7 years agomac80211: Remove superfluous ieee80211_rx_h_remove_qos_control
Helmut Schaa [Mon, 19 Dec 2011 10:39:54 +0000]
mac80211: Remove superfluous ieee80211_rx_h_remove_qos_control

This seems to not serve any purpose anymore, at least all frame
processing afterwards seems to be able to deal with QoS frames. So,
let's save the expensive memmove and just leave the QoS header in the
802.11 frame for further processing.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/padovan/blueto...
John W. Linville [Mon, 2 Jan 2012 21:43:54 +0000]
Merge branch 'master' of git://git./linux/kernel/git/padovan/bluetooth-next

7 years agoBluetooth: Fix a compile warning in RFCOMM
Gustavo F. Padovan [Tue, 27 Dec 2011 16:43:41 +0000]
Bluetooth: Fix a compile warning in RFCOMM

sock and sk were leftover from another change.

Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Incorrect address while storing LTK.
Hemant Gupta [Fri, 23 Dec 2011 05:37:24 +0000]
Bluetooth: Incorrect address while storing LTK.

This patch fixes incorrect address storage while storing
Long Term Key for LE Devices using SMP (Security Manager Protocol).
The address stored should be of remote device and not of source device.

Signed-off-by: Hemant Gupta <hemant.gupta@stericsson.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Don't disable interrupt when locking the queue
Gustavo F. Padovan [Thu, 22 Dec 2011 18:35:05 +0000]
Bluetooth: Don't disable interrupt when locking the queue

We run everything in process context now.

Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: remove *_bh usage from hci_dev_list and hci_cb_list
Gustavo F. Padovan [Thu, 22 Dec 2011 18:30:27 +0000]
Bluetooth: remove *_bh usage from hci_dev_list and hci_cb_list

They don't need to disable interrupts anymore, we only run in process
context now.

Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Remove lock from inquiry_cache
Gustavo F. Padovan [Thu, 22 Dec 2011 18:15:42 +0000]
Bluetooth: Remove lock from inquiry_cache

It was never used, so removing it.

Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: add debug output to l2cap_ack_timeout()
Gustavo F. Padovan [Thu, 22 Dec 2011 18:56:05 +0000]
Bluetooth: add debug output to l2cap_ack_timeout()

Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Remove l2cap priority from inside RFCOMM.
Gustavo F. Padovan [Tue, 20 Dec 2011 18:30:44 +0000]
Bluetooth: Remove l2cap priority from inside RFCOMM.

RFCOMM needs a proper priority mechanism inside itself and not try to use
l2cap priority to fix its own problem.

Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Add support for BCM20702A0 [0a5c:21e3]
Jesse Sung [Thu, 22 Dec 2011 02:48:47 +0000]
Bluetooth: Add support for BCM20702A0 [0a5c:21e3]

Add another vendor specific ID for BCM20702A0.

output of usb-devices:
T: Bus=06 Lev=01 Prnt=01 Port=00 Cnt=01 Dev#= 4 Spd=12 MxCh= 0
D: Ver= 2.00 Cls=ff(vend.) Sub=01 Prot=01 MxPS=64 #Cfgs= 1
P: Vendor=0a5c ProdID=21e3 Rev=01.12
S: Manufacturer=Broadcom Corp
S: Product=BCM20702A0
S: SerialNumber=9439E5CBF66C
C: #Ifs= 4 Cfg#= 1 Atr=e0 MxPwr=0mA
I: If#= 0 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=01 Prot=01 Driver=(none)
I: If#= 1 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=01 Prot=01 Driver=(none)
I: If#= 2 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=(none)
I: If#= 3 Alt= 0 #EPs= 0 Cls=fe(app. ) Sub=01 Prot=01 Driver=(none)

Signed-off-by: Wen-chien Jesse Sung <jesse.sung@canonical.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Add SMP to User Passkey and Confirm
Brian Gix [Thu, 22 Dec 2011 00:12:13 +0000]
Bluetooth: Add SMP to User Passkey and Confirm

Low Energy pairing is performed through the SMP (Security Manager Protocol)
mechanism rather than HCI.

Signed-off-by: Brian Gix <bgix@codeaurora.org>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Add MITM mechanism to LE-SMP
Brian Gix [Thu, 22 Dec 2011 00:12:12 +0000]
Bluetooth: Add MITM mechanism to LE-SMP

To achive Man-In-The-Middle (MITM) level security with Low Energy,
we have to enable User Passkey Comparison.  This commit modifies the
hard-coded JUST-WORKS pairing mechanism to support query via the MGMT
interface of Passkey comparison and User Confirmation.

Signed-off-by: Brian Gix <bgix@codeaurora.org>
Acked-by: Marcel Holtmann<marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Fix deadlocks with sock lock and L2CAP timers locks
Ulisses Furquim [Wed, 21 Dec 2011 22:02:36 +0000]
Bluetooth: Fix deadlocks with sock lock and L2CAP timers locks

When cancelling a delayed work (timer) in L2CAP we can not sleep holding
the sock mutex otherwise we might deadlock with an L2CAP timer handler.
This is possible because RX/TX and L2CAP timers run in different workqueues.
The scenario below illustrates the problem. Thus we are now avoiding to
sleep on the timers locks.

 ======================================================
 [ INFO: possible circular locking dependency detected ]
 3.1.0-05270-ga978dc7-dirty #239
 -------------------------------------------------------
 kworker/1:1/873 is trying to acquire lock:
  (sk_lock-AF_BLUETOOTH-BTPROTO_L2CAP){+.+...}, at: [<ffffffffa002ceac>] l2cap_chan_timeout+0x3c/0xe0 [bluetooth]

 but task is already holding lock:
  ((&(&chan->chan_timer)->work)){+.+...}, at: [<ffffffff81051a86>] process_one_work+0x126/0x450

 which lock already depends on the new lock.

 the existing dependency chain (in reverse order) is:

