[ath9k-devel] AR9260 hang during connection to AP

[ath9k-devel] AR9260 hang during connection to AP

[Alex Williams]

Hello! My kernel version - 2.6.29-rc4-wl - from wireless testing git repository.
My card mini PCI, based on AR9260 chipset.

00:05.0 Network controller: Atheros Communications Inc. Device 0027 (rev 01)

If I setting "iwconfig wlan1 essid id" then system is hangs and stop
reacting on keyboard and other devices...

0) insmod ath9k.ko debug=0xFFFFFFFF

1) ifconfig wlan1 up
[  161.141947]
[  161.143586]
[  161.170654] ADDRCONF(NETDEV_UP): wlan1: link is not ready

2) iwconfig wlan1 essid zx
... < many empty kernel printk mesasges >
[  181.659150]
[  181.690218] queue 3
[  181.755139]
[  181.849091]
... < many empty kernel printk mesasges during several seconds>
[  187.071684]
[  187.165765]
[  187.259766]
[  194.891316] Clocksource tsc unstable (delta = 4687389681 ns)
<and there system is totally hangs...>

//-------------------------
I have questions:
1) How to fix this problem? Did ath9k ever work with AR9260?
2) Where I can get documentation for AR5*** and AR9*** chipsets? Does
it released for public access?

[ath9k-devel] AR9260 hang during connection to AP

[Sujith]

Alex Williams wrote:
> Hello! My kernel version - 2.6.29-rc4-wl - from wireless testing git repository.
> My card mini PCI, based on AR9260 chipset.
> 00:05.0 Network controller: Atheros Communications Inc. Device 0027 (rev 01)

I think you mean AR9160.
To confirm, can you post the output of sudo lspci -v ?

> If I setting "iwconfig wlan1 essid id" then system is hangs and stop
> reacting on keyboard and other devices...
> 
> 0) insmod ath9k.ko debug=0xFFFFFFFF

0xFFFFFFFF is probably overkill.
You can modprobe with debug=0x2000, which will produce a reasonable debug output.
Also can you post the dmesg immediately after loading ath9k with this debug mask ?

> 1) ifconfig wlan1 up
> [  161.141947]
> [  161.143586]
> [  161.170654] ADDRCONF(NETDEV_UP): wlan1: link is not ready
> 
> 2) iwconfig wlan1 essid zx
> ... < many empty kernel printk mesasges >
> [  181.659150]
> [  181.690218] queue 3
> [  181.755139]
> [  181.849091]
> ... < many empty kernel printk mesasges during several seconds>
> [  187.071684]
> [  187.165765]
> [  187.259766]
> [  194.891316] Clocksource tsc unstable (delta = 4687389681 ns)
> <and there system is totally hangs...>

Are you able to at least scan, before trying to associate with an AP ?
If so, then switch to a VT and try to associate and see if the kernel panics.
A digital photograph of the panic (if one occurs) would be great.

> 1) How to fix this problem? Did ath9k ever work with AR9260?

I haven't tried ath9k with any mini-PCI card so far, will dig around to see
if I can find one.

> 2) Where I can get documentation for AR5*** and AR9*** chipsets? Does
> it released for public access?

The code is the documentation. :)

Sujith
-- 
http://sujith-m.blogspot.com

[ath9k-devel] AR9260 hang during connection to AP

[Alex Williams]

Thank you for reply!
Please, tell me: did you saw working card on ar9160 ever??? May be
with other interfaces, such as PCI or USB? Is support for ar9160
finished? Or it's now unusable?

> I think you mean AR9160.
> To confirm, can you post the output of sudo lspci -v ?

00:05.0 Network controller: Atheros Communications Inc. Device 0027 (rev 01)
        Subsystem: Atheros Communications Inc. Device 2082
        Flags: bus master, 66MHz, medium devsel, latency 64, IRQ 10
        Memory at e1020000 (32-bit, non-prefetchable) [size=64K]
        Capabilities: [44] Power Management version 2

I can read mark on the chip - ar9160-bc1a - you right.

> 0xFFFFFFFF is probably overkill.
> You can modprobe with debug=0x2000, which will produce a reasonable debug output.
> Also can you post the dmesg immediately after loading ath9k with this debug mask ?

insmod ath9k.ko debug=0x2000
[  121.839737] ath9k 0000:00:05.0: PCI INT A -> Link[LN_B] -> GSI 10
(level, low) -> IRQ 10
modprobe: FATAL: Could not load
/lib/modules/2.6.29-rc4-wl/modules.dep: No such file or directory
[  123.479199] cfg80211: Calling CRDA for country: US
[  123.507201] udev: renamed network interface wlan0 to wlan1
[  123.512965] Registered led device: ath9k-phy0::radio
[  123.546520] Registered led device: ath9k-phy0::assoc
[  123.589623] Registered led device: ath9k-phy0::tx
[  123.614561] cfg80211: Regulatory domain changed to country: US
[  123.620474]  (start_freq - end_freq @ bandwidth),
(max_antenna_gain, max_eirp)
[  123.628382]  (2402000 KHz - 2472000 KHz @ 40000 KHz), (300 mBi, 2700 mBm)
[  123.635228]  (5170000 KHz - 5250000 KHz @ 40000 KHz), (300 mBi, 1700 mBm)
[  123.642074]  (5250000 KHz - 5330000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[  123.648921]  (5490000 KHz - 5710000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
[  123.655907]  (5735000 KHz - 5835000 KHz @ 40000 KHz), (300 mBi, 3000 mBm)
[  123.685508] Registered led device: ath9k-phy0::rx
[  123.727397] phy0: Atheros AR9160 MAC/BB Rev:1 AR5133 RF Rev:b0:
mem=0xcf520000, irq=10
insert ath9k.ko
[  124.070622] ADDRCONF(NETDEV_UP): wlan1: link is not ready

> Are you able to at least scan, before trying to associate with an AP ?
> If so, then switch to a VT and try to associate and see if the kernel panics.
> A digital photograph of the panic (if one occurs) would be great.

