firefly-linux-kernel-4.4.55.git
13 years agoath: fix NULL pointer dereference on reg_notifier()
Luis R. Rodriguez [Thu, 16 Dec 2010 03:24:12 +0000 (19:24 -0800)]
ath: fix NULL pointer dereference on reg_notifier()

The reg_notifier() was recently updated as being capable of
having the request passed as NULL, fix ath to follow this API
change. Without this we end up oopsing:

BUG: unable to handle kernel NULL pointer dereference at 0000000000000004
IP: [<ffffffffa02fb8cb>] ath_reg_notifier_apply+0x5b/0xa0 [ath]
PGD b4c4c067 PUD b4c4d067 PMD 0
Oops: 0000 [#1] SMP DEBUG_PAGEALLOC
last sysfs file: /sys/devices/pci0000:00/0000:00:1b.0/uevent
CPU 1
Modules linked in: <etc>
Pid: 436, comm: modprobe Not tainted 2.6.37-rc5-wl+ #36 6460DWU/6460DWU
RIP: 0010:[<ffffffffa02fb8cb>]  [<ffffffffa02fb8cb>] ath_reg_notifier_apply+0x5b/0xa0 [ath]
RSP: 0018:ffff8800b6f6baa8  EFLAGS: 00010246
RAX: ffff8800b527b254 RBX: ffff8800b532c180 RCX: 0000000000000018
RDX: ffff8800b530c108 RSI: 0000000000000000 RDI: ffff8800b532c180
RBP: ffff8800b6f6baa8 R08: ffff8800b532f268 R09: 0000000000000235
R10: 00000000000016ad R11: 0000000000000018 R12: 0000000000000000
R13: 0000000000000016 R14: ffff8800b532f268 R15: 0000000000000011
FS:  00007f0c53104700(0000) GS:ffff8800bed00000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 0000000000000004 CR3: 00000000b6531000 CR4: 00000000000006e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process modprobe (pid: 436, threadinfo ffff8800b6f6a000, task ffff8800b404dc40)
Stack:
 ffff8800b6f6bac8 ffffffffa03ea651 ffff8800b532c180 ffff8800b527b254
 ffff8800b6f6bb38 ffffffffa01835ca ffffffffa019ed00 00000000a019ed80
 0000000000000002 ffff880000000002 ffffffffa0366140 0000000010aee572
Call Trace:
 [<ffffffffa03ea651>] ath9k_reg_notifier+0x41/0x50 [ath9k]
 [<ffffffffa01835ca>] wiphy_update_regulatory+0x4ba/0x5a0 [cfg80211]
 [<ffffffffa0366140>] ? ieee80211_register_hw+0xa0/0x5b0 [mac80211]
 [<ffffffffa0366140>] ? ieee80211_register_hw+0xa0/0x5b0 [mac80211]
 [<ffffffffa017f994>] wiphy_register+0x1d4/0x360 [cfg80211]
 [<ffffffff8114b918>] ? __kmalloc+0x108/0x1c0
 [<ffffffffa0366223>] ieee80211_register_hw+0x183/0x5b0 [mac80211]
 [<ffffffffa03eb49b>] ath9k_init_device+0x66b/0x850 [ath9k]
 [<ffffffffa03f9dd6>] ath_pci_probe+0x2f6/0x3c0 [ath9k]
 [<ffffffff81037529>] ? default_spin_lock_flags+0x9/0x10
 [<ffffffff812e19cf>] local_pci_probe+0x5f/0xd0
 [<ffffffff812e2bf1>] pci_device_probe+0x101/0x120
 [<ffffffff81390aca>] ? driver_sysfs_add+0x7a/0xb0
 [<ffffffff81390c26>] driver_probe_device+0x96/0x1c0
 [<ffffffff81390deb>] __driver_attach+0x9b/0xa0
 [<ffffffff81390d50>] ? __driver_attach+0x0/0xa0
 [<ffffffff81390008>] bus_for_each_dev+0x68/0x90
 [<ffffffff81390a4e>] driver_attach+0x1e/0x20
 [<ffffffff81390309>] bus_add_driver+0xe9/0x290
 [<ffffffffa0407000>] ? ath9k_init+0x0/0x4d [ath9k]
 [<ffffffff81391130>] driver_register+0x80/0x150
 [<ffffffffa0407000>] ? ath9k_init+0x0/0x4d [ath9k]
 [<ffffffffa0407000>] ? ath9k_init+0x0/0x4d [ath9k]
 [<ffffffff812e2e76>] __pci_register_driver+0x56/0xd0
 [<ffffffffa03f9ec3>] ath_pci_init+0x23/0x30 [ath9k]
 [<ffffffffa040702b>] ath9k_init+0x2b/0x4d [ath9k]
 [<ffffffff81002053>] do_one_initcall+0x43/0x190
 [<ffffffff8109fb5b>] sys_init_module+0xbb/0x200
 [<ffffffff8100c042>] system_call_fastpath+0x16/0x1b
Code: <who even reads this anyway? haha, ok you do>
RIP  [<ffffffffa02fb8cb>] ath_reg_notifier_apply+0x5b/0xa0 [ath]
 RSP <ffff8800b6f6baa8>
CR2: 0000000000000004
---[ end trace 6d03d3c7eda9f06b ]---

Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agocfg80211: fix null pointer dereference with a custom regulatory request
Luis R. Rodriguez [Thu, 16 Dec 2010 03:24:11 +0000 (19:24 -0800)]
cfg80211: fix null pointer dereference with a custom regulatory request

Once we moved the core regulatory request to the queue and let
the scheduler process it last_request will have been left NULL
until the schedular decides to process the first request. When
this happens and we are loading a driver with a custom regulatory
request like all Atheros drivers we end up with a NULL pointer
dereference. We fix this by checking if the request was a
custom one.

