi: sdio: new SDIO card id for 4165 series

Add a new SDIO card id intended for the new 4165 series.

Signed-off-by: Oren Givon <oren.givon@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: declare d0i3 support for IWL_DEVICE_8000
Eliad Peller [Wed, 25 Jun 2014 15:09:58 +0000 (18:09 +0300)]
iwlwifi: declare d0i3 support for IWL_DEVICE_8000

Signed-off-by: Eliad Peller <eliadx.peller@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: disconnect TDLS peers before channel switch
Arik Nemtsov [Sun, 14 Sep 2014 16:13:54 +0000 (19:13 +0300)]
iwlwifi: mvm: disconnect TDLS peers before channel switch

In case of channel switch, we need to teardown the TDLS peers.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: declare support for VHT BF info in radiotap
Eyal Shapira [Tue, 18 Nov 2014 14:43:55 +0000 (16:43 +0200)]
iwlwifi: mvm: declare support for VHT BF info in radiotap

The driver knows whether an rx frame was beamformed and marks
it in the radiotap VHT flags. However it should also declare
that it knows to extract this info otherwise this gets discarded
by sniffers like Wireshark.

Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: add support to MFUART loading notification
Chaya Rachel Ivgy [Sat, 15 Nov 2014 19:08:29 +0000 (21:08 +0200)]
iwlwifi: mvm: add support to MFUART loading notification

Add support to MFUART loading notification (print notification
data with IWL_DEBUG_INFO)

Signed-off-by: Chaya Rachel Ivgy <chaya.rachel.ivgi@intel.com>
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: call iwl_trans_suspend/resume
Eliad Peller [Tue, 30 Sep 2014 07:02:13 +0000 (10:02 +0300)]
iwlwifi: mvm: call iwl_trans_suspend/resume

call the appropriate iwl_trans_suspend/resume ops
on iwl_mvm_suspend/resume calls.

Signed-off-by: Eliad Peller <eliadx.peller@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: trans: add suspend/resume ops
Eliad Peller [Wed, 1 Oct 2014 12:21:28 +0000 (15:21 +0300)]
iwlwifi: trans: add suspend/resume ops

Add suspend/resume trans ops that will be called from
mac80211's suspend/resume ops.

Signed-off-by: Eliad Peller <eliadx.peller@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: treat netdetect wake up separately
Luciano Coelho [Thu, 20 Nov 2014 06:59:51 +0000 (08:59 +0200)]
iwlwifi: mvm: treat netdetect wake up separately

When the device wakes up due to netdetect, we need to query different
things from the firmware than when it wakes up with a normal WoWLAN.
To make this easier, separate the netdetect wake up handling from the
rest.  For now, we don't send netdetect as a wake up reason, treating
it as a non-wireless wake up.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: refactor iwl_mvm_query_wakeup_reasons()
Luciano Coelho [Thu, 20 Nov 2014 08:14:44 +0000 (10:14 +0200)]
iwlwifi: mvm: refactor iwl_mvm_query_wakeup_reasons()

Refactor the iwl_mvm_query_wakeup_reasons() function to split the part
that gets the firmware status from the part that sets up the WoWLAN
status.  This will allow netdetect to reuse the code.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: refactor wowlan and netdetect configuration when suspending
Luciano Coelho [Wed, 19 Nov 2014 20:35:37 +0000 (22:35 +0200)]
iwlwifi: mvm: refactor wowlan and netdetect configuration when suspending

We need to send a WOWLAN_CONFIGURATION command also for netdetect and
configure the rfkill release trigger if needed.  To do so, refactor
the code that configures wowlan and netdetect when suspending and send
the WOWLAN_CONFIGURATION command also for net_detect.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: add support for net detect
Luciano Coelho [Fri, 24 Oct 2014 07:39:51 +0000 (10:39 +0300)]
iwlwifi: mvm: add support for net detect

Add the net detect WoWLAN flag to indicate support and use the
nd_config from the WoWLAN configuration to start net detect, if it is
set.  The WoWLAN configuration takes precedence over the debugfs
configuration.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: protect session during CSA
Luciano Coelho [Mon, 10 Nov 2014 09:10:21 +0000 (11:10 +0200)]
iwlwifi: mvm: protect session during CSA