To perform scan, we need to "up" interface. Strange, but now, after
ifconfig wlan1 up - system hang deadly without any explanations! T_T
But sometimes it continue work after module load...

>> 1) How to fix this problem? Did ath9k ever work with AR9260?
> I haven't tried ath9k with any mini-PCI card so far, will dig around to see
> if I can find one.

OK

>> 2) Where I can get documentation for AR5*** and AR9*** chipsets? Does
>> it released for public access?
> The code is the documentation. :)

Really??? ^_^ It's impossible! Is there some datasheet on ar9160???
How about register addresses and their purpose? How can you write
driver for device if you do not know how it actually works? Writing
drivers it's like an embedded programming on microcontrolles without
OS - you should know every register and it's purpose. And even this
knowledge not guarantee errors proof... Sorry, I'm just wondering how
this possible... I heards that for ath9k Atheros released
specifications, or they writing in-house driver?.. Please, more info,
how it's really developing... =)

P.S.
How to enable additional debug info? Can I use KDBG + /dev/ttyS0 to
search for the problem? :)
P.P.S.
Please, give me a link to git tutorial. I want to modifi some wireless
files, but keep ability to update them by "git pull"... Sorry, I'm
newbie...

[ath9k-devel] AR9260 hang during connection to AP

[Sujith]

Alex Williams wrote:
> Thank you for reply!
> Please, tell me: did you saw working card on ar9160 ever??? May be
> with other interfaces, such as PCI or USB? Is support for ar9160
> finished? Or it's now unusable?

No idea. But I found a mini-PCI AR9160 card and it works fine here.

> insmod ath9k.ko debug=0x2000
> [  121.839737] ath9k 0000:00:05.0: PCI INT A -> Link[LN_B] -> GSI 10
> (level, low) -> IRQ 10
> modprobe: FATAL: Could not load
> /lib/modules/2.6.29-rc4-wl/modules.dep: No such file or directory
> [  123.479199] cfg80211: Calling CRDA for country: US
> [  123.507201] udev: renamed network interface wlan0 to wlan1
> [  123.512965] Registered led device: ath9k-phy0::radio
> [  123.546520] Registered led device: ath9k-phy0::assoc
> [  123.589623] Registered led device: ath9k-phy0::tx
> [  123.614561] cfg80211: Regulatory domain changed to country: US
> [  123.620474]  (start_freq - end_freq @ bandwidth),
> (max_antenna_gain, max_eirp)
> [  123.628382]  (2402000 KHz - 2472000 KHz @ 40000 KHz), (300 mBi, 2700 mBm)
> [  123.635228]  (5170000 KHz - 5250000 KHz @ 40000 KHz), (300 mBi, 1700 mBm)
> [  123.642074]  (5250000 KHz - 5330000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
> [  123.648921]  (5490000 KHz - 5710000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
> [  123.655907]  (5735000 KHz - 5835000 KHz @ 40000 KHz), (300 mBi, 3000 mBm)
> [  123.685508] Registered led device: ath9k-phy0::rx
> [  123.727397] phy0: Atheros AR9160 MAC/BB Rev:1 AR5133 RF Rev:b0:
> mem=0xcf520000, irq=10
> insert ath9k.ko
> [  124.070622] ADDRCONF(NETDEV_UP): wlan1: link is not ready

CONFIG_ATH9K_DEBUG should be enabled in your kernel .config to enable debugging.
See: http://wireless.kernel.org/en/users/Drivers/ath9k

> > Are you able to at least scan, before trying to associate with an AP ?
> > If so, then switch to a VT and try to associate and see if the kernel panics.
> > A digital photograph of the panic (if one occurs) would be great.
> 
> To perform scan, we need to "up" interface. Strange, but now, after
> ifconfig wlan1 up - system hang deadly without any explanations! T_T
> But sometimes it continue work after module load...

Did you switch to a VT to see if a panic happens on bringing up the interface ?

> Really??? ^_^ It's impossible! Is there some datasheet on ar9160???
> How about register addresses and their purpose? How can you write
> driver for device if you do not know how it actually works? Writing
> drivers it's like an embedded programming on microcontrolles without
> OS - you should know every register and it's purpose. And even this
> knowledge not guarantee errors proof... Sorry, I'm just wondering how
> this possible... I heards that for ath9k Atheros released
> specifications, or they writing in-house driver?.. Please, more info,
> how it's really developing... =)
> 

AR9160 is supported by ath9k, so no worries here.

> How to enable additional debug info? Can I use KDBG + /dev/ttyS0 to
> search for the problem? :)

Search for serial console / netconsole debugging guides on the net.

> Please, give me a link to git tutorial. I want to modifi some wireless
> files, but keep ability to update them by "git pull"... Sorry, I'm
> newbie...

http://wireless.kernel.org/en/developers/Documentation/git-guide

Sujith
-- 
http://sujith-m.blogspot.com

[ath9k-devel] AR9260 hang during connection to AP

[Alex Williams]

> No idea. But I found a mini-PCI AR9160 card and it works fine here.

Thanks! This is a very important information! I think I should try
other hardware and mini-PCI card... Can you tell model and
manufacturer of your card? Are you using wireless-testing git?

> CONFIG_ATH9K_DEBUG should be enabled in your kernel .config to enable debugging.
> See: http://wireless.kernel.org/en/users/Drivers/ath9k

Yes, I'm already have this option enabled this in my kernel config...

> Did you switch to a VT to see if a panic happens on bringing up the interface ?