BUG: unable to handle kernel NULL pointer dereference at 0000000000000004
IP: [<ffffffffa016de87>] freq_reg_info_regd.clone.2+0x27/0x130 [cfg80211]
PGD 71f91067 PUD 712b2067 PMD 0
Oops: 0000 [#1] PREEMPT SMP
last sysfs file: /sys/devices/pci0000:00/0000:00:1d.7/usb2/2-1/firmware/2-1/loading
CPU 0
Modules linked in: ath9k_htc(+) ath9k_common ath9k_hw ath <etc>
Pid: 3094, comm: insmod Tainted: G        W   2.6.37-rc5-wl #16 INVALID/28427ZQ
RIP: 0010:[<ffffffffa016de87>]  [<ffffffffa016de87>] freq_reg_info_regd.clone.2+0x27/0x130 [cfg80211]
RSP: 0018:ffff88007045db78  EFLAGS: 00010282
RAX: 0000000000000000 RBX: ffffffffa047d9a0 RCX: ffff88007045dbd0
RDX: 0000000000004e20 RSI: 000000000024cde0 RDI: ffff8800700483e0
RBP: ffff88007045db98 R08: ffffffffa02f5b40 R09: 0000000000000001
R10: 000000000000000e R11: 0000000000000001 R12: 0000000000000000
R13: ffff88007004e3b0 R14: 0000000000000000 R15: ffff880070048340
FS:  00007f635a707700(0000) GS:ffff880077400000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 0000000000000004 CR3: 00000000708a9000 CR4: 00000000000006f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process insmod (pid: 3094, threadinfo ffff88007045c000, task ffff8800713e3ec0)
Stack:
 ffffffffa047d9a0 0000000000000000 ffff88007004e3b0 0000000000000000
 ffff88007045dc08 ffffffffa016e147 000000007045dc08 0000000000000002
 ffff8800700483e0 ffffffffa02f5b40 ffff88007045dbd8 0000000000000000
Call Trace:
 [<ffffffffa016e147>] wiphy_apply_custom_regulatory+0x137/0x1d0 [cfg80211]
 [<ffffffffa047a690>] ? ath9k_reg_notifier+0x0/0x50 [ath9k_htc]
 [<ffffffffa02f47f7>] ath_regd_init+0x347/0x430 [ath]
 [<ffffffffa047b1f5>] ath9k_htc_probe_device+0x6c5/0x960 [ath9k_htc]
 [<ffffffffa0472a2c>] ath9k_htc_hw_init+0xc/0x30 [ath9k_htc]
 [<ffffffffa04747e6>] ath9k_hif_usb_probe+0x216/0x3b0 [ath9k_htc]
 [<ffffffffa03bb6bc>] usb_probe_interface+0x10c/0x210 [usbcore]
 [<ffffffff812aec26>] driver_probe_device+0x96/0x1c0
 [<ffffffff812aedf3>] __driver_attach+0xa3/0xb0
 [<ffffffff812aed50>] ? __driver_attach+0x0/0xb0
 [<ffffffff812adaae>] bus_for_each_dev+0x5e/0x90
 [<ffffffff812ae8c9>] driver_attach+0x19/0x20
 [<ffffffff812ae438>] bus_add_driver+0x168/0x320
 [<ffffffff812af071>] driver_register+0x71/0x140
 [<ffffffff811fc4a8>] ? __raw_spin_lock_init+0x38/0x70
 [<ffffffffa03ba39c>] usb_register_driver+0xdc/0x190 [usbcore]
 [<ffffffffa03a2000>] ? ath9k_htc_init+0x0/0x4f [ath9k_htc]
 [<ffffffffa047499e>] ath9k_hif_usb_init+0x1e/0x20 [ath9k_htc]
 [<ffffffffa03a202b>] ath9k_htc_init+0x2b/0x4f [ath9k_htc]
 [<ffffffff8100212f>] do_one_initcall+0x3f/0x180
 [<ffffffff8109ef5b>] sys_init_module+0xbb/0x200
 [<ffffffff8100bf52>] system_call_fastpath+0x16/0x1b
Code: <etc, who cares>
RIP  [<ffffffffa016de87>] freq_reg_info_regd.clone.2+0x27/0x130 [cfg80211]
 RSP <ffff88007045db78>
CR2: 0000000000000004
---[ end trace 79e4193601c8b713 ]---

Reported-by: Sujith Manoharan <Sujith.Manoharan@atheros.com>
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: Configure appropriate Tx power when PAPRD fails
Vasanthakumar Thiagarajan [Wed, 15 Dec 2010 15:30:53 +0000 (07:30 -0800)]
ath9k_hw: Configure appropriate Tx power when PAPRD fails

Target Tx power available in eeprom is for PAPRD. If PAPRD
fails, paprd scale factor needs to be detected from this
target tx power.

Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: Disable PAPRD for rates with low Tx power
Vasanthakumar Thiagarajan [Wed, 15 Dec 2010 15:30:52 +0000 (07:30 -0800)]
ath9k_hw: Disable PAPRD for rates with low Tx power

When the drop in Tx power for a particular mcs rate exceeds
the paprd scale factor, paprd may not work properly. Disable
paprd for any such rates.

Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: Add a helper to get paprd scale factor
Vasanthakumar Thiagarajan [Wed, 15 Dec 2010 15:30:51 +0000 (07:30 -0800)]
ath9k_hw: Add a helper to get paprd scale factor

Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: Tx IQ cal changes for AR9003
Vasanthakumar Thiagarajan [Wed, 15 Dec 2010 15:30:50 +0000 (07:30 -0800)]
ath9k_hw: Tx IQ cal changes for AR9003

Add multiple Tx IQ cal support to improve EVM accross
different power levels.

Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: Move get_streams() to hw.h
Vasanthakumar Thiagarajan [Wed, 15 Dec 2010 15:30:49 +0000 (07:30 -0800)]
ath9k_hw: Move get_streams() to hw.h

This helper can be used in multiple places. Also make
it inline returning u8.

Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: Remove unnecessary Rx IQ cal register configuration in ar9003_hw_tx_iq_cal()
Vasanthakumar Thiagarajan [Wed, 15 Dec 2010 15:30:48 +0000 (07:30 -0800)]
ath9k_hw: Remove unnecessary Rx IQ cal register configuration in ar9003_hw_tx_iq_cal()

Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: Remove delay during regwrite of analog shift registers
Vasanthakumar Thiagarajan [Wed, 15 Dec 2010 15:30:47 +0000 (07:30 -0800)]
ath9k_hw: Remove delay during regwrite of analog shift registers

This is not needed for AR9003.

Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agonl80211: Add notification for dropped Deauth/Disassoc
Jouni Malinen [Wed, 15 Dec 2010 22:52:40 +0000 (00:52 +0200)]
nl80211: Add notification for dropped Deauth/Disassoc

Add a new notification to indicate that a received, unprotected
Deauthentication or Disassociation frame was dropped due to
management frame protection being in use. This notification is
needed to allow user space (e.g., wpa_supplicant) to implement
SA Query procedure to recover from association state mismatch
between an AP and STA.

This is needed to avoid getting stuck in non-working state when MFP
(IEEE 802.11w) is used and a protected Deauthentication or
Disassociation frame is dropped for any reason. After that, the
station would silently discard any unprotected Deauthentication or
Disassociation frame that could be indicating that the AP does not
have association for the STA (when the Reason Code would be 6 or 7).
IEEE Std 802.11w-2009, 11.13 describes this recovery mechanism.

Signed-off-by: Jouni Malinen <j@w1.fi>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoMerge branch 'wl12xx-next' of git://git.kernel.org/pub/scm/linux/kernel/git/luca...
John W. Linville [Thu, 16 Dec 2010 20:21:33 +0000 (15:21 -0500)]
Merge branch 'wl12xx-next' of git://git./linux/kernel/git/luca/wl12xx

13 years agoiwlwifi: Fix error: struct iwl_lq_sta has no member named dbg_fixed_rate
Sedat Dilek [Thu, 16 Dec 2010 11:46:23 +0000 (12:46 +0100)]
iwlwifi: Fix error: struct iwl_lq_sta has no member named dbg_fixed_rate

While compiling linux-next (next-20101216) I fell over this breakage:
...
drivers/net/wireless/iwlwifi/iwl-agn-rs.c: In function ‘iwl_rs_rate_init’:
drivers/net/wireless/iwlwifi/iwl-agn-rs.c:2876:8: error: ‘struct iwl_lq_sta’ has no member named ‘dbg_fixed_rate’

dbg_fixed_rate is only used when CONFIG_MAC80211_DEBUGFS is set:

[ drivers/net/wireless/iwlwifi/iwl-agn-rs.h ]
...
 #ifdef CONFIG_MAC80211_DEBUGFS
       struct dentry *rs_sta_dbgfs_scale_table_file;
       struct dentry *rs_sta_dbgfs_stats_table_file;
       struct dentry *rs_sta_dbgfs_rate_scale_data_file;
       struct dentry *rs_sta_dbgfs_tx_agg_tid_en_file;
       u32 dbg_fixed_rate;
 #endif

The issue was introduced by commit a1da077bc36368eb7d6312e7e49260f0a3d92c77:
"iwlwifi: clear dbg_fixed_rate during init"

Signed-off-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agortlwifi: convert to __packed notation
John W. Linville [Thu, 16 Dec 2010 14:20:16 +0000 (09:20 -0500)]
rtlwifi: convert to __packed notation

Use "__packed" instead of "__attribute__ ((packed))"...

Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agowireless:mac80211: kill unuse macro MESH_CFG_CMP_LEN in mesh.h
Shan Wei [Wed, 15 Dec 2010 06:27:28 +0000 (14:27 +0800)]
wireless:mac80211: kill unuse macro MESH_CFG_CMP_LEN in mesh.h

Commit 00d3f14c has removed the references of this macro,
but left it only. So remove this definition.

commit 00d3f14cf9f12c21428121026a5e1d5f65926447
Author: Johannes Berg <johannes@sipsolutions.net>
Date:   Tue Feb 10 21:26:00 2009 +0100

    mac80211: use cfg80211s BSS infrastructure

    Remove all the code from mac80211 to keep track of BSSes
    and use the cfg80211-provided code completely.

Signed-off-by: Shan Wei <shanwei@cn.fujitsu.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: Add comments for making pm-qos as modparam
Mohammed Shafi Shajakhan [Wed, 15 Dec 2010 07:32:46 +0000 (13:02 +0530)]
ath9k: Add comments for making pm-qos as modparam

PM-QOS value can be user specified via module parameter.
This patch adds few comments regarding this in the driver
code.

Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_htc: Remove PCI specific configuration
Sujith Manoharan [Wed, 15 Dec 2010 02:17:23 +0000 (07:47 +0530)]
ath9k_htc: Remove PCI specific configuration

This is not required for USB devices.

Signed-off-by: Sujith Manoharan <Sujith.Manoharan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomac80211: Add timeout to BA session start API
Sujith Manoharan [Wed, 15 Dec 2010 02:17:10 +0000 (07:47 +0530)]
mac80211: Add timeout to BA session start API

Allow drivers or rate control algorithms to specify BlockAck session
timeout when initiating an ADDBA transaction. This is useful in cases
where maintaining persistent BA sessions does not incur any overhead.

The current timeout value of 5000 TUs is retained for all non ath9k/ath9k_htc
drivers.

Signed-off-by: Sujith Manoharan <Sujith.Manoharan@atheros.com>
Reviewed-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agonl80211: advertise maximum remain-on-channel duration
Johannes Berg [Tue, 14 Dec 2010 16:54:28 +0000 (17:54 +0100)]
nl80211: advertise maximum remain-on-channel duration

With the upcoming hardware offload implementation,
some devices will have a different maximum duration
for the remain-on-channel command. Advertise the
maximum duration in mac80211, and make mac80211 set
it.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoiwlagn: fix witespace damage
Wey-Yi Guy [Tue, 14 Dec 2010 15:38:58 +0000 (07:38 -0800)]
iwlagn: fix witespace damage

patch "iwlagn: check ready in iwlagn_bss_info_changed()" introduce
whitespace, fix it

please merge with the previous patch

Reported 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>
13 years agoath9k: Make PM-QOS value as user configurable
Mohammed Shafi Shajakhan [Tue, 14 Dec 2010 07:48:28 +0000 (13:18 +0530)]
ath9k: Make PM-QOS value as user configurable

This patch allows the pm-qos value to be user configurable by making it as
a module parameter.This will help our customers to configure the pm-qos
value according to the effect in throughput due to the DMA latency problem
which was observed in Intel Pinetrail platforms.
The tested value of '55' will be filled as the default
pm-qos-value incase the user does not specifies pm-qos value as a
module parameter.
example usage: sudo modprobe ath9k pmqos=65

Cc: Senthilkumar Balasubramanian <Senthilkumar.Balasubramanian@Atheros.com>
Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoMerge branch 'wireless-next-2.6' of git://git.kernel.org/pub/scm/linux/kernel/git...
John W. Linville [Wed, 15 Dec 2010 21:37:55 +0000 (16:37 -0500)]
Merge branch 'wireless-next-2.6' of git://git./linux/kernel/git/iwlwifi/iwlwifi-2.6

Conflicts:
drivers/net/wireless/iwlwifi/iwl-1000.c
drivers/net/wireless/iwlwifi/iwl-6000.c
drivers/net/wireless/iwlwifi/iwl-agn-eeprom.c
drivers/net/wireless/iwlwifi/iwl-core.h

13 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
John W. Linville [Wed, 15 Dec 2010 21:33:28 +0000 (16:33 -0500)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-2.6

Conflicts:
drivers/net/wireless/iwlwifi/iwl-1000.c
drivers/net/wireless/iwlwifi/iwl-6000.c
drivers/net/wireless/iwlwifi/iwl-core.h

13 years agortl8192ce: Add new driver
Larry Finger [Wed, 8 Dec 2010 17:12:31 +0000 (11:12 -0600)]
rtl8192ce: Add new driver

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agowl12xx_sdio_test: rename files to match current style
Luciano Coelho [Wed, 15 Dec 2010 14:10:12 +0000 (16:10 +0200)]
wl12xx_sdio_test: rename files to match current style

Change some file names and Kconfig settings so that this new module matches
the new way of using wl12xx instead of wl1271.

Also fix SDIO power enabling and disabling to match the latest way of doing
it.

Cc: Roger Quadros <roger.quadros@nokia.com>
Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
13 years agowl12xx: Change TX queue to be per AC
Juuso Oikarinen [Mon, 13 Dec 2010 07:52:37 +0000 (09:52 +0200)]
wl12xx: Change TX queue to be per AC

With the current single-queue implementation traffic priorization is not
working correctly - when using multiple BE streams and one, say VI stream,
the VI stream will share bandwidth almost equally with the BE streams.

To fix the issue, implement per AC queues, which are emptied in priority
order to the firmware. To keep it relatively simple, maintain a global
buffer count and global queue stop/wake instead of per-AC.

With these changes, priorization appears to work just fine.

Signed-off-by: Juuso Oikarinen <juuso.oikarinen@nokia.com>
Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
13 years agowl12xx: allow runtime changing of debug_level
Eliad Peller [Sun, 12 Dec 2010 10:15:35 +0000 (12:15 +0200)]
wl12xx: allow runtime changing of debug_level

Currently, the debug level is set in compilation time (by the DEBUG_LEVEL
const). This method has the advantage of compiling only the relevant
messages, while optimizing out the unused ones.

In order to allow runtime control over the debug_level, while optimizing
out messages when debug messages are not needed, we combine some methods:
1. use dynamic_debug (pr_debug) rather then printk.
2. add debug_level module param in order to set debug level during insmod.
3. add debug_level sysfs file in order to allow dynamic control over the
   debug level.

Since patches for pr_debug_hex_dump() implementation haven't been applied yet,
we are still temporarly using print_hex_dump().

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
13 years agowl1271: fixed problem with WPS IEs in probe requests
Guy Eilam [Thu, 9 Dec 2010 14:54:59 +0000 (16:54 +0200)]
wl1271: fixed problem with WPS IEs in probe requests

Inclusion of a WPS IE in probe requests caused a problem
in the driver due to the maximum size of the probe request
template and the max_scan_ie_len values at initialization.