 -> #1 ((&(&chan->chan_timer)->work)){+.+...}:
        [<ffffffff8106b276>] check_prevs_add+0xf6/0x170
        [<ffffffff8106b903>] validate_chain+0x613/0x790
        [<ffffffff8106dfee>] __lock_acquire+0x4be/0xac0
        [<ffffffff8106ec2d>] lock_acquire+0x8d/0xb0
        [<ffffffff81052a6f>] wait_on_work+0x4f/0x160
        [<ffffffff81052ca3>] __cancel_work_timer+0x73/0x80
        [<ffffffff81052cbd>] cancel_delayed_work_sync+0xd/0x10
        [<ffffffffa002f2ed>] l2cap_chan_connect+0x22d/0x470 [bluetooth]
        [<ffffffffa002fb51>] l2cap_sock_connect+0xb1/0x140 [bluetooth]
        [<ffffffff8130811b>] kernel_connect+0xb/0x10
        [<ffffffffa00cf98a>] rfcomm_session_create+0x12a/0x1c0 [rfcomm]
        [<ffffffffa00cfbe7>] __rfcomm_dlc_open+0x1c7/0x240 [rfcomm]
        [<ffffffffa00d07c2>] rfcomm_dlc_open+0x42/0x70 [rfcomm]
        [<ffffffffa00d3b03>] rfcomm_sock_connect+0x103/0x150 [rfcomm]
        [<ffffffff8130bd7e>] sys_connect+0xae/0xc0
        [<ffffffff813368d2>] compat_sys_socketcall+0xb2/0x220
        [<ffffffff813b2089>] sysenter_dispatch+0x7/0x30

 -> #0 (sk_lock-AF_BLUETOOTH-BTPROTO_L2CAP){+.+...}:
        [<ffffffff8106b16d>] check_prev_add+0x6cd/0x6e0
        [<ffffffff8106b276>] check_prevs_add+0xf6/0x170
        [<ffffffff8106b903>] validate_chain+0x613/0x790
        [<ffffffff8106dfee>] __lock_acquire+0x4be/0xac0
        [<ffffffff8106ec2d>] lock_acquire+0x8d/0xb0
        [<ffffffff8130d91a>] lock_sock_nested+0x8a/0xa0
        [<ffffffffa002ceac>] l2cap_chan_timeout+0x3c/0xe0 [bluetooth]
        [<ffffffff81051ae4>] process_one_work+0x184/0x450
        [<ffffffff8105276e>] worker_thread+0x15e/0x340
        [<ffffffff81057bb6>] kthread+0x96/0xa0
        [<ffffffff813b1ef4>] kernel_thread_helper+0x4/0x10

 other info that might help us debug this:

  Possible unsafe locking scenario:

        CPU0                    CPU1
        ----                    ----
   lock((&(&chan->chan_timer)->work));
                                lock(sk_lock-AF_BLUETOOTH-BTPROTO_L2CAP);
                                lock((&(&chan->chan_timer)->work));
   lock(sk_lock-AF_BLUETOOTH-BTPROTO_L2CAP);

  *** DEADLOCK ***

 2 locks held by kworker/1:1/873:
  #0:  (events){.+.+.+}, at: [<ffffffff81051a86>] process_one_work+0x126/0x450
  #1:  ((&(&chan->chan_timer)->work)){+.+...}, at: [<ffffffff81051a86>] process_one_work+0x126/0x450

 stack backtrace:
 Pid: 873, comm: kworker/1:1 Not tainted 3.1.0-05270-ga978dc7-dirty #239
 Call Trace:
  [<ffffffff813a0f6e>] print_circular_bug+0xd2/0xe3
  [<ffffffff8106b16d>] check_prev_add+0x6cd/0x6e0
  [<ffffffff8106b276>] check_prevs_add+0xf6/0x170
  [<ffffffff8106b903>] validate_chain+0x613/0x790
  [<ffffffff8106dfee>] __lock_acquire+0x4be/0xac0
  [<ffffffff8130d8f6>] ? lock_sock_nested+0x66/0xa0
  [<ffffffff8106ea30>] ? lock_release_nested+0x100/0x110
  [<ffffffff8130d8f6>] ? lock_sock_nested+0x66/0xa0
  [<ffffffff8106ec2d>] lock_acquire+0x8d/0xb0
  [<ffffffffa002ceac>] ? l2cap_chan_timeout+0x3c/0xe0 [bluetooth]
  [<ffffffff8130d91a>] lock_sock_nested+0x8a/0xa0
  [<ffffffffa002ceac>] ? l2cap_chan_timeout+0x3c/0xe0 [bluetooth]
  [<ffffffff81051a86>] ? process_one_work+0x126/0x450
  [<ffffffffa002ceac>] l2cap_chan_timeout+0x3c/0xe0 [bluetooth]
  [<ffffffff81051ae4>] process_one_work+0x184/0x450
  [<ffffffff81051a86>] ? process_one_work+0x126/0x450
  [<ffffffffa002ce70>] ? l2cap_security_cfm+0x4e0/0x4e0 [bluetooth]
  [<ffffffff8105276e>] worker_thread+0x15e/0x340
  [<ffffffff81052610>] ? manage_workers+0x110/0x110
  [<ffffffff81057bb6>] kthread+0x96/0xa0
  [<ffffffff813b1ef4>] kernel_thread_helper+0x4/0x10
  [<ffffffff813af69d>] ? retint_restore_args+0xe/0xe
  [<ffffffff81057b20>] ? __init_kthread_worker+0x70/0x70
  [<ffffffff813b1ef0>] ? gs_change+0xb/0xb

Signed-off-by: Ulisses Furquim <ulisses@profusion.mobi>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Make HCI call directly into SCO and L2CAP event functions
Ulisses Furquim [Wed, 21 Dec 2011 12:11:33 +0000]
Bluetooth: Make HCI call directly into SCO and L2CAP event functions

The struct hci_proto and all related register/unregister and dispatching
code was removed. HCI core code now call directly the SCO and L2CAP
event functions.

Signed-off-by: Ulisses Furquim <ulisses@profusion.mobi>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Remove magic numbers from le scan cmd
Andrei Emeltchenko [Mon, 19 Dec 2011 14:14:18 +0000]
Bluetooth: Remove magic numbers from le scan cmd

Make code readable by removing magic numbers.

Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: fix bt_accept_dequeue() to work in process context
Gustavo F. Padovan [Tue, 20 Dec 2011 19:15:56 +0000]
Bluetooth: fix bt_accept_dequeue() to work in process context

No local_bh_disable is needed there once we run everything in process
context. The same goes for the replacement of bh_lock_sock() by
lock_sock().

Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agomac80211: Keep skb->piority for relayed frames in AP mode
Helmut Schaa [Wed, 21 Dec 2011 08:11:35 +0000]
mac80211: Keep skb->piority for relayed frames in AP mode

When mac80211 relays a frame from STA1 to STA2 in AP mode it will get
re-classified in the tx path. Unfortunately the frame protocol field
is always set to ETH_P_8023 while the classification only kicks in
for ETH_P_IP. Hence, a high priority frame from STA1 will be send to
STA2 as best effort.