Yes, I using serial port connection to "target" (x86 SBC) with miniCPI
card installed + minicom software in VT102 mode on the "host"...
Usually, when kernel panic happening, I see it on serial port console.
But in this situation I did not see kernel panic message - system just
hang up silently... =(

> AR9160 is supported by ath9k, so no worries here.

OK, I understand ;)

> Search for serial console / netconsole debugging guides on the net.
> http://wireless.kernel.org/en/developers/Documentation/git-guide

OK, I'll search. Thank you for the link :)

[ath9k-devel] AR9260 hang during connection to AP

[Alex Williams]

> No idea. But I found a mini-PCI AR9160 card and it works fine here.

Thanks again! My problem solved. I'm just switched to single board
computer from other manufacturer and now my AR9160 card works just
fine on all allowed frequencies, even > 5.8 GHz :)

[ath9k-devel] AR9260 hang during connection to AP

[Steve Brown]

(forgot to cc the list)

Sujith wrote:
> Alex Williams wrote:
>   
>> Thank you for reply!
>> Please, tell me: did you saw working card on ar9160 ever??? May be
>> with other interfaces, such as PCI or USB? Is support for ar9160
>> finished? Or it's now unusable?
>>     
>
> No idea. But I found a mini-PCI AR9160 card and it works fine here.
>
>   
>> insmod ath9k.ko debug=0x2000
>> [  121.839737] ath9k 0000:00:05.0: PCI INT A -> Link[LN_B] -> GSI 10
>> (level, low) -> IRQ 10
>> modprobe: FATAL: Could not load
>> /lib/modules/2.6.29-rc4-wl/modules.dep: No such file or directory
>> [  123.479199] cfg80211: Calling CRDA for country: US
>> [  123.507201] udev: renamed network interface wlan0 to wlan1
>> [  123.512965] Registered led device: ath9k-phy0::radio
>> [  123.546520] Registered led device: ath9k-phy0::assoc
>> [  123.589623] Registered led device: ath9k-phy0::tx
>> [  123.614561] cfg80211: Regulatory domain changed to country: US
>> [  123.620474]  (start_freq - end_freq @ bandwidth),
>> (max_antenna_gain, max_eirp)
>> [  123.628382]  (2402000 KHz - 2472000 KHz @ 40000 KHz), (300 mBi, 2700 mBm)
>> [  123.635228]  (5170000 KHz - 5250000 KHz @ 40000 KHz), (300 mBi, 1700 mBm)
>> [  123.642074]  (5250000 KHz - 5330000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
>> [  123.648921]  (5490000 KHz - 5710000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
>> [  123.655907]  (5735000 KHz - 5835000 KHz @ 40000 KHz), (300 mBi, 3000 mBm)
>> [  123.685508] Registered led device: ath9k-phy0::rx
>> [  123.727397] phy0: Atheros AR9160 MAC/BB Rev:1 AR5133 RF Rev:b0:
>> mem=0xcf520000, irq=10
>> insert ath9k.ko
>> [  124.070622] ADDRCONF(NETDEV_UP): wlan1: link is not ready
>>     
>
> CONFIG_ATH9K_DEBUG should be enabled in your kernel .config to enable debugging.
> See: http://wireless.kernel.org/en/users/Drivers/ath9k
>
>   
>>> Are you able to at least scan, before trying to associate with an AP ?
>>> If so, then switch to a VT and try to associate and see if the kernel panics.
>>> A digital photograph of the panic (if one occurs) would be great.
>>>       
>> To perform scan, we need to "up" interface. Strange, but now, after
>> ifconfig wlan1 up - system hang deadly without any explanations! T_T
>> But sometimes it continue work after module load...
>>     
>
> Did you switch to a VT to see if a panic happens on bringing up the interface ?
>
>   
>> Really??? ^_^ It's impossible! Is there some datasheet on ar9160???
>> How about register addresses and their purpose? How can you write
>> driver for device if you do not know how it actually works? Writing
>> drivers it's like an embedded programming on microcontrolles without
>> OS - you should know every register and it's purpose. And even this
>> knowledge not guarantee errors proof... Sorry, I'm just wondering how
>> this possible... I heards that for ath9k Atheros released
>> specifications, or they writing in-house driver?.. Please, more info,
>> how it's really developing... =)
>>
>>     
>
> AR9160 is supported by ath9k, so no worries here.
>
>   
>> How to enable additional debug info? Can I use KDBG + /dev/ttyS0 to
>> search for the problem? :)
>>     
>
> Search for serial console / netconsole debugging guides on the net.
>
>   
>> Please, give me a link to git tutorial. I want to modifi some wireless
>> files, but keep ability to update them by "git pull"... Sorry, I'm
>> newbie...
>>     
>
> http://wireless.kernel.org/en/developers/Documentation/git-guide
>
> Sujith
>   
I'm having the same problem. The driver loads and the device scans. The 
oops happens while networkmanager is trying to associate with an AP.

01:0b.0 Network controller: Atheros Communications Inc. AR9160 
802.11abgn Wireless PCI Adapter (rev 01)
01:0b.0 0280: 168c:0027 (rev 01)
   Subsystem: 168c:2082
   Flags: bus master, 66MHz, medium devsel, latency 64, IRQ 5
   Memory at fe5e0000 (32-bit, non-prefetchable) [size=64K]
   Capabilities: [44] Power Management version 2
   Kernel modules: ath9k

The card is a Sparklan WMIA-199N.