Increased the size of probe request template
to the maximum size allowed by the firmware.
Struct wl12xx_probe_req_template, which was only used
for calculating the max size of the probe request template,
is no longer used and needed.

max_scan_ie_len is used for validating the size of
additional IEs in scan requests.
Initialized the max_scan_ie_len field to the maximum size
of the probe request template minus the ieee80211 header size.

Signed-off-by: Guy Eilam <guy@wizery.com>
Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
13 years agowl12xx: add auto-arp support
Eliad Peller [Thu, 9 Dec 2010 09:31:27 +0000 (11:31 +0200)]
wl12xx: add auto-arp support

The auto-arp feature of wl12xx allows the firmware to automatically
response to arp requests asking for its ip.

in order to use it, we configure the arp response template and
enable the corresponding bit in wl1271_acx_arp_filter (along with
passing its ip)

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
13 years agowl1271_sdio_test: Add module for sdio RX/TX testing
Roger Quadros [Wed, 1 Dec 2010 09:58:54 +0000 (11:58 +0200)]
wl1271_sdio_test: Add module for sdio RX/TX testing

This module enables individually generating RX and TX traffic
over the SDIO bus on which the WL1271 chipset is connected.
This is required to perform RF interference testing.

The module takes 2 module parameters 'rx' and 'tx'.

To generate RX traffic:
modprobe wl1271_sdio_test rx=1
To generate TX traffic:
modprobe wl1271_sdio_test tx=1

To generate both RX & TX traffic, set both rx and tx to 1.

You can change the testing configuration at runtime by changing
the rx & tx values at /sys/modules/wl1271_sdio_test/

To stop testing simply unload the module.

Signed-off-by: Roger Quadros <roger.quadros@nokia.com>
Reviewed-by: Carlos Chinea <carlos.chinea@nokia.com>
Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
13 years agowl12xx: disable 11a channels when wl->enable_11a is known
Luciano Coelho [Fri, 3 Dec 2010 15:05:40 +0000 (17:05 +0200)]
wl12xx: disable 11a channels when wl->enable_11a is known

Disabling the 11a channels when not supported in the reg_notify function was
not working as it should, because when the driver is initiailizing (and
registering itself with mac80211), it would get the reg notification too
early.  At that point the driver wouldn't have received the NVS yet, so it
wouldn't know whether 11a was supported.

To fix this, we disable 11a channels when we read the NVS instead.  Also, it
is easier (and still safe) to set n_channels to zero instead of setting the
disabled flag on every 11a channel.

Signed-off-by: Luciano Coelho <luciano.coelho@nokia.com>
13 years agop54usb: New USB ID for Gemtek WUBI-100GW
Eduardo Costa [Tue, 14 Dec 2010 20:37:59 +0000 (14:37 -0600)]
p54usb: New USB ID for Gemtek WUBI-100GW

This USB ID is for the WUBI-100GW 802.11g Wireless LAN USB Device that
uses p54usb.

Signed-off-by: Larry Finger <Larry.Finger@lwfinger.net>
Signed-off-by: Eduardo Costa <ecosta.tmp@gmail.com>
Cc: Stable <stable@kernel.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoiwlagn: code clean up to remove duplicate code
Wey-Yi Guy [Wed, 8 Dec 2010 15:51:50 +0000 (07:51 -0800)]
iwlagn: code clean up to remove duplicate code

Multiple devices use almost the same .cfg with minor differences.
Use macro and remove the duplication. By doing this, reduce
the chance for mistake while modify .cfg parameters

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
13 years agoiwlwifi: keep track of bt coex enable/disable stage
Wey-Yi Guy [Wed, 8 Dec 2010 23:34:52 +0000 (15:34 -0800)]
iwlwifi: keep track of bt coex enable/disable stage

For debugging purpose, keep track of the bt coex enable/disable state.

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
13 years agoiwlwifi: do not reload fw if WiMAX own the RF
Wey-Yi Guy [Tue, 7 Dec 2010 16:06:31 +0000 (08:06 -0800)]
iwlwifi: do not reload fw if WiMAX own the RF

For WiFi/WiMAX combo devices, if WiMAX own the RF, WiFi driver
try to access RF and fail. This is the W/A to To avoid WiFi keep
reloading firmware and try to access RF again.

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
13 years agoiwlagn: More detail tx power logging
Wey-Yi Guy [Fri, 3 Dec 2010 18:33:36 +0000 (10:33 -0800)]
iwlagn: More detail tx power logging

For enhanced tx power table in EEPROM, add more detail logging
to help debugging

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
13 years agoiwlwifi: use IWL_DEBUG_EEPROM for EEPROM related info
Wey-Yi Guy [Fri, 3 Dec 2010 18:33:35 +0000 (10:33 -0800)]
iwlwifi: use IWL_DEBUG_EEPROM for EEPROM related info

For logging EEPROM related info, instead of using IWL_DEBUG_INFO,
use the dedicated logging (IWL_DEBUG_EEPROM) for easier debugging

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
13 years agoiwlwifi: add new EEPROM debug log type
Wey-Yi Guy [Fri, 3 Dec 2010 18:33:34 +0000 (10:33 -0800)]
iwlwifi: add new EEPROM debug log type

Adding new debug type to log EEPROM related data

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
13 years agoiwlagn: remove old EEPROM TX power reading
Johannes Berg [Thu, 9 Dec 2010 20:56:42 +0000 (12:56 -0800)]
iwlagn: remove old EEPROM TX power reading

This removes the old TX power reading code, it isn't
necessary since the new code is able to read all the
various EEPROM layouts due to relying on information
contained in the EEPROM.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
13 years agoiwlagn: implement layout-agnostic EEPROM reading
Johannes Berg [Thu, 9 Dec 2010 17:30:14 +0000 (09:30 -0800)]
iwlagn: implement layout-agnostic EEPROM reading

The current EEPROM reading code has some layout
assumptions that now turned out to be false with
some newer versions of the EEPROM. Luckily, we
can avoid all such assumptions by using data in
the EEPROM itself, so implement using that.

However, for risk mitigation purposes, keep the
old reading code for current hardware for now.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
13 years agoiwlagn: rename enhanced txpower fields
Johannes Berg [Tue, 30 Nov 2010 19:03:44 +0000 (11:03 -0800)]
iwlagn: rename enhanced txpower fields

Some fields we didn't previously use from the
enhanced TX power structure will be needed in
the next patch, so rename them to their correct
names to be able to use them and change code
reading them accordingly.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
13 years agoiwlagn: rx antenna diversity
Wey-Yi Guy [Mon, 6 Dec 2010 19:51:39 +0000 (11:51 -0800)]
iwlagn: rx antenna diversity

For the new 1x1 devices, hw and uCode will support rx
antenna diversity, but we need to indicate 1x1 device to
AccessPoint to make sure it won't use MIMO.