Instead of running classification on the frame just use the same
priority as STA1 did. Do this by adding 256 to the skb->priority
to allow cfg80211_classify8021d to shortcut frame classification.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agomwifiex: fix issues in band configuration code
Amitkumar Karwar [Wed, 21 Dec 2011 07:47:22 +0000]
mwifiex: fix issues in band configuration code

Currently due to following issues in the code even if device is
configured in B only, G only or BG mode using iw bitrates command,
ibss is getting created in BGN mode.

1) mwifiex_channels_to_cfg80211_channel_type() routine gives channel
type as NL80211_CHAN_HT20 for non-HT channel as well, because driver
doesn't store HT information provided by stack for the channel.
This issue is fixed by maintaining channel type information in
'adapter->channel_type'.
2) Band configuration is unnecessarily overwritten with BGN/AN while
setting channel.

This patch makes sure that "adapter->config_bands" correctly gets
modified while setting channel.

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agomwifiex: use IEEE80211_HT_PARAM_CHA_SEC_* macros
Amitkumar Karwar [Wed, 21 Dec 2011 07:47:21 +0000]
mwifiex: use IEEE80211_HT_PARAM_CHA_SEC_* macros

Replace driver specific macros with the corresponding
IEEE80211_HT_PARAM_CHA_SEC_* macros defined in ieee80211.h.
Also, rename 'adapter->chan_offset' to 'adapter->sec_chan_offset'
for consistency.

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agomwifiex: remove redundant band config code
Amitkumar Karwar [Wed, 21 Dec 2011 07:47:20 +0000]
mwifiex: remove redundant band config code

struct mwifiex_ds_band_cfg and mwifiex_set_radio_band_cfg() routine
are unnecessary. It can be done with simple equivalant code.

Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agoatheros: force endian checks on atheros wireless drivers
Luis R. Rodriguez [Tue, 20 Dec 2011 19:18:48 +0000]
atheros: force endian checks on atheros wireless drivers

Please NACK nasty patches.

Cc: Kalle Valo <kvalo@qca.qualcomm.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agomac80211: ibss should not purge clients it is not responsible for
Marek Lindner [Tue, 20 Dec 2011 15:16:52 +0000]
mac80211: ibss should not purge clients it is not responsible for

The IBSS merge code calls ieee80211_sta_expire() with a relatively
short expire timeout that purges other clients prematurely. The
expire function has to check that only the clients belonging to
the vif in question are purged.

Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agomwl8k: Call ieee80211_stop_tx_ba_cb_irqsafe for already deleted BA stream
Yogesh Ashok Powar [Tue, 20 Dec 2011 06:07:08 +0000]
mwl8k: Call ieee80211_stop_tx_ba_cb_irqsafe for already deleted BA stream

When stack calls ampdu_action with action = IEEE80211_AMPDU_TX_STOP
for a stream that has already been removed from the driver, call
ieee80211_tx_ba_stop_irqsafe to clear the stream in the stack.

Signed-off-by: Yogesh Ashok Powar <yogeshp@marvell.com>
Signed-off-by: Nishant Sarmukadam <nishants@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agoath9k: fix roadkill due to Joe's patch on ath_dbg() changes
Luis R. Rodriguez [Tue, 20 Dec 2011 03:30:43 +0000]
ath9k: fix roadkill due to Joe's patch on ath_dbg() changes

Joe changed ath_dbg() to simpify code but while his patch was
being merged dfs.c was born and as such did not get the change
Joe envisioned. This fixes that. Test compiled with:

 make allmodconfig

Cc: Joe Perches <joe@perches.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: John W. Linville <linville@tuxdriver.com>
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Luis R. Rodriguez <mcgrof@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agoiwlagn: fix (remove) use of PAGE_SIZE
Johannes Berg [Mon, 19 Dec 2011 22:00:59 +0000]
iwlagn: fix (remove) use of PAGE_SIZE

The ICT code erroneously uses PAGE_SIZE. The bug
is that PAGE_SIZE isn't necessarily 4096, so on
such platforms this code will not work correctly
as we'll try to attempt to read an index in the
table that the device never wrote, it always has
4096-byte pages.

Additionally, the manual alignment code here is
unnecessary -- Documentation/DMA-API-HOWTO.txt
states:
  The cpu return address and the DMA bus master address are both
  guaranteed to be aligned to the smallest PAGE_SIZE order which
  is greater than or equal to the requested size.  This invariant
  exists (for example) to guarantee that if you allocate a chunk
  which is smaller than or equal to 64 kilobytes, the extent of the
  buffer you receive will not cross a 64K boundary.

Just use appropriate new constants and get rid of
the alignment code.

Cc: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Cc: stable@vger.kernel.org
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agort2800: Add support for the Fujitsu Stylistic Q550
Alan Cox [Mon, 19 Dec 2011 21:07:33 +0000]
rt2800: Add support for the Fujitsu Stylistic Q550

Just another USB identifier.

Signed-off-by: Alan Cox <alan@linux.intel.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agolibertas: clean up scan thread handling
Andres Salomon [Mon, 19 Dec 2011 20:22:58 +0000]
libertas: clean up scan thread handling

The libertas scan thread expects priv->scan_req to be non-NULL.  In theory,
it should always be set.  In practice, we've seen the following oops:

[ 8363.067444] Unable to handle kernel NULL pointer dereference at virtual address 00000004
[ 8363.067490] pgd = c0004000
[ 8363.078393] [00000004] *pgd=00000000
[ 8363.086711] Internal error: Oops: 17 [#1] PREEMPT
[ 8363.091375] Modules linked in: fuse libertas_sdio libertas psmouse mousedev ov7670 mmp_camera joydev videobuf2_core videobuf2_dma_sg videobuf2_memops [last unloaded: scsi_wait_scan]
[ 8363.107490] CPU: 0    Not tainted  (3.0.0-gf7ccc69 #671)
[ 8363.112799] PC is at lbs_scan_worker+0x108/0x5a4 [libertas]
[ 8363.118326] LR is at 0x0
[ 8363.120836] pc : [<bf03a854>]    lr : [<00000000>]    psr: 60000113
[ 8363.120845] sp : ee66bf48  ip : 00000000  fp : 00000000
[ 8363.120845] r10: ee2c2088  r9 : c04e2efc  r8 : eef97005
[ 8363.132231] r7 : eee0716f  r6 : ee2c02c0  r5 : ee2c2088  r4 : eee07160
[ 8363.137419] r3 : 00000000  r2 : a0000113  r1 : 00000001  r0 : eee07160
[ 8363.143896] Flags: nZCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment kernel
[ 8363.157630] Control: 10c5387d  Table: 2e754019  DAC: 00000015
[ 8363.163334] Process kworker/u:1 (pid: 25, stack limit = 0xee66a2f8)

While I've not found a smoking gun, there are two places that raised red flags
for me.  The first is in _internal_start_scan, when we queue up a scan; we
first queue the worker, and then set priv->scan_req.  There's theoretically
a 50mS delay which should be plenty, but doing things that way just seems
racy (and not in the good way).