When another vif is also running during a channel switch, we need to
use a session protection when we move to the new channel, so that we
don't miss the beacons.  Without this, sometimes the other vif
repeatedly gets time exactly when we should be hearing the beacons,
preventing channel switch from completing.  Adding a session
protection that lasts from the moment the channel changes until 2
TBTTs later, ensures that we will hear the beacons on the destination
channel.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
9 years agoiwlwifi: mvm: clear TE data if CSA time event fails to start
Luciano Coelho [Mon, 10 Nov 2014 09:10:20 +0000 (11:10 +0200)]
iwlwifi: mvm: clear TE data if CSA time event fails to start

If setting the CSA time event fails, we must clear the TE data,
otherwise we'll try to remove it when, for instance, a disconnection
occurs, causing a SYSASSERT.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
9 years agoiwlwifi: mvm: disable beacon filtering during CSA
Luciano Coelho [Mon, 10 Nov 2014 09:10:18 +0000 (11:10 +0200)]
iwlwifi: mvm: disable beacon filtering during CSA

After a channel switch, transmission on the new channel is only
started once we see a beacon on it.  Thus, beacon filtering needs to
be disabled during channel switch so that mac80211 receives this
beacon and finishes the process.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
9 years agoiwlwifi: mvm: add a channel_switch op to bypass mac80211 timer
Luciano Coelho [Mon, 10 Nov 2014 09:10:17 +0000 (11:10 +0200)]
iwlwifi: mvm: add a channel_switch op to bypass mac80211 timer

We need to call ieee80211_chswitch_done() ourselves just when the
absence TE started, so we perform the actual context switch early
enough.  To do so, add a dummy channel_switch op, which will cause
mac80211 to skip the countdown timer and allow us to call
ieee80211_chswitch_done() to complete the operation.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
9 years agoiwlwifi: mvm: finalize on post_switch instead of unassign
Luciano Coelho [Mon, 10 Nov 2014 09:10:16 +0000 (11:10 +0200)]
iwlwifi: mvm: finalize on post_switch instead of unassign

Now that mac80211 waits for the first beacon on the new channel before
calling post_channel_switch, the reconfiguration of the firmware
should be done in the post_channel_switch operation instead of when
assigning the vif to the new context.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
9 years agoiwlwifi: mvm: schedule CSA time event a bit before beacon 1
Luciano Coelho [Mon, 10 Nov 2014 09:10:15 +0000 (11:10 +0200)]
iwlwifi: mvm: schedule CSA time event a bit before beacon 1

Instead of using a hardcoded number of TUs before beacon 0 as the time
to start the absence and actual channel switch, calculate it in
relation to the beacon interval.  We use 10 TUs + beacon interval
before beacon 0 to target a bit before beacon 1.  This gives us enough
time to switch to the new channel before the AP/GO switches.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
9 years agoiwlwifi: mvm: add CSA absent time event for clients
Luciano Coelho [Mon, 10 Nov 2014 09:10:14 +0000 (11:10 +0200)]
iwlwifi: mvm: add CSA absent time event for clients

Add an absent time event when pre_channel_switch is called and use the
time event started indication to set the disable_tx bit instead of
doing it in unassign_vif().  This is done so that the firmware queues
are stopped before the actual switch takes place to avoid losing
packets while the AP/GO is performing its actual switch.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
9 years agoiwlwifi: mvm: use switching_chanctx argument instead of csa_active
Luciano Coelho [Mon, 10 Nov 2014 09:10:13 +0000 (11:10 +0200)]
iwlwifi: mvm: use switching_chanctx argument instead of csa_active

Now that all CSA flows are using the switch_vif_chanctx op, we can
rely on the switching_chanctx boolean that is passed to the
__iwl_mvm_assign_vif_chanctx() and __iwl_mvm_unassign_vif_chanctx()
functions to decide whether the context switch flows need to be
executed.  In this way we make the chanctx switch flow more generic,
without having to rely on the csa_active flag being set.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
9 years agoiwlwifi: mvm: disable PS during channel switch
Luciano Coelho [Mon, 10 Nov 2014 09:10:12 +0000 (11:10 +0200)]
iwlwifi: mvm: disable PS during channel switch

Disable PS when pre_channel_switch is called and add the
post_channel_switch operation to re-enable PS when the channel switch
is completed.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
9 years agoiwlwifi: mvm: Handle failed beacon transmissions during CSA
Andrei Otcheretianski [Mon, 10 Nov 2014 09:10:11 +0000 (11:10 +0200)]
iwlwifi: mvm: Handle failed beacon transmissions during CSA

The spec requires to decrement the CSA counters based on TBTT,
regardless if the beacon was actually transmitted. Previously, the fw
would send beacon notifications only for successfully transmitted
beacons. This behavior resulted in inaccurate CSA countdown. In order
to address this issue, the fw was changed to send beacon
notifications also for not transmitted beacons. Such notifications
have TX_STATUS_INTERNAL_ABORT (0x92).
Don't start the CSA countdown before first successfully transmitted
beacon, in order to guarantee that the CSA is announced for a
required period.