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
13 years agoiwlagn: fix debug variable access
Johannes Berg [Tue, 30 Nov 2010 21:24:36 +0000 (13:24 -0800)]
iwlagn: fix debug variable access

The compiler correctly warns:
iwl-agn-tx.c: In function ‘iwlagn_tx_status_reply_compressed_ba’:
iwl-agn-tx.c:1240: warning: ‘bitmap’ may be used uninitialized in this function

Move the debug print to the branch that reads the
bitmap, and move the variables too so it's more
obvious where they are needed.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
13 years agoiwlagn: check ready in iwlagn_bss_info_changed()
Shanyu Zhao [Thu, 2 Dec 2010 19:02:28 +0000 (11:02 -0800)]
iwlagn: check ready in iwlagn_bss_info_changed()

In function iwlagn_bss_info_changed(), we need to check if the driver
is ready before doing real work. Also, the previously put WARN() is
removed because the vif is not guaranteed to be valid. uCode restart
routine will clear the vif.

Signed-off-by: Shanyu Zhao <shanyu.zhao@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
13 years agoiwlwifi: clear dbg_fixed_rate during init
Shanyu Zhao [Thu, 2 Dec 2010 19:02:54 +0000 (11:02 -0800)]
iwlwifi: clear dbg_fixed_rate during init

This prevent bad fixed_rate keeps crashing uCode in an endless loop.

Signed-off-by: Shanyu Zhao <shanyu.zhao@intel.com>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
13 years agoiwlagn: change led compensation for 6005 and 6030 devices
Wey-Yi Guy [Mon, 29 Nov 2010 18:45:16 +0000 (10:45 -0800)]
iwlagn: change led compensation for 6005 and 6030 devices

For both 6005 and 6030 devices, change the led compensation to 57

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
13 years agort2x00: Fix firmware loading regression on x86_64.
Wolfgang Kufner [Mon, 13 Dec 2010 11:39:39 +0000 (12:39 +0100)]
rt2x00: Fix firmware loading regression on x86_64.

Commit 6175ddf06b6172046a329e3abfd9c901a43efd2e changes the way
memcpy_toio() works for x86_64, causing firmware loading to fail for
some Ralink WLAN devices with the rt2800pci driver since linux 2.6.34.
This causes the log message: "phy0 -> rt2800pci_load_firmware: Error -
PBF system register not ready.".

Fix this by using __iowrite32_copy instead of memcpy_toio().

Signed-off-by: Wolfgang Kufner <wolfgang.kufner@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Pad beacon to multiple of 32 bits.
Wolfgang Kufner [Mon, 13 Dec 2010 11:39:12 +0000 (12:39 +0100)]
rt2x00: Pad beacon to multiple of 32 bits.

Pad beacon to a multiple of 32 bits in preparation for the change
from memcpy_toio() to __iowrite32_copy() in register_multiwrite().

Signed-off-by: Wolfgang Kufner <wolfgang.kufner@gmail.com>
Acked-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: remove stray semicolon
Johannes Stezenbach [Mon, 13 Dec 2010 11:38:49 +0000 (12:38 +0100)]
rt2x00: remove stray semicolon

The stray semicolon after DEBUG_PRINTK_MSG causes things
like "if (...) WARNING(...); else {}" to fail with syntax error.

Signed-off-by: Johannes Stezenbach <js@sig21.net>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Fix WMM Queue naming
Ivo van Doorn [Mon, 13 Dec 2010 11:36:38 +0000 (12:36 +0100)]
rt2x00: Fix WMM Queue naming

The Queue names were incorrectly copied from the legacy drivers,
as a result the queue names were inversed to what was expected.

This renames the queues using this mapping:
QID_AC_BK -> QID_AC_VO (priority 0)
QID_AC_BE -> QID_AC_VI (priority 1)
QID_AC_VI -> QID_AC_BE (priority 2)
QID_AC_VO -> QID_AC_BK (priority 3)

Note that this was a naming problem only, which didn't affect
the assignment of frames to their respective queues.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Introduce extra queue entry sanity flag
Ivo van Doorn [Mon, 13 Dec 2010 11:36:18 +0000 (12:36 +0100)]
rt2x00: Introduce extra queue entry sanity flag

Add a queue entry flag ENTRY_DATA_STATUS_PENDING,
which can be used to indicate a queue entry has
returned from the hardware and is waiting for
status processing. Using this flag we can add
some extra sanity checks to prevent queue corruption.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Acked-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Cleanup RX index counting
Ivo van Doorn [Mon, 13 Dec 2010 11:36:00 +0000 (12:36 +0100)]
rt2x00: Cleanup RX index counting

Add the rt2x00_dmastart function to rt2x00lib which
marks the queue_entry as "owned by device", and increased
the Q_INDEX number.

This cleanups up the index handling by rt2x00lib which
at until so far used hackish approaches to keep the
RX queue index numbering sane.

The rt2x00pci.c changes are from Helmut Schaa

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Add "flush" queue command
Ivo van Doorn [Mon, 13 Dec 2010 11:35:40 +0000 (12:35 +0100)]
rt2x00: Add "flush" queue command

Add a new command to the queue handlers: "flush",
this moves the flush() callback from mac80211
into rt2x00queue and adds support for flushing
the RX queue as well.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Acked-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Protect queue control with mutex
Ivo van Doorn [Mon, 13 Dec 2010 11:35:17 +0000 (12:35 +0100)]
rt2x00: Protect queue control with mutex

Add wrapper functions in rt2x00queue.c to
start & stop queues. This control must be protected
using a mutex.

Queues can also be paused which will halt the flow
of packets between the driver and mac80211. This doesn't
require a mutex protection.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Reorganize queue callback functions
Ivo van Doorn [Mon, 13 Dec 2010 11:34:54 +0000 (12:34 +0100)]
rt2x00: Reorganize queue callback functions

As part of the queue refactoring, change the queue callback
function names to have 3 different actions: start, kick & stop.

We can now also remove the STATE_RADIO_RX_ON/STATE_RADIO_RX_OFF
device_state flags, and replace the usage with using the
start_queue/stop_queue callback functions.
This streamlines the RX queue handling to the
similar approach as all other queues.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Acked-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Introduce 3 queue commands in drivers (start, kick, stop).
Ivo van Doorn [Mon, 13 Dec 2010 11:34:22 +0000 (12:34 +0100)]
rt2x00: Introduce 3 queue commands in drivers (start, kick, stop).

As part of the queue refactoring, we now introduce
3 queue commands: start, kick, stop.

 - Start: will enable a queue, for TX this will
   not mean anything, while for beacons and RX
   this will update the registers to enable the queue.
 - Kick: This will kick all pending frames to
   the hardware. This is needed for the TX queue
   to push all frames to the HW after the queue
   has been started
 - Stop: This will stop the queue in the hardware,
   and cancel any pending work (So this doesn't
   mean the queue is empty after a stop!).

Move all code from the drivers into the appropriate
functions, and link those calls to the old rt2x00lib
callback functions (we will fix this later when we
refactor the queue control inside rt2x00lib).

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Acked-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: trivial: add missing \n on warnings
Johannes Stezenbach [Mon, 13 Dec 2010 11:34:00 +0000 (12:34 +0100)]
rt2x00: trivial: add missing \n on warnings

Signed-off-by: Johannes Stezenbach <js@sig21.net>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Don't frequently reset beacon interval in AdHoc mode
Helmut Schaa [Mon, 13 Dec 2010 11:33:36 +0000 (12:33 +0100)]
rt2x00: Don't frequently reset beacon interval in AdHoc mode

Commit 0204464329c17ba6d293e1899f71223599a0e582 "Check for specific changed
flags when updating the erp config" changed the way in which a new beacon
interval gets handled. However, due to a bug in rt2800usb and rt2800pci the
beacon interval was reset during each scan, thus causing problems in AdHoc
mode.