The second is in the scan worker thread itself.  Depending on the state of
priv->scan_channel, we cancel pending scan runs and then requeue a run in
300mS.  We then send the scan command down to the hardware, sleep, and if
we get scan results for all the desired channels, we set priv->scan_req to
NULL.  However, it that's happened in less than 300mS, what happens with
the pending scan run?

This patch addresses both of those concerns.  With the patch applied, we
have not seen the oops in the past two weeks.

Signed-off-by: Andres Salomon <dilinger@queued.net>
Cc: stable@kernel.org
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agoath9k: classify DFS debug header further
Luis R. Rodriguez [Mon, 19 Dec 2011 16:11:54 +0000]
ath9k: classify DFS debug header further

DFS_DEBUG_H is very generic, instead use something more specific
to ath9k such as ATH9K_DFS_DEBUG_H.

Reported-by: Julian Calaby <julian.calaby@gmail.com>
Signed-off-by: Luis R. Rodriguez <mcgrof@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agoMerge branch 'for-linville' of git://git.kernel.org/pub/scm/linux/kernel/git/luca...
John W. Linville [Wed, 21 Dec 2011 20:02:18 +0000]
Merge branch 'for-linville' of git://git./linux/kernel/git/luca/wl12xx

7 years agoBluetooth: Always compile SCO and L2CAP in Bluetooth Core
Ulisses Furquim [Wed, 21 Dec 2011 03:32:09 +0000]
Bluetooth: Always compile SCO and L2CAP in Bluetooth Core

The handling of SCO audio links and the L2CAP protocol are essential to
any system with Bluetooth thus are always compiled in from now on.

Signed-off-by: Ulisses Furquim <ulisses@profusion.mobi>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Remove global mutex hci_task_lock
Ulisses Furquim [Tue, 20 Dec 2011 19:10:51 +0000]
Bluetooth: Remove global mutex hci_task_lock

The hci_task_lock mutex (previously a lock) was supposed to protect the
register/unregister of HCI protocols against RX/TX tasks. This will not
be needed anymore because SCO and L2CAP will always be compiled.

Moreover, with the recent move of RX/TX to workqueues per device the
global hci_task_lock was causing starvation between different HCI
devices.

Signed-off-by: Ulisses Furquim <ulisses@profusion.mobi>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: btusb: fix bInterval for high/super speed isochronous endpoints
Bing Zhao [Wed, 21 Dec 2011 02:19:00 +0000]
Bluetooth: btusb: fix bInterval for high/super speed isochronous endpoints

For high-speed/super-speed isochronous endpoints, the bInterval
value is used as exponent, 2^(bInterval-1). Luckily we have
usb_fill_int_urb() function that handles it correctly. So we just
call this function to fill in the RX URB.

Cc: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agowl12xx: mark no sched scan only after FW event
Eyal Shapira [Tue, 20 Dec 2011 12:55:38 +0000]
wl12xx: mark no sched scan only after FW event

stop sched scan isn't an immediate operation
and we need to wait for PERIODIC_SCAN_COMPLETE_EVENT_ID
after sending a stop before changing internal state
and notifying upper layers.
Not doing this caused problems when canceling an existing sched
scan and immediately requesting to start a new one
with a different configuration as the FW was still
in the middle of the previous sched scan.

Signed-off-by: Eyal Shapira <eyal@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>

7 years agowl12xx: fix sched scan of DFS channels
Eyal Shapira [Tue, 20 Dec 2011 10:04:01 +0000]
wl12xx: fix sched scan of DFS channels

DFS channels weren't scanned properly because
min/max_duration weren't set for these channels
even though they're required by the FW.
The change sets passive_duration and min/max_duration
for all channels as the FW uses the correct parameters
according to the channel type.

Signed-off-by: Eyal Shapira <eyal@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>

7 years agowl12xx: add missing copyright notice
Luciano Coelho [Mon, 19 Dec 2011 14:31:55 +0000]
wl12xx: add missing copyright notice

The wl12xx_platform_data.c file did not have a proper copyright
notice.

Signed-off-by: Luciano Coelho <coelho@ti.com>

7 years agowl12xx: check the actual vif operstate in wl1271_dev_notify
Eliad Peller [Sun, 18 Dec 2011 18:25:45 +0000]
wl12xx: check the actual vif operstate in wl1271_dev_notify

The current wl1271_dev_notify implementation sets the
new operstate to all associated stations (while only
a specific vif was changed).

Until we'll have a method to get the actual vif from
the given dev, check the current operstate of each vif.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>

7 years agowl12xx: stop device role on remove_interface
Eliad Peller [Sun, 18 Dec 2011 18:25:44 +0000]
wl12xx: stop device role on remove_interface

When removing a sta/ibss role, the device role has to
stopped (and disabled) as well.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>

7 years agowl12xx: fix checking of started dev role
Eliad Peller [Sun, 18 Dec 2011 18:25:43 +0000]
wl12xx: fix checking of started dev role

dev_role_id only indicates whether the dev role
is enabled, not started (e.g. on IBSS merge,
the device role is enabled, but not started).

Checking for any role in ROC (in order to determine
whether dev role was started) is wrong as well,
especially in multi-vif env.

Check for started dev role only by checking the dev_hlid.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>

7 years agowl12xx: flush packets before stopping dev role
Eliad Peller [Sun, 18 Dec 2011 18:25:42 +0000]
wl12xx: flush packets before stopping dev role

During sta disconnection, a deauth packet is being queued to
the dev role queue. However, the dev role is being stopped
before the packet was sent.