The machine is a P4 2.80GHz running ubuntu and v2.6.29-rc5-14641-g562bdf5

The driver on loading:
Feb 19 09:16:23 ubuntu kernel: [ 4676.662134] ath9k 0000:01:0b.0: PCI 
INT A -> Link[LNKG] -> GSI 5 (level, low) -> IRQ 5
Feb 19 09:16:24 ubuntu kernel: [ 4679.635453] phy2: Selected rate 
control algorithm 'ath9k_rate_control'
Feb 19 09:16:24 ubuntu kernel: [ 4679.676985] cfg80211: Calling CRDA for 
country: AM
Feb 19 09:16:24 ubuntu kernel: [ 4679.677041] Registered led device: 
ath9k-phy2::radio
Feb 19 09:16:24 ubuntu kernel: [ 4679.677060] Registered led device: 
ath9k-phy2::assoc
Feb 19 09:16:24 ubuntu kernel: [ 4679.677078] Registered led device: 
ath9k-phy2::tx
Feb 19 09:16:24 ubuntu kernel: [ 4679.677103] Registered led device: 
ath9k-phy2::rx
Feb 19 09:16:24 ubuntu kernel: [ 4679.677113] phy2: Atheros AR9160 
MAC/BB Rev:1 AR5133 RF Rev:b0: mem=0xf80c0000, irq=5
Feb 19 09:16:24 ubuntu kernel: [ 4679.701984] udev: renamed network 
interface wlan0 to wlan2


The netconsole output is attached.

Hope I haven't missed anything.