Signed-off-by: Andrei Otcheretianski <andrei.otcheretianski@intel.com>
9 years agoiwlwifi: mvm: return the actual error code when switch_vif_chanctx fails
Luciano Coelho [Mon, 10 Nov 2014 09:10:10 +0000 (11:10 +0200)]
iwlwifi: mvm: return the actual error code when switch_vif_chanctx fails

We have code to recover and go back to the original channel context if
something fails in the middle of switch_vif_chanctx, but we return the
error code of the recover calls instead of the original code, so if
the recovery succeeds, we will return 0 (success).  Fix this by not
assigning the return value of the recovery calls to ret.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
9 years agoiwlwifi: mvm: add support for CHANCTX_SWMODE_REASSIGN_VIF
Luciano Coelho [Mon, 10 Nov 2014 09:10:09 +0000 (11:10 +0200)]
iwlwifi: mvm: add support for CHANCTX_SWMODE_REASSIGN_VIF

Add support to reassign vif in switch_vif_chanctx.  This is similar to
the existing CHANCTX_SWMODE_SWAP_CONTEXTS mode, but doesn't delete the
old context nor creates a new one, doing to switch between two
existing contexts.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
9 years agoiwlwifi: mvm: refactor iwl_mvm_switch_vif_chanctx to support other modes
Luciano Coelho [Mon, 10 Nov 2014 09:10:08 +0000 (11:10 +0200)]
iwlwifi: mvm: refactor iwl_mvm_switch_vif_chanctx to support other modes

Currently we only support the CHANCTX_SWMODE_SWAP_CONTEXTS mode, but
we need to support other modes as well.  Spin a new function off in
order to make it easier to support other modes.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
9 years agoiwlwifi: mvm: only save csa_vif in AP/GO mode
Luciano Coelho [Mon, 10 Nov 2014 09:10:07 +0000 (11:10 +0200)]
iwlwifi: mvm: only save csa_vif in AP/GO mode

We only need the csa_vif in AP/GO modes, and assigning for other
interfaces may cause problems, because csa_vif is never cleared.  To
prevent this, only assign the value if the iftype is
NL80211_IFTYPE_AP.  Use a switch to do this, even though, for now,
only the AP interface type is handled, because soon other interface
types will be added as well.

Additionally, convert the WARN() in the error case when a
channel-switch is already running to WARN_ONCE().

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
9 years agoiwlwifi: mvm: use new pre_channel_switch op instead of channel_switch_beacon
Luciano Coelho [Mon, 10 Nov 2014 09:10:06 +0000 (11:10 +0200)]
iwlwifi: mvm: use new pre_channel_switch op instead of channel_switch_beacon

A new callback has been added to prepare the device for a channel
switch.  Use the new callback instead of the old channel_switch_beacon
operation.

This makes it possible to remove the channel_switch_beacon operation
from mac80211.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
9 years agoiwlwifi: mvm: implement mac80211 TDLS channel-switch APIs
Arik Nemtsov [Thu, 23 Oct 2014 15:03:10 +0000 (18:03 +0300)]
iwlwifi: mvm: implement mac80211 TDLS channel-switch APIs

Maintain a TDLS channel-switch state and update it according to
notifications from FW and timeouts. Explicitly check all state
transitions are valid.
When switching is initiated by mac80211, use a delayed work to
periodically reschedule it from iwlwifi.
Give the FW mac80211 generated TDLS channel-switch request/response
templates. It will change appropriate values (switch timings) and Tx
them at appropriate times.

Enable the channel switch wiphy capability bit when the FW supports it.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: block TID when using TDLS
Liad Kaufman [Mon, 6 Oct 2014 17:08:56 +0000 (19:08 +0200)]
iwlwifi: mvm: block TID when using TDLS

Block TID 4 (VI) on a vif when using TDLS. This TID will be used by FW
for TDLS channel-switch requests and PTI requests.

Signed-off-by: Liad Kaufman <liad.kaufman@intel.com>
Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: use private TFD queues for TDLS stations
Arik Nemtsov [Sun, 21 Sep 2014 16:10:04 +0000 (19:10 +0300)]
iwlwifi: mvm: use private TFD queues for TDLS stations

When adding a TDLS station, allocate 4 new queues for it. Configure them
to FW and enable them. On station removal, drain the queues if needed
and disable them when empty.