Flush the tx queue before stopping the dev role.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>

7 years agowl12xx: make WL1271_FLAG_IDLE flag per-vif
Eliad Peller [Sun, 18 Dec 2011 18:25:41 +0000]
wl12xx: make WL1271_FLAG_IDLE flag per-vif

This flag should be set per-vif, rather than globally.

Rename the flag to indicate IN_USE (rather than IDLE), as
in the default configuration (i.e. flag is clear) the vif
should be idle.

Change all the bit operations (and elp conditions) appropriately.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>

7 years agowl12xx: remove redundant code from wl1271_op_conf_tx
Eliad Peller [Sun, 18 Dec 2011 18:25:40 +0000]
wl12xx: remove redundant code from wl1271_op_conf_tx

Since the conf_tx callback passes the vif as param,
we must have been added first (and mac80211 verifies it).

Remove the handling of such case.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>

7 years agowl12xx: implement change_interface
Eliad Peller [Mon, 19 Dec 2011 10:00:03 +0000]
wl12xx: implement change_interface

Implement the change_interface callback by simply removing the
current vif and adding a new one after updating the vif type.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>

7 years agoBluetooth: Rename info_work to info_timer
Gustavo F. Padovan [Tue, 20 Dec 2011 12:57:28 +0000]
Bluetooth: Rename info_work to info_timer

It makes more sense this way, since info_timer is a timer using delayed
work API.

Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: convert security timer to delayed_work
Gustavo F. Padovan [Tue, 20 Dec 2011 12:57:27 +0000]
Bluetooth: convert security timer to delayed_work

This one also needs to run in process context

Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Move l2cap_{set,clear}_timer to l2cap.h
Gustavo F. Padovan [Tue, 20 Dec 2011 12:57:26 +0000]
Bluetooth: Move l2cap_{set,clear}_timer to l2cap.h

It is the only place where it is used.

Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Clean up magic pointers
Andrei Emeltchenko [Mon, 19 Dec 2011 14:31:30 +0000]
Bluetooth: Clean up magic pointers

Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Check for flow control mode
Andrei Emeltchenko [Mon, 19 Dec 2011 14:31:29 +0000]
Bluetooth: Check for flow control mode

Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Initialize default flow control mode
Andrei Emeltchenko [Mon, 19 Dec 2011 14:31:28 +0000]
Bluetooth: Initialize default flow control mode

Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoBluetooth: Split ctrl init to BREDR and AMP parts
Andrei Emeltchenko [Mon, 19 Dec 2011 14:31:27 +0000]
Bluetooth: Split ctrl init to BREDR and AMP parts

Current controller initialization is moved tp bredr_init and new
function added amp_init to handle later AMP init sequence. Current
AMP init sequence include Reset and Read Local Version.

Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Acked-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Gustavo F. Padovan <padovan@profusion.mobi>

7 years agoiwlwifi: Execute runtime calibration always
Venkataraman, Meenakshi [Thu, 15 Dec 2011 00:54:21 +0000]
iwlwifi: Execute runtime calibration always

Runtime DC calibration was previously conditional. Remove this
behaviour, as new devices support runtime DC calibration,
while older devices ignore the runtime DC calibration request.

This patch addresses low TX throughput issues seen with the 6205.

Signed-off-by: Meenakshi Venkataraman <meenakshi.venkataraman@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>

7 years agoiwlwifi: use bus_get_hw_id for IWL_TM_CMD_APP2DEV_GET_DEVICE_ID
Wey-Yi Guy [Sat, 10 Dec 2011 19:33:53 +0000]
iwlwifi: use bus_get_hw_id for IWL_TM_CMD_APP2DEV_GET_DEVICE_ID

instead of doing all the work in IWL_TM_CMD_APP2DEV_GET_DEVICE_ID, just use
the information from bus_get_hw_id()

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>

7 years agoiwlwifi: set hw_version in wiphy
Wey-Yi Guy [Sat, 10 Dec 2011 19:33:52 +0000]
iwlwifi: set hw_version in wiphy

Set the hw_version in wiphy structure

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>

7 years agoiwlwifi: deliver hw version in both string and u32 format
Wey-Yi Guy [Wed, 14 Dec 2011 15:41:36 +0000]
iwlwifi: deliver hw version in both string and u32 format

Add function to get hw version in both strind and u32 format

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>

7 years agoiwlagn: finer-grained HT disable
Johannes Berg [Tue, 13 Dec 2011 08:07:40 +0000]
iwlagn: finer-grained HT disable

At least while debugging, a lot of people use
11n_disable=1 to disable HT completely. To be
able to figure out what parts of HT cause the
problems we see, make the parameter a bitmap,
allowing to disable all of HT and aggregation
(TX/RX) separately.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>

7 years agoiwlagn: add IRQ tracing
Johannes Berg [Fri, 9 Dec 2011 15:26:13 +0000]
iwlagn: add IRQ tracing

The legacy IRQs could be read from a trace by their
IO accesses, but reading the ICT doesn't leave any
trace (pun intended ;-) ) so in order to see what
input they get we need to add specific tracepoints.

While at it, fix whitespace in two related places.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>

7 years agomac80211: Fixing sparse warning at sta_info.c
Yogesh Ashok Powar [Fri, 16 Dec 2011 06:17:15 +0000]
mac80211: Fixing sparse warning at sta_info.c

The commit 42624d4913a00219a8fdbb4bafd634d1d843be85
created following sparse warning
>net/mac80211/sta_info.c:965:24: warning: incorrect type in assignment (different address spaces)
>net/mac80211/sta_info.c:965:24:    expected struct tid_ampdu_tx *tid_tx
>net/mac80211/sta_info.c:965:24:    got struct tid_ampdu_tx [noderef] <asn:4>*<noident>

Making use of rcu_dereference_protected to fix the problem.