Steve



------------------------------------------------------------------------
Feb 19 12:51:12 192.168.1.7 [ 307.761386] ------------[ cut here ]------------
Feb 19 12:51:12 192.168.1.7 [ 307.761399] kernel BUG at drivers/net/wireless/ath9k/rc.c:745!
Feb 19 12:51:12 192.168.1.7 [ 307.761404] invalid opcode: 0000 [#1] 
Feb 19 12:51:12 192.168.1.7 
Feb 19 12:51:12 192.168.1.7 [ 307.761411] last sysfs file: /sys/devices/pci0000:00/0000:00:1e.0/0000:01:0b.0/net/wlan2/flags
Feb 19 12:51:12 192.168.1.7 [ 307.761417] Modules linked in:
Feb 19 12:51:12  arc4
Feb 19 12:51:12  ecb
Feb 19 12:51:12  ath9k
Feb 19 12:51:12  mac80211
Feb 19 12:51:12  rfkill
Feb 19 12:51:12  cfg80211
Feb 19 12:51:12  led_class
Feb 19 12:51:12  netconsole
Feb 19 12:51:12  configfs
Feb 19 12:51:12  i915
Feb 19 12:51:12  drm
Feb 19 12:51:12  i2c_algo_bit
Feb 19 12:51:12  i2c_core
Feb 19 12:51:12  af_packet
Feb 19 12:51:12  ppdev
Feb 19 12:51:12  speedstep_lib
Feb 19 12:51:12  cpufreq_conservative
Feb 19 12:51:12  cpufreq_powersave
Feb 19 12:51:12  cpufreq_userspace
Feb 19 12:51:12  cpufreq_ondemand
Feb 19 12:51:12  cpufreq_stats
Feb 19 12:51:12  freq_table
Feb 19 12:51:13  rfcomm
Feb 19 12:51:13  sco
Feb 19 12:51:13  bridge
Feb 19 12:51:13  stp
Feb 19 12:51:13  bnep
Feb 19 12:51:13  l2cap
Feb 19 12:51:13  bluetooth
Feb 19 12:51:13  binfmt_misc
Feb 19 12:51:13  ipv6
Feb 19 12:51:13  battery
Feb 19 12:51:13  sbs
Feb 19 12:51:13  sbshc
Feb 19 12:51:13  video
Feb 19 12:51:13  output
Feb 19 12:51:13  container
Feb 19 12:51:13  pci_slot
Feb 19 12:51:13  snd_intel8x0
Feb 19 12:51:13  snd_ac97_codec
Feb 19 12:51:13  ac97_bus
Feb 19 12:51:13  snd_pcm_oss
Feb 19 12:51:13  snd_mixer_oss
Feb 19 12:51:13  snd_pcm
Feb 19 12:51:13  snd_seq_dummy
Feb 19 12:51:13  snd_seq_oss
Feb 19 12:51:13  snd_seq_midi
Feb 19 12:51:13  snd_rawmidi
Feb 19 12:51:13  snd_seq_midi_event
Feb 19 12:51:13  snd_seq
Feb 19 12:51:13  snd_timer
Feb 19 12:51:13  snd_seq_device
Feb 19 12:51:13  snd
Feb 19 12:51:13  psmouse
Feb 19 12:51:14  serio_raw
Feb 19 12:51:14  pcspkr
Feb 19 12:51:14  soundcore
Feb 19 12:51:14  snd_page_alloc
Feb 19 12:51:14  evdev
Feb 19 12:51:14  iTCO_wdt
Feb 19 12:51:14  iTCO_vendor_support
Feb 19 12:51:14  button
Feb 19 12:51:14  shpchp
Feb 19 12:51:14  pci_hotplug
Feb 19 12:51:14  intel_agp
Feb 19 12:51:14  agpgart
Feb 19 12:51:14  parport_pc
Feb 19 12:51:14  lp
Feb 19 12:51:14  parport
Feb 19 12:51:14  ac
Feb 19 12:51:14  iptable_filter
Feb 19 12:51:14  ip_tables
Feb 19 12:51:14  x_tables
Feb 19 12:51:14  ext3
Feb 19 12:51:14  jbd
Feb 19 12:51:14  mbcache
Feb 19 12:51:14  usbhid
Feb 19 12:51:14  hid
Feb 19 12:51:14  sr_mod
Feb 19 12:51:14  cdrom
Feb 19 12:51:14  sd_mod
Feb 19 12:51:14  crc_t10dif
Feb 19 12:51:14  sg
Feb 19 12:51:14  ata_piix
Feb 19 12:51:14  pata_acpi
Feb 19 12:51:14  ata_generic
Feb 19 12:51:14  libata
Feb 19 12:51:15  scsi_mod
Feb 19 12:51:15  ehci_hcd
Feb 19 12:51:15  uhci_hcd
Feb 19 12:51:15  e100
Feb 19 12:51:15  mii
Feb 19 12:51:15  usbcore
Feb 19 12:51:15  thermal
Feb 19 12:51:15  processor
Feb 19 12:51:15  fan
Feb 19 12:51:15  fuse
Feb 19 12:51:15 192.168.1.7 
Feb 19 12:51:15 192.168.1.7 [ 307.761748] 
Feb 19 12:51:15 192.168.1.7 [ 307.761753] Pid: 5611, comm: phy0 Not tainted (2.6.29-rc5-wl #6) To Be Filled By O.E.M.
Feb 19 12:51:15 192.168.1.7 [ 307.761758] EIP: 0060:[<f87a7bcf>] EFLAGS: 00010246 CPU: 0
Feb 19 12:51:15 192.168.1.7 [ 307.761771] EIP is at ath_get_rate+0x28f/0x620 [ath9k]
Feb 19 12:51:15 192.168.1.7 [ 307.761776] EAX: 00000000 EBX: 0000001c ECX: 00000000 EDX: f87c4980
Feb 19 12:51:15 192.168.1.7 [ 307.761780] ESI: 0000001c EDI: f87c4980 EBP: dfeb7d2c ESP: dfeb7cdc
Feb 19 12:51:15 192.168.1.7 [ 307.761786]  DS: 007b ES: 007b FS: 0000 GS: 0000 SS: 0068
Feb 19 12:51:15 192.168.1.7 [ 307.761792] Process phy0 (pid: 5611, ti=dfeb6000 task=dfd63cf0 task.ti=dfeb6000)
Feb 19 12:51:15 192.168.1.7 [ 307.761795] Stack:
Feb 19 12:51:15 192.168.1.7 [ 307.761799]  df307d08
Feb 19 12:51:15  c013fcc3
Feb 19 12:51:15  ee059860
Feb 19 12:51:15  dfeb7d08
Feb 19 12:51:15  00000046
Feb 19 12:51:15  2aeb7d20
Feb 19 12:51:15  dfd32800
Feb 19 12:51:15  d870eb00
Feb 19 12:51:15 192.168.1.7 
Feb 19 12:51:15 192.168.1.7 [ 307.761824] 
Feb 19 12:51:15  00000148
Feb 19 12:51:15  ee14fe00
Feb 19 12:51:16  e7dda410
Feb 19 12:51:16  ee14fe20
Feb 19 12:51:16  00002863
Feb 19 12:51:16  00001e50
Feb 19 12:51:16  f87c4980
Feb 19 12:51:16  0000001c
Feb 19 12:51:16 192.168.1.7 
Feb 19 12:51:16 192.168.1.7 [ 307.761931] 
Feb 19 12:51:16  1d1c000a
Feb 19 12:51:16  f87c3500
Feb 19 12:51:16  ee14fe20
Feb 19 12:51:16  ee1b0b80
Feb 19 12:51:16  dfeb7d50
Feb 19 12:51:16  f845d82d
Feb 19 12:51:16  dfeb7ddc
Feb 19 12:51:16  f87c3500
Feb 19 12:51:16 192.168.1.7 
Feb 19 12:51:16 192.168.1.7 [ 307.761964] Call Trace:
Feb 19 12:51:16 192.168.1.7 [ 307.761968] 
Feb 19 12:51:16  [<c013fcc3>] ? getnstimeofday+0x53/0x110
Feb 19 12:51:16 192.168.1.7 [ 307.761986] 
Feb 19 12:51:16  [<f845d82d>] ? rate_control_get_rate+0xbd/0xd0 [mac80211]
Feb 19 12:51:16 192.168.1.7 [ 307.762011] 
Feb 19 12:51:16  [<f8464d38>] ? invoke_tx_handlers+0x648/0x10b0 [mac80211]
Feb 19 12:51:16 192.168.1.7 [ 307.762029] 
Feb 19 12:51:16  [<c0199043>] ? __slab_alloc+0xb3/0x4f0
Feb 19 12:51:16 192.168.1.7 [ 307.762040] 
Feb 19 12:51:16  [<c02ce4f8>] ? skb_release_data+0x68/0xa0
Feb 19 12:51:16 192.168.1.7 [ 307.762048] 
Feb 19 12:51:17  [<f8464184>] ? __ieee80211_tx_prepare+0x184/0x340 [mac80211]
Feb 19 12:51:17 192.168.1.7 [ 307.762068] 
Feb 19 12:51:17  [<f8466cac>] ? ieee80211_master_start_xmit+0x24c/0x550 [mac80211]
Feb 19 12:51:17 192.168.1.7 [ 307.762091] 
Feb 19 12:51:17  [<c02d797b>] ? dev_hard_start_xmit+0x25b/0x2f0
Feb 19 12:51:17 192.168.1.7 [ 307.762102] 
Feb 19 12:51:17  [<c02e8bd4>] ? __qdisc_run+0x194/0x1f0
Feb 19 12:51:17 192.168.1.7 [ 307.762174] 
Feb 19 12:51:17  [<c01023f4>] ? __switch_to+0xa4/0x190
Feb 19 12:51:17 192.168.1.7 [ 307.762180] 
Feb 19 12:51:17  [<c02d89ff>] ? dev_queue_xmit+0x32f/0x580
Feb 19 12:51:17 192.168.1.7 [ 307.762187] 
Feb 19 12:51:17  [<c02ced2d>] ? __alloc_skb+0x4d/0x110
Feb 19 12:51:17 192.168.1.7 [ 307.762195] 
Feb 19 12:51:17  [<f84689c2>] ? ieee80211_tx_skb+0x52/0x60 [mac80211]
Feb 19 12:51:17 192.168.1.7 [ 307.762212] 
Feb 19 12:51:17  [<f8458ded>] ? ieee80211_send_nullfunc+0xbd/0x100 [mac80211]
Feb 19 12:51:17 192.168.1.7 [ 307.762227] 
Feb 19 12:51:17  [<c01209cb>] ? __cond_resched+0x1b/0x40
Feb 19 12:51:17 192.168.1.7 [ 307.762235] 
Feb 19 12:51:17  [<f8455b2a>] ? ieee80211_scan_completed+0x2da/0x360 [mac80211]
Feb 19 12:51:17 192.168.1.7 [ 307.762251] 
Feb 19 12:51:17  [<f8455bb0>] ? ieee80211_scan_work+0x0/0x190 [mac80211]
Feb 19 12:51:17 192.168.1.7 [ 307.762266] 
Feb 19 12:51:17  [<f8455d09>] ? ieee80211_scan_work+0x159/0x190 [mac80211]
Feb 19 12:51:17 192.168.1.7 [ 307.762287] 
Feb 19 12:51:17  [<c0356f8b>] ? schedule+0x31b/0x4a0
Feb 19 12:51:17 192.168.1.7 [ 307.762297] 
Feb 19 12:51:17  [<f8455bb0>] ? ieee80211_scan_work+0x0/0x190 [mac80211]
Feb 19 12:51:17 192.168.1.7 [ 307.762318] 
Feb 19 12:51:18  [<c013460b>] ? run_workqueue+0x5b/0x110
Feb 19 12:51:18 192.168.1.7 [ 307.762387] 
Feb 19 12:51:18  [<c0134adf>] ? worker_thread+0x7f/0xe0
Feb 19 12:51:18 192.168.1.7 [ 307.762394] 
Feb 19 12:51:18  [<c0138150>] ? autoremove_wake_function+0x0/0x50
Feb 19 12:51:18 192.168.1.7 [ 307.762402] 
Feb 19 12:51:18  [<c0134a60>] ? worker_thread+0x0/0xe0
Feb 19 12:51:18 192.168.1.7 [ 307.762407] 
Feb 19 12:51:18  [<c0137c6a>] ? kthread+0x3a/0x70
Feb 19 12:51:18 192.168.1.7 [ 307.762416] 
Feb 19 12:51:18  [<c0137c30>] ? kthread+0x0/0x70
Feb 19 12:51:18 192.168.1.7 [ 307.762422] 
Feb 19 12:51:18  [<c0103d27>] ? kernel_thread_helper+0x7/0x10
Feb 19 12:51:18 192.168.1.7 [ 307.762430] Code: 
Feb 19 12:51:18 192.168.1.7 8b
Feb 19 12:51:18 192.168.1.7 44
Feb 19 12:51:18 192.168.1.7 c2
Feb 19 12:51:18 192.168.1.7 04
Feb 19 12:51:18 192.168.1.7 85
Feb 19 12:51:18 192.168.1.7 c0
Feb 19 12:51:18 192.168.1.7 74
Feb 19 12:51:18 192.168.1.7 09
Feb 19 12:51:18 192.168.1.7 8b
Feb 19 12:51:18 192.168.1.7 5d
Feb 19 12:51:18 192.168.1.7 c8
Feb 19 12:51:18 192.168.1.7 80
Feb 19 12:51:18 192.168.1.7 7b
Feb 19 12:51:18 192.168.1.7 55
Feb 19 12:51:18 192.168.1.7 00