Fix this by not cleaning up the beacon interval when killing the beacon queue
but just prevent the device from sending out beacons.

Reported-by: Wolfgang Kufner <wolfgang.kufner@gmail.com>
Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Ensure TX-ed frames are returned in the original state.
Gertjan van Wingerde [Mon, 13 Dec 2010 11:33:12 +0000 (12:33 +0100)]
rt2x00: Ensure TX-ed frames are returned in the original state.

Recent changes to the TX-done code of rt2x00 resulted in TX-ed frames not
being returned to mac80211 in the original state, and therefore with
insufficient headroom for re-transmissions.

Fix this by reverting the changes done and by ensuring we remove the inserted
L2pad by moving the header backwards instead of the data forwards.

At the same time also make sure that the rt2x00queue_remove_l2pad will not
move any memory when a frame has no data at all.

Signed-off-by: Gertjan van Wingerde <gwingerde@gmail.com>
Acked-by: Helmut Schaa <helmut.schaa@googlemail.com>
Cc: Jay Hung <Jay_Hung@ralinktech.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: fix hang when unplugging USB device in use
Johannes Stezenbach [Mon, 13 Dec 2010 11:32:49 +0000 (12:32 +0100)]
rt2x00: fix hang when unplugging USB device in use

When an rt2x00 USB device is unplugged while in use, it reliably
hangs the whole system.  After some time the watchdog prints:

BUG: soft lockup - CPU#0 stuck for 64s! [kworker/u:0:5]
...
[<c01a88d8>] (usb_submit_urb+0x0/0x2ac) from [<bf0e752c>] (rt2x00usb_kick_rx_entry+0xb4/0xe8 [rt2x00usb])
[<bf0e7478>] (rt2x00usb_kick_rx_entry+0x0/0xe8 [rt2x00usb]) from [<bf0e7588>] (rt2x00usb_clear_entry+x28/0x2c [rt2x00usb])
[<bf0e7560>] (rt2x00usb_clear_entry+0x0/0x2c [rt2x00usb]) from [<bf0d5bc4>] (rt2x00lib_rxdone+0x2e0/0x2f8 [rt2x00lib])
[<bf0d58e4>] (rt2x00lib_rxdone+0x0/0x2f8 [rt2x00lib]) from [<bf0e7e00>] (rt2x00usb_work_rxdone+0x54/0x74 [rt2x00usb])
[<bf0e7dac>] (rt2x00usb_work_rxdone+0x0/0x74 [rt2x00usb]) from [<c00542b4>] (process_one_work+0x20c/0x35c)

Clear the DEVICE_STATE_PRESENT flag when usb_submit_urb()
returns -ENODEV to fix this.

Signed-off-by: Johannes Stezenbach <js@sig21.net>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Add RF chip definition
RA-Jay Hung [Mon, 13 Dec 2010 11:32:22 +0000 (12:32 +0100)]
rt2x00: Add RF chip definition

Add RF chip definition

Signed-off-by: RA-Jay Hung <jay_hung@ralinktech.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Implement get_survey callback for rt2800
Helmut Schaa [Mon, 13 Dec 2010 11:31:58 +0000 (12:31 +0100)]
rt2x00: Implement get_survey callback for rt2800

Implement the get_survey callback to allow user space to read statistics
about the current channel condition.

Signed-off-by: Helmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: Add rt2800 EEPROM definition
RA-Jay Hung [Mon, 13 Dec 2010 11:31:27 +0000 (12:31 +0100)]
rt2x00: Add rt2800 EEPROM definition

Add and modify NIC Configuration and LED definition of EEPROM

Signed-off-by: RA-Jay Hung <jay_hung@ralinktech.com>
Acked-by: Gertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: update AR9003 initvals to improve carrier leak calibration/correction
Felix Fietkau [Mon, 13 Dec 2010 07:40:56 +0000 (08:40 +0100)]
ath9k_hw: update AR9003 initvals to improve carrier leak calibration/correction

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: update AR9003 initvals for improved radar detection
Felix Fietkau [Mon, 13 Dec 2010 07:40:55 +0000 (08:40 +0100)]
ath9k_hw: update AR9003 initvals for improved radar detection

Reduces the likelihood of false pulse detects in the hardware

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: fix PA predistortion training power selection
Felix Fietkau [Mon, 13 Dec 2010 07:40:54 +0000 (08:40 +0100)]
ath9k_hw: fix PA predistortion training power selection

The EEPROM contains scale factors for the tx power, which define
the range of allowable difference between target power and training
power. If the difference is too big, PA predistortion cannot be used.
For 2.4 GHz there is only one scale factor, for 5 GHz there are
three, depending on the specific frequency range.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: fix the PA predistortion rate mask
Felix Fietkau [Mon, 13 Dec 2010 07:40:53 +0000 (08:40 +0100)]
ath9k_hw: fix the PA predistortion rate mask

The EEPROM PAPRD rate mask fields only contain mask values for actual
rates in the low 25 bits. The upper bits are reserved for tx power
scale values. Add the proper mask definitions and use them before
writing the values to the register.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: fix PA predistortion thermal measurement handling
Felix Fietkau [Mon, 13 Dec 2010 07:40:52 +0000 (08:40 +0100)]
ath9k: fix PA predistortion thermal measurement handling

To be able to measure the thermal values correctly for PAPRD, we need
to send training frames before setting up the gain table for the measurement,
and then again afterwards for the actual training.

For further improvement, send training frames at MCS0 instead of 54 MBit/s
legacy. That way we can use the No-ACK flag for the transmission, which
speeds up PAPRD training in general, as the hardware won't have to
retransmit and wait for ACK timeout (was previously set to 4 * 6
transmission attempts).

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: fix the slot time setting for long distance links
Felix Fietkau [Mon, 13 Dec 2010 07:40:51 +0000 (08:40 +0100)]
ath9k_hw: fix the slot time setting for long distance links

Testing shows that adjusting the slot time based on the coverage class
produces very high latencies and very low throughput on long distance links.

Adjusting only the ACK timeout and leaving the slot time at the regular
values - while technically not optimal for CSMA - works a lot better on
long links (tested with 10 km distance)

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: initialize ah->slottime
Felix Fietkau [Mon, 13 Dec 2010 07:40:50 +0000 (08:40 +0100)]
ath9k_hw: initialize ah->slottime

(u32) -1 is not particularly useful as a slottime default, so even though
the ath9k_hw default should never get used, it's better to pick something
sane here.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_htc: Add Ubiquiti wifistation ext to supported devices
Sujith Manoharan [Mon, 13 Dec 2010 02:09:32 +0000 (07:39 +0530)]
ath9k_htc: Add Ubiquiti wifistation ext to supported devices

Signed-off-by: Sujith Manoharan <Sujith.Manoharan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: remove ah->txpower_indexoffset
Felix Fietkau [Sat, 11 Dec 2010 23:51:14 +0000 (00:51 +0100)]
ath9k_hw: remove ah->txpower_indexoffset

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: remove ah->beacon_interval
Felix Fietkau [Sat, 11 Dec 2010 23:51:13 +0000 (00:51 +0100)]
ath9k_hw: remove ah->beacon_interval

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: clean up SREV version checks
Felix Fietkau [Sat, 11 Dec 2010 23:51:12 +0000 (00:51 +0100)]
ath9k_hw: clean up SREV version checks