V2:
 - Replacing rcu_dereference with rcu_dereference_protected
   as suggested by Johannes.
 - Adding mutex_lock/unlock to satisfy the condition at
   rcu_dereference_protected

Cc: Nishant Sarmukadam <nishants@marvell.com>
Reported-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Yogesh Ashok Powar <yogeshp@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agoath9k: fix tx locking issues
Felix Fietkau [Mon, 19 Dec 2011 15:45:54 +0000]
ath9k: fix tx locking issues

The commit "ath9k: simplify tx locking" introduced a soft lockup triggered
by mac80211 sending a BAR frame triggered by a driver call to
ieee80211_tx_send_bar or ieee80211_tx_status.
Fix these issues by queueing processed tx status skbs and submitting them
to mac80211 outside of the lock.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Mohammed Shafi Shajakhan <mohammed@qca.qualcomm.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agoath9k_hw: make bluetooth coexistence support optional at compile time
Felix Fietkau [Sat, 17 Dec 2011 15:47:56 +0000]
ath9k_hw: make bluetooth coexistence support optional at compile time

Many systems (e.g. embedded systems) do not have wifi modules connected to
bluetooth modules, so bluetooth coexistence is irrelevant there. With the
addition of MCI support, ath9k picked up quite a bit of extra code that
can be compiled out this way.

This patch redefines ATH9K_HW_CAP_MCI and adds an inline wrapper for
querying the bluetooth coexistence scheme, allowing the compiler to
eliminate code that uses it, with only very little use of #ifdef.

On MIPS this reduces the total size for the modules by about 20k.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agob43: N-PHY: reorder functions: random cleanups
Rafał Miłecki [Sat, 17 Dec 2011 12:57:27 +0000]
b43: N-PHY: reorder functions: random cleanups

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agob43: N-PHY: reorder functions: move rest of workarounds
Rafał Miłecki [Sat, 17 Dec 2011 12:57:26 +0000]
b43: N-PHY: reorder functions: move rest of workarounds

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agob43: N-PHY: reorder functions: move RSSI calibration
Rafał Miłecki [Sat, 17 Dec 2011 12:57:25 +0000]
b43: N-PHY: reorder functions: move RSSI calibration

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agob43: N-PHY: split gain control workarounds
Rafał Miłecki [Sat, 17 Dec 2011 12:57:24 +0000]
b43: N-PHY: split gain control workarounds

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agob43: N-PHY: reorder functions: collect RSSI selects
Rafał Miłecki [Sat, 17 Dec 2011 12:57:23 +0000]
b43: N-PHY: reorder functions: collect RSSI selects

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agob43: N-PHY: reorder functions: collect samples ones
Rafał Miłecki [Sat, 17 Dec 2011 12:57:22 +0000]
b43: N-PHY: reorder functions: collect samples ones

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agob43: N-PHY: reorder functions: collect radio ones
Rafał Miłecki [Sat, 17 Dec 2011 12:57:21 +0000]
b43: N-PHY: reorder functions: collect radio ones

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agob43: N-PHY: reorder functions: put basic ones at beginning
Rafał Miłecki [Sat, 17 Dec 2011 12:57:20 +0000]
b43: N-PHY: reorder functions: put basic ones at beginning

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Acked-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agortlwifi: Fix locking problem introduces with commit 6539306b
Larry Finger [Sat, 17 Dec 2011 03:17:16 +0000]
rtlwifi: Fix locking problem introduces with commit 6539306b

When I tested commit 6539306, I did not notice that loading an out-of-tree
module turns off lockdep testing in kernel 3.2. For that reason, I missed
the kernel WARNING shown below:

The solution fixes the warning by partially reverting commit 6539306.

[   84.168146] ------------[ cut here ]------------
[   84.168155] WARNING: at kernel/mutex.c:198 mutex_lock_nested+0x309/0x310()
[   84.168158] Hardware name: HP Pavilion dv2700 Notebook PC
[   84.168161] Modules linked in: nfs lockd auth_rpcgss nfs_acl sunrpc af_packet cpufreq_conservative cpufreq_userspace cpufreq_powersave powernow_k8 mperf e
xt3 jbd ide_cd_mod cdrom snd_hda_codec_conexant arc4 rtl8192ce ide_pci_generic rtl8192c_common rtlwifi snd_hda_intel mac80211 snd_hda_codec snd_pcm snd_timer
 amd74xx ide_core cfg80211 k8temp snd joydev soundcore hwmon battery forcedeth i2c_nforce2 sg rfkill ac serio_raw snd_page_alloc button video i2c_core ipv6 a
utofs4 ext4 mbcache jbd2 crc16 sd_mod ahci ohci_hcd libahci libata scsi_mod ehci_hcd usbcore usb_common fan processor thermal
[   84.168231] Pid: 1218, comm: kworker/u:2 Not tainted 3.2.0-rc5-wl+ #155
[   84.168234] Call Trace:
[   84.168240]  [<ffffffff81048aaa>] warn_slowpath_common+0x7a/0xb0
[   84.168245]  [<ffffffff81048af5>] warn_slowpath_null+0x15/0x20
[   84.168249]  [<ffffffff813811f9>] mutex_lock_nested+0x309/0x310
[   84.168269]  [<ffffffffa00793f9>] ? rtl_ips_nic_on+0x49/0xb0 [rtlwifi]
[   84.168277]  [<ffffffffa00793f9>] rtl_ips_nic_on+0x49/0xb0 [rtlwifi]
[   84.168284]  [<ffffffffa007ab85>] rtl_pci_tx+0x1b5/0x560 [rtlwifi]
[   84.168291]  [<ffffffffa007635a>] rtl_op_tx+0x9a/0xa0 [rtlwifi]
[   84.168359]  [<ffffffffa043cf51>] __ieee80211_tx+0x181/0x2b0 [mac80211]
[   84.168375]  [<ffffffffa043ef06>] ieee80211_tx+0xf6/0x120 [mac80211]
[   84.168391]  [<ffffffffa043ee49>] ? ieee80211_tx+0x39/0x120 [mac80211]
[   84.168408]  [<ffffffffa043f80b>] ieee80211_xmit+0xdb/0x100 [mac80211]
[   84.168425]  [<ffffffffa043f730>] ? ieee80211_skb_resize.isra.26+0xb0/0xb0 [mac80211]
[   84.168441]  [<ffffffffa0440b2a>] ieee80211_tx_skb_tid+0x5a/0x70 [mac80211]
[   84.168458]  [<ffffffffa0443da2>] ieee80211_send_auth+0x152/0x1b0 [mac80211]
[   84.168474]  [<ffffffffa042e169>] ieee80211_work_work+0x1049/0x1860 [mac80211]
[   84.168489]  [<ffffffffa042d120>] ? free_work+0x20/0x20 [mac80211]
[   84.168504]  [<ffffffffa042d120>] ? free_work+0x20/0x20 [mac80211]
[   84.168510]  [<ffffffff81065ffc>] process_one_work+0x17c/0x530
[   84.168514]  [<ffffffff81065f92>] ? process_one_work+0x112/0x530
[   84.168519]  [<ffffffff81066994>] worker_thread+0x164/0x350
[   84.168524]  [<ffffffff8108420d>] ? trace_hardirqs_on+0xd/0x10
[   84.168528]  [<ffffffff81066830>] ? manage_workers.isra.28+0x220/0x220
[   84.168533]  [<ffffffff8106bc17>] kthread+0x87/0x90
[   84.168539]  [<ffffffff813854b4>] kernel_thread_helper+0x4/0x10
[   84.168543]  [<ffffffff81382bdd>] ? retint_restore_args+0xe/0xe
[   84.168547]  [<ffffffff8106bb90>] ? __init_kthread_worker+0x70/0x70
[   84.168552]  [<ffffffff813854b0>] ? gs_change+0xb/0xb
[   84.168554] ---[ end trace f25a4fdc768c028f ]---