Message from syslogd at 192.168.1.7 at Feb 19 12:51:24 ...
 f>
Feb 19 12:51:24 192.168.1.7 74
Feb 19 12:51:24 192.168.1.7 38
Feb 19 12:51:24 192.168.1.7 8b
Feb 19 12:51:24 192.168.1.7 75
Feb 19 12:51:24 192.168.1.7 ec
Feb 19 12:51:24 192.168.1.7 8b
Feb 19 12:51:24 192.168.1.7 7d
Feb 19 12:51:24 192.168.1.7 e8
Feb 19 12:51:24 192.168.1.7 8d
Feb 19 12:51:24 192.168.1.7 04
Feb 19 12:51:24 192.168.1.7 b6
Feb 19 12:51:24 192.168.1.7 8b
Feb 19 12:51:24 192.168.1.7 44
Feb 19 12:51:24 192.168.1.7 c7
Feb 19 12:51:24 192.168.1.7 08
Feb 19 12:51:24 192.168.1.7 85
Feb 19 12:51:24 192.168.1.7 c0
Feb 19 12:51:24 192.168.1.7 74
Feb 19 12:51:24 192.168.1.7 09
Feb 19 12:51:24 192.168.1.7 8b
Feb 19 12:51:24 192.168.1.7 45
Feb 19 12:51:24 192.168.1.7 c8
Feb 19 12:51:24 192.168.1.7 80
Feb 19 12:51:24 192.168.1.7 78
Feb 19 12:51:24 192.168.1.7 55
Feb 19 12:51:24 192.168.1.7 00
Feb 19 12:51:24 192.168.1.7 75
Feb 19 12:51:24 192.168.1.7 1e
Feb 19 12:51:24 192.168.1.7 f>
Feb 19 12:51:24 192.168.1.7 0b
Feb 19 12:51:24 192.168.1.7 eb
Feb 19 12:51:24 192.168.1.7 fe
Feb 19 12:51:25 192.168.1.7 90
Feb 19 12:51:25 192.168.1.7 8d
Feb 19 12:51:25 192.168.1.7 74
Feb 19 12:51:25 192.168.1.7 26
Feb 19 12:51:25 192.168.1.7 00
Feb 19 12:51:25 192.168.1.7 38
Feb 19 12:51:25 192.168.1.7 d3
Feb 19 12:51:25 192.168.1.7 0f
Feb 19 12:51:25 192.168.1.7 8e
Feb 19 12:51:25 192.168.1.7 5e
Feb 19 12:51:25 192.168.1.7 fe
Feb 19 12:51:25 192.168.1.7 ff
Feb 19 12:51:25 192.168.1.7 ff
Feb 19 12:51:25 192.168.1.7 89
Feb 19 12:51:25 192.168.1.7 d3
Feb 19 12:51:25 192.168.1.7 8d
Feb 19 12:51:25 192.168.1.7 b6
Feb 19 12:51:25 192.168.1.7 
Feb 19 12:51:25 192.168.1.7 [ 307.763062] EIP: [<f87a7bcf>] 
Feb 19 12:51:25 192.168.1.7 ath_get_rate+0x28f/0x620 [ath9k]
Feb 19 12:51:25  SS: ESP 0068:dfeb7cdc
Feb 19 12:51:25 192.168.1.7 [ 307.763091] ---[ end trace fdaf4ee84daa2dc6 ]---
=======================================