There's no need to have separate callbacks for pre-AR9003 vs AR9003
SREV version checks, so just merge those into one function.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: remove antenna configuration eeprom ops and variables
Felix Fietkau [Sat, 11 Dec 2010 23:51:11 +0000 (00:51 +0100)]
ath9k_hw: remove antenna configuration eeprom ops and variables

AR9280 based hardware with 3 antennas and slow antenna diversity has
not been seen in the wild and ath9k does not support that form of
antenna diversity, so remove the EEPROM ops for it.
These EEPROM ops are currently only used for setting the
AR_PHY_SWITCH_COM register, which is being done in the EEPROM specific
file already.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: merge the ar9287 version of ath9k_hw_get_gain_boundaries_pdadcs
Felix Fietkau [Sat, 11 Dec 2010 23:51:10 +0000 (00:51 +0100)]
ath9k_hw: merge the ar9287 version of ath9k_hw_get_gain_boundaries_pdadcs

Also add a comment about a potential array overrun that needs to
be reviewed.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: merge ath9k_hw_get_gain_boundaries_pdadcs between eeprom_def.c and eeprom_4k.c
Felix Fietkau [Sat, 11 Dec 2010 23:51:09 +0000 (00:51 +0100)]
ath9k_hw: merge ath9k_hw_get_gain_boundaries_pdadcs between eeprom_def.c and eeprom_4k.c

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: clean up duplicate and unnused eeprom related defines
Felix Fietkau [Sat, 11 Dec 2010 23:51:08 +0000 (00:51 +0100)]
ath9k_hw: clean up duplicate and unnused eeprom related defines

AR*_MAX_RATE_POWER => MAX_RATE_POWER
AR*_EEPROM_MODAL_SPURS => AR_EEPROM_MODAL_SPURS
AR*_OPFLAGS_* => AR5416_OPFLAGS_*
...

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k_hw: only use the PCIe disable register write sequence for AR5416
Felix Fietkau [Sat, 11 Dec 2010 23:51:07 +0000 (00:51 +0100)]
ath9k_hw: only use the PCIe disable register write sequence for AR5416

Newer chips do not need this, and maybe these register writes could have
negative side effects on newer hardware.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: fix sequence number assigment for non-AMPDU QoS data frames
Felix Fietkau [Fri, 10 Dec 2010 20:16:46 +0000 (21:16 +0100)]
ath9k: fix sequence number assigment for non-AMPDU QoS data frames

wireless-testing commit 04caf863750bc7e042d1e8d57e5ce9d6326ab435
('ath9k: more tx setup cleanups') merged tx path code for HT vs
non-HT frames, however it did not pass the tid pointer to
ath_tx_send_normal, causing an inconsistency between AMPDU vs
non-AMPDU sequence number handling.
Fix this by always passing in the tid pointer for all QoS data frames.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomac80211: support IBSS RSN with SW crypto
Johannes Berg [Fri, 10 Dec 2010 16:10:44 +0000 (17:10 +0100)]
mac80211: support IBSS RSN with SW crypto

When software crypto is used, mac80211 will
support IBSS RSN, it doesn't depend on the
driver in that case.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: Use power save wrappers for TSF get/set
Sujith Manoharan [Fri, 10 Dec 2010 05:57:06 +0000 (11:27 +0530)]
ath9k: Use power save wrappers for TSF get/set

The HW has to be awake when accessing registers.

Signed-off-by: Sujith Manoharan <Sujith.Manoharan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoieee80211: add Parameter Set Count bitmask
Bing Zhao [Fri, 10 Dec 2010 02:24:41 +0000 (18:24 -0800)]
ieee80211: add Parameter Set Count bitmask

WMM IE QoS Info field lower 4 bits: Parameter Set Count

Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomac80211 default tx_last_beacon false (congestion)
Tim Harvey [Thu, 9 Dec 2010 21:15:45 +0000 (13:15 -0800)]
mac80211 default tx_last_beacon false (congestion)

The 802.11 spec states that the STA that generated the last Beacon frame shall
be the STA that response to a probe request.  This is important for congestion
reduction when a probe request is received - only 1 node in an adhoc BSS
will transmit a response.  While mac80211 drivers should provide the
tx_last_beacon function to report if they transmitted the last beacon many
do not.  As an attempt to reduce probe response congestion default this
to 0 such that a node not implementing this capability does not contribute
to unnecessary congestion.

In a modern medium sized office environment I see upwards of 100 probe
requests per second received at a given node from various hardware/OS/drivers
doing zeroconf 'active probing' as opposed to passively listening for beacons.
With a modest 10-node adhoc network consisting of drivers that do not implement
this tx_last_beacon feature, I have seen this result in the simultaneous xmit
of probe responses accumulating to 500 probe responses per second because of
collisions which brings the adhoc network to its knees as well as causes
needless congestion.

Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agob43: rename TMS defines, drop useless condition from core reset
Rafał Miłecki [Thu, 9 Dec 2010 19:56:01 +0000 (20:56 +0100)]
b43: rename TMS defines, drop useless condition from core reset

As discussed we do not know band width at core reset time and it is not a good
idea to reset whole just to change band. So just set unconditionally 20 MHz
band width as default during core reset.

As for defines PHY clock changed to band width in specs and it makes much more
sens to call defines by band width which is self-explainable. Updated specs do
not mention 0 value, but comparing to old ones you can notice lineral relation
between PHY clock speed and band width. So it makes sense for 0x0 value to be
10 MHz band width.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agob43: N-PHY: use designed function and macro for writing tables
Rafał Miłecki [Thu, 9 Dec 2010 19:56:00 +0000 (20:56 +0100)]
b43: N-PHY: use designed function and macro for writing tables

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agob43: N-PHY: one more fix for order of tables initialization
Rafał Miłecki [Thu, 9 Dec 2010 19:55:59 +0000 (20:55 +0100)]
b43: N-PHY: one more fix for order of tables initialization

I missed that part in previous reordering.

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agob43: N-PHY: use correct bit for controlling MAC and PHY
Rafał Miłecki [Thu, 9 Dec 2010 19:55:58 +0000 (20:55 +0100)]
b43: N-PHY: use correct bit for controlling MAC and PHY

Signed-off-by: Rafał Miłecki <zajec5@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomac80211: support separate default keys
Johannes Berg [Thu, 9 Dec 2010 18:49:02 +0000 (19:49 +0100)]
mac80211: support separate default keys

Add support for split default keys (unicast
and multicast) in mac80211.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agocfg80211/nl80211: separate unicast/multicast default TX keys
Johannes Berg [Thu, 9 Dec 2010 18:58:59 +0000 (19:58 +0100)]
cfg80211/nl80211: separate unicast/multicast default TX keys

Allow userspace to specify that a given key
is default only for unicast and/or multicast
transmissions. Only WEP keys are for both,
WPA/RSN keys set here are GTKs for multicast
only. For more future flexibility, allow to
specify all combiations.

Wireless extensions can only set both so use
nl80211; WEP keys (connect keys) must be set
as default for both (but 802.1X WEP is still
possible).

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomac80211: clean up RX key checks
Johannes Berg [Thu, 9 Dec 2010 18:49:00 +0000 (19:49 +0100)]
mac80211: clean up RX key checks

Using the default key for "any key set" isn't
quite what we should do. It works, but with the
upcoming changes it makes life unnecessarily
complex, so do something better here and really
check for "any key".