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Cc: Stanislaw Gruska <sgruszka@redhat.com>
Cc: Chaoming Li <chaoming_li@realsil.com.cn>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: add Kconfig option for SDIO bus support
Franky Lin [Sat, 17 Dec 2011 02:37:20 +0000]
brcm80211: fmac: add Kconfig option for SDIO bus support

This patch adds a Kconfig option for SDIO bus support and abstracts
a build subset correspondingly. It's the final patch of fullmac bus
interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: exclude unnecessary header files
Franky Lin [Sat, 17 Dec 2011 02:37:19 +0000]
brcm80211: fmac: exclude unnecessary header files

This patch removes some headers files include lines from sdio layer
code. This is part of the fullmac bus interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: move idle macros to dhd_sdio.c
Franky Lin [Sat, 17 Dec 2011 02:37:18 +0000]
brcm80211: fmac: move idle macros to dhd_sdio.c

The idle macros are only used by dhd_sdio.c. It's more appropriate
to place them in dhd_sdio.c instead of dhd.h.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: move debug level macros to dhd_dbg.h
Franky Lin [Sat, 17 Dec 2011 02:37:17 +0000]
brcm80211: fmac: move debug level macros to dhd_dbg.h

Debug message level macros are used for debug purpose. It would be
more appropriate to place them at dhd_dbg.h.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: abstract ctrl frames interface function pointers
Franky Lin [Sat, 17 Dec 2011 02:37:16 +0000]
brcm80211: fmac: abstract ctrl frames interface function pointers

Abstract bus layer brcmf_bus_txctl/brcmf_bus_rxctl function
pointers for common layer. This patch is part of the fullmac bus
interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: abstract bus_txdata interface function pointer
Franky Lin [Sat, 17 Dec 2011 02:37:15 +0000]
brcm80211: fmac: abstract bus_txdata interface function pointer

Abstract bus layer brcmf_bus_txdata function pointer for common
layer. This patch is part of the fullmac bus interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: abstract bus_init interface function pointer
Franky Lin [Sat, 17 Dec 2011 02:37:14 +0000]
brcm80211: fmac: abstract bus_init interface function pointer

Abstract bus layer brcmf_bus_init function pointer for common
layer. This patch is part of the fullmac bus interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: move common layer bus interface context to dhd_bus.h
Franky Lin [Sat, 17 Dec 2011 02:37:13 +0000]
brcm80211: fmac: move common layer bus interface context to dhd_bus.h

dhd_bus.h is the header file for bus interface. Move functions
declarations and brcmf_bus structure to there. This is part of
the fullmac bus interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: abstract bus_stop interface function pointer
Franky Lin [Sat, 17 Dec 2011 02:37:12 +0000]
brcm80211: fmac: abstract bus_stop interface function pointer

Common layer should use interface function pointer stored in
brcmf_bus to invoke corresponding interface function in bus layer.
This patch is part of the fullmac bus interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: move sdio related macros to sdio_host.h
Franky Lin [Sat, 17 Dec 2011 02:37:11 +0000]
brcm80211: fmac: move sdio related macros to sdio_host.h

dhd_bus.h will be used as the shared header file for common layer
and bus layer. It should not contain any sdio specific macros.
This patch moves them to sdio_host.h as part of the fullmac bus
interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: stop referencing brcmf_sdio in common layer
Franky Lin [Sat, 17 Dec 2011 02:37:10 +0000]
brcm80211: fmac: stop referencing brcmf_sdio in common layer

brcmf_sdio is the data structure for sdio bus layer. Stop
referencing brcmf_sdio from common layer. This patch is part of
the fullmac bus interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: stop referencing brcmf_pub in bus layer
Franky Lin [Sat, 17 Dec 2011 02:37:09 +0000]
brcm80211: fmac: stop referencing brcmf_pub in bus layer

brcmf_pub is the data structure for common layer. Since brcmf_bus
should be the only structure shared by common layer and bus layer,
stop referencing brcmf_pub from bus layer. This patch is part of
the fullmac bus interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: move tx flow ctrl flag to bus layer
Franky Lin [Sat, 17 Dec 2011 02:37:08 +0000]
brcm80211: fmac: move tx flow ctrl flag to bus layer

txoff is the flow control flag for transmit used in sdio layer.
Move it to bus layer data structure brcmf_sdio. Also flag
management code is moved out of brcmf_txflowcontrol(). This is
part of the fullmac bus interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: move packet realloc stats to struct brcmf_bus
Franky Lin [Sat, 17 Dec 2011 02:37:07 +0000]
brcm80211: fmac: move packet realloc stats to struct brcmf_bus

tx_realloc is used by both common layer and bus layer. This patch
moves it to interface structure brcmf_bus as part of the fullmac
bus interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: move dongle statistics to struct brcmf_bus
Franky Lin [Sat, 17 Dec 2011 02:37:06 +0000]
brcm80211: fmac: move dongle statistics to struct brcmf_bus

Dongle statistics are shared data between common layer and bus
layer. This patch places them in bus interface structure brcmf_bus
as part of the fullmac bus interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: remove duplicate statistics from driver data structure
Franky Lin [Sat, 17 Dec 2011 02:37:05 +0000]
brcm80211: fmac: remove duplicate statistics from driver data structure

Some dongle statistics are stored in two places and synced when net
device status inquired. There is no need to do it this way any more.
Direct all usage to dongle stats structure in order to increase
readability.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: move driver up status to struct brcmf_bus
Franky Lin [Sat, 17 Dec 2011 02:37:04 +0000]
brcm80211: fmac: move driver up status to struct brcmf_bus