messages

[ath9k-devel] AR9260 hang during connection to AP

[Steve Brown]

Sujith wrote:
> Alex Williams wrote:
>   
>> Thank you for reply!
>> Please, tell me: did you saw working card on ar9160 ever??? May be
>> with other interfaces, such as PCI or USB? Is support for ar9160
>> finished? Or it's now unusable?
>>     
>
> No idea. But I found a mini-PCI AR9160 card and it works fine here.
>
>   
>> insmod ath9k.ko debug=0x2000
>> [  121.839737] ath9k 0000:00:05.0: PCI INT A -> Link[LN_B] -> GSI 10
>> (level, low) -> IRQ 10
>> modprobe: FATAL: Could not load
>> /lib/modules/2.6.29-rc4-wl/modules.dep: No such file or directory
>> [  123.479199] cfg80211: Calling CRDA for country: US
>> [  123.507201] udev: renamed network interface wlan0 to wlan1
>> [  123.512965] Registered led device: ath9k-phy0::radio
>> [  123.546520] Registered led device: ath9k-phy0::assoc
>> [  123.589623] Registered led device: ath9k-phy0::tx
>> [  123.614561] cfg80211: Regulatory domain changed to country: US
>> [  123.620474]  (start_freq - end_freq @ bandwidth),
>> (max_antenna_gain, max_eirp)
>> [  123.628382]  (2402000 KHz - 2472000 KHz @ 40000 KHz), (300 mBi, 2700 mBm)
>> [  123.635228]  (5170000 KHz - 5250000 KHz @ 40000 KHz), (300 mBi, 1700 mBm)
>> [  123.642074]  (5250000 KHz - 5330000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
>> [  123.648921]  (5490000 KHz - 5710000 KHz @ 40000 KHz), (300 mBi, 2000 mBm)
>> [  123.655907]  (5735000 KHz - 5835000 KHz @ 40000 KHz), (300 mBi, 3000 mBm)
>> [  123.685508] Registered led device: ath9k-phy0::rx
>> [  123.727397] phy0: Atheros AR9160 MAC/BB Rev:1 AR5133 RF Rev:b0:
>> mem=0xcf520000, irq=10
>> insert ath9k.ko
>> [  124.070622] ADDRCONF(NETDEV_UP): wlan1: link is not ready
>>     
>
> CONFIG_ATH9K_DEBUG should be enabled in your kernel .config to enable debugging.
> See: http://wireless.kernel.org/en/users/Drivers/ath9k
>
>   
>>> Are you able to at least scan, before trying to associate with an AP ?
>>> If so, then switch to a VT and try to associate and see if the kernel panics.
>>> A digital photograph of the panic (if one occurs) would be great.
>>>       
>> To perform scan, we need to "up" interface. Strange, but now, after
>> ifconfig wlan1 up - system hang deadly without any explanations! T_T
>> But sometimes it continue work after module load...
>>     
>
> Did you switch to a VT to see if a panic happens on bringing up the interface ?
>
>   
>> Really??? ^_^ It's impossible! Is there some datasheet on ar9160???
>> How about register addresses and their purpose? How can you write
>> driver for device if you do not know how it actually works? Writing
>> drivers it's like an embedded programming on microcontrolles without
>> OS - you should know every register and it's purpose. And even this
>> knowledge not guarantee errors proof... Sorry, I'm just wondering how
>> this possible... I heards that for ath9k Atheros released
>> specifications, or they writing in-house driver?.. Please, more info,
>> how it's really developing... =)
>>
>>     
>
> AR9160 is supported by ath9k, so no worries here.
>
>   
>> How to enable additional debug info? Can I use KDBG + /dev/ttyS0 to
>> search for the problem? :)
>>     
>
> Search for serial console / netconsole debugging guides on the net.
>
>   
>> Please, give me a link to git tutorial. I want to modifi some wireless
>> files, but keep ability to update them by "git pull"... Sorry, I'm
>> newbie...
>>     
>
> http://wireless.kernel.org/en/developers/Documentation/git-guide
>
> Sujith
>   
For me, the ASSERT at rc.c:745 goes off.

I instrumented it. The results are attached.

Steve



-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: ath9k-rc-assert.txt
Url: http://lists.ath9k.org/pipermail/ath9k-devel/attachments/20090220/7549cb03/attachment-0001.txt 

[ath9k-devel] AR9260 hang during connection to AP

[Sujith]

Steve Brown wrote:
> For me, the ASSERT at rc.c:745 goes off.
> 
> I instrumented it. The results are attached.

Thanks for the trace. We'll work on this.