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: clean up hardware code for beacon handling
Mohammed Shafi Shajakhan [Thu, 9 Dec 2010 16:18:54 +0000 (21:48 +0530)]
ath9k: clean up hardware code for beacon handling

The registers TBTT_TIMER ,DMA_BEACON_ALERT ,NEXT_SWBA are need to be
configured only for AP and IBSS mode.

SWBA register is used for generating software interrupts so that beacon
frames will be created by the software.DMA beacon alert register is
to indicate the hardware to DMA the contents of beacon buffer to PCU buffer
and TBTT to start transmitting the packet buffer to the base band.
Clearly these things are not needed for station/monitor mode so
remove configuring them.

Cc: doug dahlby <ddahlby@atheros.com>
Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agocfg80211: update information elements in cached BSS struct
Sven Neumann [Thu, 9 Dec 2010 14:05:24 +0000 (15:05 +0100)]
cfg80211: update information elements in cached BSS struct

When a cached BSS struct is updated because a new beacon was received,
the code replaces the cached information elements by the IEs from the
new beacon. However it did not update the pub.information_elements
and pub.len_information_elements fields leaving them either pointing
to the old beacon IEs or in an inconsistent state where the data is
replaced by the new beacon IEs but len_information_elements still has
its value from the first beacon.

Fix this by updating the information elements fields if they are
pointing to beacon IEs.

Signed-off-by: Sven Neumann <s.neumann@raumfeld.com>
Reviewed-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath: Missed to clear key4 of micentry
Rajkumar Manoharan [Wed, 8 Dec 2010 14:31:05 +0000 (20:01 +0530)]
ath: Missed to clear key4 of micentry

key4 of micentry is used, if ATH_CRYPT_CAP_MIC_COMBINED is set.
But is not cleared on key cache reset.

Cc: stable@kernel.org
Signed-off-by: Rajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: Add change_interface callback
Rajkumar Manoharan [Wed, 8 Dec 2010 14:08:55 +0000 (19:38 +0530)]
ath9k: Add change_interface callback

Add support to change interface type
without bringing down the interface.

Signed-off-by: Rajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agocfg80211: Add antenna availability information
Bruno Randolf [Wed, 8 Dec 2010 04:59:24 +0000 (13:59 +0900)]
cfg80211: Add antenna availability information

Add a field to wiphy for the hardware to report the availble antennas for
configuration. Only if this is set to something bigger than zero, will the
anntenna configuration ops be executed.

Allthough this could be a simple number of antennas, I defined it as a bitmap
of antennas which are available for configuration, since it's more consistent
with the rest of the antenna API and there could be cases where the
hardware allows only configuration of certain antennas. As it does not make
much of a difference in size or normal usage, I think it's better to be able to
support this, in case the need arises.

The antenna configuration is now also checked against the availabe antennas and
rejected if it does not match.

Signed-off-by: Bruno Randolf <br1@einfach.org>
--
v3: always apply available antenna mask (for "all" antennas case).

v2: reject antenna configurations which don't match the available antennas
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: fix assumptions for idle calls on suspend/resume
Luis R. Rodriguez [Tue, 7 Dec 2010 23:13:20 +0000 (15:13 -0800)]
ath9k: fix assumptions for idle calls on suspend/resume

mac80211 will notify drivers when to go idle and ath9k
assumed that it would get further notifications for idle
states after a device stop() config call but as per agreed
semantics the idle state of the radio is left up to driver
after mac80211 issues the stop() callback. The driver is
resposnbile for ensuring the device remains idle after
that even between suspend / resume calls.

This fixes suspend/resume when you issue suspend and resume
twice on ath9k when ath9k_stop() was already called. We need
to put the radio to full sleep in order for resume to work
correctly.

What might seem fishy is we are turning the radio off
after resume. The reason why we do this is because we know
we should not have anything enabled after a mac80211 tells
us to stop(), if we resume and never get a start() we won't
get another stop() by mac80211 so to be safe always bring
the 802.11 device with the radio disabled after resume,
this ensures that if we suspend we already have the radio
disabled and only a start() will ever trigger it on.

Cc: stable@kernel.org
Cc: Paul Stewart <pstew@google.com>
Cc: Amod Bodas <amod.bodas@atheros.com>
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: Fix power save count imbalance on ath_radio_enable()
Luis R. Rodriguez [Tue, 7 Dec 2010 23:13:19 +0000 (15:13 -0800)]
ath9k: Fix power save count imbalance on ath_radio_enable()

Upon a failure we never call ath9k_ps_restore() on ath_radio_enable(),
this will throw off the sc->ps_usecount. When the sc->ps_usecount
is > 0 we never put the chip to full sleep. This drains battery,
and will also make the chip fail upon resume with:

ath: Starting driver with initial channel: 5745 MHz
ath: timeout (100000 us) on reg 0x7000: 0xdeadbeef & 0x00000003 != 0x00000000

This would make the chip useless upon resume.

I cannot prove this can happen but in theory it is so best to
avoid this race completely and not have users complain about
a broken device after resume.

Cc: stable@kernel.org
Cc: Paul Stewart <pstew@google.com>
Cc: Amod Bodas <amod.bodas@atheros.com>
Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomac80211: avoid calling ieee80211_work_work unconditionally
Herton Ronaldo Krzesinski [Mon, 13 Dec 2010 13:43:51 +0000 (11:43 -0200)]
mac80211: avoid calling ieee80211_work_work unconditionally

On suspend, there might be usb wireless drivers which wrongly trigger
the warning in ieee80211_work_work. If an usb driver doesn't have a
suspend hook, the usb stack will disconnect the device. On disconnect,
a mac80211 driver calls ieee80211_unregister_hw, which calls dev_close,
which calls ieee80211_stop, and in the end calls ieee80211_work_purge->
ieee80211_work_work.

The problem is that this call to ieee80211_work_purge comes after
mac80211 is suspended, triggering the warning even when we don't have
work queued in work_list (the expected case when already suspended),
because it always calls ieee80211_work_work.

So, just call ieee80211_work_work in ieee80211_work_purge if we really
have to abort work. This addresses the warning reported at
https://bugzilla.kernel.org/show_bug.cgi?id=24402

Signed-off-by: Herton Ronaldo Krzesinski <herton@mandriva.com.br>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agop54usb: add 5 more USBIDs
Christian Lamparter [Sat, 11 Dec 2010 11:19:48 +0000 (12:19 +0100)]
p54usb: add 5 more USBIDs

This patch adds five more USBIDs to the table.

Source:
http://www.linuxant.com/pipermail/driverloader/2005q3/002307.html
http://wireless.kernel.org/en/users/Drivers/p54/devices (by M. Davis)

Cc: <stable@kernel.org>
Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomac80211: Fix NULL-pointer deference on ibss merge when not ready
Tim Harvey [Thu, 9 Dec 2010 18:43:13 +0000 (10:43 -0800)]
mac80211: Fix NULL-pointer deference on ibss merge when not ready

dev_open will eventually call ieee80211_ibss_join which sets up the
skb used for beacons/probe-responses however it is possible to
receive beacons that attempt to merge before this occurs causing
a null pointer dereference.  Check ssid_len as that is the last
thing set in ieee80211_ibss_join.

This occurs quite easily in the presence of adhoc nodes with hidden SSID's

revised previous patch to check further up based on irc feedback

Signed-off-by: Tim Harvey <harvey.tim@gmail.com>
Reviewed-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>