Driver up/down status to network interface need to be shared by
common layer and bus layer. Move it to bus interface structure
brcmf_bus as part of the fullmac bus interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: move maxctl to struct brcmf_bus
Franky Lin [Sat, 17 Dec 2011 02:37:03 +0000]
brcm80211: fmac: move maxctl to struct brcmf_bus

maxctl is the max size of rxctl request from protocol layer to bus
layer. Move it to bus interface structure brcmf_bus. This is part
of the fullmac bus interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: move sdio used statistics to struct brcmf_sdio
Franky Lin [Sat, 17 Dec 2011 02:37:02 +0000]
brcm80211: fmac: move sdio used statistics to struct brcmf_sdio

Some statistics only used by sdio modules. Move them to sdio layer
private structure brcmf_sdio. This is part of the fullmac bus
interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: change function proto_hdrpull parameter
Franky Lin [Sat, 17 Dec 2011 02:37:01 +0000]
brcm80211: fmac: change function proto_hdrpull parameter

Change parameter to device pointer for bus layer interface function
brcmf_proto_hdrpull This is part of the fullmac bus interface
refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: change function rx_frame parameter
Franky Lin [Sat, 17 Dec 2011 02:37:00 +0000]
brcm80211: fmac: change function rx_frame parameter

Change parameter to device pointer for bus layer interface function
brcmf_rx_frame. This is part of the fullmac bus interface
refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: change function brcmf_c_prec_enq parameter
Franky Lin [Sat, 17 Dec 2011 02:36:59 +0000]
brcm80211: fmac: change function brcmf_c_prec_enq parameter

Change parameter to device pointer for bus layer interface function
brcmf_c_prec_enq. This is part of the fullmac bus interface
refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: change function txcomplete parameter
Franky Lin [Sat, 17 Dec 2011 02:36:58 +0000]
brcm80211: fmac: change function txcomplete parameter

Change parameter to device pointer for bus layer interface function
brcmf_txcomplete. This is part of the fullmac bus interface
refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: change function txflowcontrol parameter
Franky Lin [Sat, 17 Dec 2011 02:36:57 +0000]
brcm80211: fmac: change function txflowcontrol parameter

Change parameter to device pointer for bus layer interface function
brcmf_txflowcontrol. This is part of the fullmac bus interface
refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: change function brcmf_detach parameter
Franky Lin [Sat, 17 Dec 2011 02:36:56 +0000]
brcm80211: fmac: change function brcmf_detach parameter

Change parameter to device pointer for bus layer interface function
brcmf_detach. This is part of the fullmac bus interface refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: change function bus_start parameter
Franky Lin [Sat, 17 Dec 2011 02:36:55 +0000]
brcm80211: fmac: change function bus_start parameter

Change parameter to device pointer for bus layer interface function
brcmf_bus_start. This is part of the fullmac bus interface
refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: change function add_if parameter
Franky Lin [Sat, 17 Dec 2011 02:36:54 +0000]
brcm80211: fmac: change function add_if parameter

Change parameter to device pointer for bus layer interface function
brcmf_add_if. This is part of the fullmac bus interface
refactoring.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: register a dummy irq handler for SDIO function 2
Franky Lin [Sat, 17 Dec 2011 02:36:53 +0000]
brcm80211: fmac: register a dummy irq handler for SDIO function 2

When there is data available in dongle for driver to fetch, an
interrupt will be triggered and both SDIO function 1 and 2 will be
flagged by default in bcm4329. These two interrupt flags are
identical and only need to be handled once. Since there is no way
to turn off one flag from the dongle side, a dummy handler for
function 2 interrupt is placed.

Reported-by: Denis 'GNUtoo' Carikli <GNUtoo@no-log.org>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: remove oneline proto block functions
Franky Lin [Sat, 17 Dec 2011 02:36:52 +0000]
brcm80211: fmac: remove oneline proto block functions

brcmf_os_proto_block and brcmf_os_proto_unblock are oneline
functions handling proto_block mutex. Place the mutex interface
call inline to increase readability.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agobrcm80211: fmac: unify common layer driver data structure
Franky Lin [Sat, 17 Dec 2011 02:36:51 +0000]
brcm80211: fmac: unify common layer driver data structure

No need to split data structure for common layer into brcmf_pub and
brcmf_info. Absorb brcmf_info into brcmf_pub to increase code
readability.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Arend van Spriel <arend@broadcom.com>
Reviewed-by: Alwin Beukers <alwin@broadcom.com>
Signed-off-by: Franky Lin <frankyl@broadcom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agoNFC: use kfree_skb() for sk_buffs
Dan Carpenter [Fri, 16 Dec 2011 20:26:00 +0000]
NFC: use kfree_skb() for sk_buffs

This is a struct sk_buff pointer and it should be freed with kfree_skb()
instead of kfree().

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Samuel Ortiz <sameo@linux.intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agoNFC: double unlock in nfc_llcp_recv_connect()
Dan Carpenter [Fri, 16 Dec 2011 20:25:29 +0000]
NFC: double unlock in nfc_llcp_recv_connect()

We unlock inside the if block on the other side of this if else
statement.  It could result in calling mutex_unlock() twice.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Samuel Ortiz <sameo@linux.intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

7 years agob43: avoid packet losses in the dma worker code.
francesco.gringoli@ing.unibs.it [Fri, 16 Dec 2011 17:34:56 +0000]
b43: avoid packet losses in the dma worker code.

Following Rafal request, we verified that on "modern" CPUs using one
or more workers is equivalent. Here is patch V3 that addresses the
packet loss bug in the dma engine using only one worker.

-------

This patch addresses a bug in the dma worker code that keeps draining
packets even when the hardware queues are full. In such cases packets
can not be passed down to the device and are erroneusly dropped by the
code.

This problem was already discussed here

http://www.mail-archive.com/b43-dev@lists.infradead.org/msg01413.html

and acknowledged by Michael.

Number of hardware queues is now defined in b43.h (B43_QOS_QUEUE_NUM).

Acknowledgements to Riccardo Paolillo <riccardo.paolillo@gmail.com> and
Michele Orru <michele.orru@hotmail.it>

Signed-off-by: Francesco Gringoli <francesco.gringoli@ing.unibs.it>
Signed-off-by: John W. Linville <linville@tuxdriver.com>