Make sure to flush all packets in the private queues of TDLS stations in
the mac80211 flush() callback.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: disconnect TDLS peers on reconfig
Arik Nemtsov [Sun, 14 Sep 2014 16:13:54 +0000 (19:13 +0300)]
iwlwifi: mvm: disconnect TDLS peers on reconfig

When TDLS peers are present the FW will send packets on a dedicated
TID vs. the peer when performing TDLS channel-switches. The driver
configures the TID on connection to the peer and the FW is responsible
for maintaining the state of QoS seqno and PN/IV for encryption.

If the FW asserts, the driver cannot correctly reconfigure the starting
seqno/PN to the reloaded FW, thus forcing us to reconnect the peer.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: allow private per-STA TFD queues
Arik Nemtsov [Sun, 21 Sep 2014 16:07:30 +0000 (19:07 +0300)]
iwlwifi: mvm: allow private per-STA TFD queues

TDLS stations will have private queues, so consider them as well when
allocating a new one. Consolidate the HW-queue iterating code into
a single exported function, to be used by the TDLS code in the future.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: configure TDLS peers to FW
Arik Nemtsov [Mon, 15 Sep 2014 15:48:59 +0000 (18:48 +0300)]
iwlwifi: mvm: configure TDLS peers to FW

Send a dedicated TDLS_CONFIG command when a TDLS peer joins/leaves. The
fields for the command are mostly place-holders, as most of the FW
functionality is not implemented. In the future the dedicated FW TID
will be used for channel-switching and buffer-sta functionality.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: add TDLS channel switch FW APIs
Arik Nemtsov [Thu, 11 Sep 2014 10:10:08 +0000 (13:10 +0300)]
iwlwifi: mvm: add TDLS channel switch FW APIs

Add a channel-switch command and a switch-start notification. Also add a
FW TLV bit indicating TDLS channel switching support.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: declare TDLS support
Arik Nemtsov [Sun, 11 May 2014 15:13:04 +0000 (18:13 +0300)]
iwlwifi: mvm: declare TDLS support

The driver/FW combination now support TDLS.
Check the FW support with the TLV bit.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: pull SNAP header into skb->head
Johannes Berg [Thu, 13 Nov 2014 10:06:43 +0000 (11:06 +0100)]
iwlwifi: mvm: pull SNAP header into skb->head

When we pre-populate the skb->head for the stack, we only pull
in the 802.11 header including crypto (assuming the packet isn't
short enough to be in there completely.) This is fine, but in
ieee80211_data_to_8023() we later unconditionally pull 8 more
bytes for the SNAP header and ethertype field (except for mesh
or 4-addr, where it's even more, but we don't care as much about
them).

Avoid the additional later pull by pulling in those 8 bytes here.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Reviewed-by: IdoX Yariv <ido@wizery.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: pull crypto header into skb->head
Johannes Berg [Wed, 12 Nov 2014 16:12:05 +0000 (17:12 +0100)]
iwlwifi: mvm: pull crypto header into skb->head

When we pre-populate the skb->head for the stack, we only pull
in the 802.11 header (assuming the packet isn't short enough to
be in there completely.) This is fine, but in many cases we'll
pull in the crypto headers pretty much immediately afterwards,
so to avoid that pull in the crypto header early.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Reviewed-by: IdoX Yariv <ido@wizery.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: build mac80211 rx_status in place
Johannes Berg [Wed, 12 Nov 2014 15:45:12 +0000 (16:45 +0100)]
iwlwifi: build mac80211 rx_status in place

Instead of building the rx_status on the stack and then
copying it to the skb, allocate the skb a bit earlier
and then build the rx_status in place.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Reviewed-by: IdoX Yariv <ido@wizery.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: add WEP RX hardware offload support
Johannes Berg [Wed, 12 Nov 2014 22:54:48 +0000 (23:54 +0100)]
iwlwifi: mvm: add WEP RX hardware offload support

In the original driver, we decided to not implement WEP RX hardware
offload because of a quirk with the firmware API - it allows setting
global WEP keys that then get used for all virtual interfaces, which
is clearly wrong if more than one exists, and it allows setting per-
station keys but then separates multicast and unicast keys.

In order to implement WEP RX hardware offload, work around these
limitations by uploading each WEP key twice, once as multicast and
once as unicast, but point them both to the same key slot (offset)
and use the same key material so the slot overwrite on the second
upload doesn't actually change anything. Upon removal, also remove
the key twice so the station no longer references it.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>