Sujith

[ath9k-devel] AR9260 hang during connection to AP

[Sujith]

Steve Brown wrote:
> Feb 20 11:01:29 ubuntu kernel: [15783.068821] ath9k: Set channel: 2437 MHz
> Feb 20 11:01:29 ubuntu kernel: [15783.068825] ath9k: tx chmask: 1, rx chmask: 1
> Feb 20 11:01:29 ubuntu kernel: [15783.068909] ath9k: (2412 MHz) -> (2437 MHz), chanwidth: 0
> Feb 20 11:01:29 ubuntu kernel: [15783.069918] ath9k: RX filter 0x0 bssid 00:0e:8e:1d:f5:5b aid 0x0
> Feb 20 11:01:29 ubuntu kernel: [15783.070935] ath9k: Set channel: 2437 MHz
> Feb 20 11:01:29 ubuntu kernel: [15783.070941] ath9k: tx chmask: 1, rx chmask: 1
> Feb 20 11:01:29 ubuntu kernel: [15783.071026] ath9k: (2437 MHz) -> (2437 MHz), chanwidth: 0

Is the AP on channel 6 ? Configured with dynamic 20/40 ?

> Feb 20 11:01:29 ubuntu kernel: [15783.074932] ath9k: RX filter 0x0 bssid 00:0e:8e:1d:f5:5b aid 0x0
> Feb 20 11:01:29 ubuntu kernel: [15783.075828] ath9k: Configure tx [queue/halq] [0/3],  aifs: 2, cw_min: 15, cw_max: 1023, txop: 0
> Feb 20 11:01:29 ubuntu kernel: [15783.075842] ath9k: Configure tx [queue/halq] [1/2],  aifs: 2, cw_min: 15, cw_max: 1023, txop: 0
> Feb 20 11:01:29 ubuntu kernel: [15783.075853] ath9k: Configure tx [queue/halq] [2/1],  aifs: 2, cw_min: 15, cw_max: 1023, txop: 0
> Feb 20 11:01:29 ubuntu kernel: [15783.075864] ath9k: Configure tx [queue/halq] [3/0],  aifs: 2, cw_min: 15, cw_max: 1023, txop: 0
> Feb 20 11:01:29 ubuntu kernel: [15783.075880] wlan2: authenticate with AP 00:0e:8e:1d:f5:5b
> Feb 20 11:01:29 ubuntu kernel: [15783.078537] wlan2: authenticated
> Feb 20 11:01:29 ubuntu kernel: [15783.078540] wlan2: associate with AP 00:0e:8e:1d:f5:5b
> Feb 20 11:01:29 ubuntu kernel: [15783.081874] wlan2: RX AssocResp from 00:0e:8e:1d:f5:5b (capab=0x411 status=0 aid=2)
> Feb 20 11:01:29 ubuntu kernel: [15783.081878] wlan2: associated
> Feb 20 11:01:29 ubuntu kernel: [15783.081889] ath9k: Choosing rate table for mode: 10
                                                       ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 

This is the bug, the AP advertises 20/40, but sets the operating channel to 20 Mhz,
but ath9k chooses the wrong rate table since it doesn't honor the AP's current channel width.

Sujith

[ath9k-devel] AR9260 hang during connection to AP

[Steve Brown]

Sujith wrote:
> Steve Brown wrote:
>   
>> Feb 20 11:01:29 ubuntu kernel: [15783.068821] ath9k: Set channel: 2437 MHz
>> Feb 20 11:01:29 ubuntu kernel: [15783.068825] ath9k: tx chmask: 1, rx chmask: 1
>> Feb 20 11:01:29 ubuntu kernel: [15783.068909] ath9k: (2412 MHz) -> (2437 MHz), chanwidth: 0
>> Feb 20 11:01:29 ubuntu kernel: [15783.069918] ath9k: RX filter 0x0 bssid 00:0e:8e:1d:f5:5b aid 0x0
>> Feb 20 11:01:29 ubuntu kernel: [15783.070935] ath9k: Set channel: 2437 MHz
>> Feb 20 11:01:29 ubuntu kernel: [15783.070941] ath9k: tx chmask: 1, rx chmask: 1
>> Feb 20 11:01:29 ubuntu kernel: [15783.071026] ath9k: (2437 MHz) -> (2437 MHz), chanwidth: 0
>>     
>
> Is the AP on channel 6 ? Configured with dynamic 20/40 ?
>   
hostapd.conf
channel=6
ht_capab=[HT40-][SHORT-GI-40][DSSS_CCK-40]

iw phy phy1 info
Wiphy phy1
        Band 1:
                HT capabilities: 0x104e
                         * 20/40 MHz operation
                         * SM PS disabled
                         * 40 MHz short GI
                         * max A-MSDU len 3839
                         * DSSS/CCK 40 MHz
                HT A-MPDU factor: 0x0003 (65535 bytes)
                HT A-MPDU density: 0x0006 (8 usec)
                HT MCS set: ff ff 00 00 00 00 00 00 00 00 00 00 01 00 00 00

>   
>> Feb 20 11:01:29 ubuntu kernel: [15783.074932] ath9k: RX filter 0x0 bssid 00:0e:8e:1d:f5:5b aid 0x0
>> Feb 20 11:01:29 ubuntu kernel: [15783.075828] ath9k: Configure tx [queue/halq] [0/3],  aifs: 2, cw_min: 15, cw_max: 1023, txop: 0
>> Feb 20 11:01:29 ubuntu kernel: [15783.075842] ath9k: Configure tx [queue/halq] [1/2],  aifs: 2, cw_min: 15, cw_max: 1023, txop: 0
>> Feb 20 11:01:29 ubuntu kernel: [15783.075853] ath9k: Configure tx [queue/halq] [2/1],  aifs: 2, cw_min: 15, cw_max: 1023, txop: 0
>> Feb 20 11:01:29 ubuntu kernel: [15783.075864] ath9k: Configure tx [queue/halq] [3/0],  aifs: 2, cw_min: 15, cw_max: 1023, txop: 0
>> Feb 20 11:01:29 ubuntu kernel: [15783.075880] wlan2: authenticate with AP 00:0e:8e:1d:f5:5b
>> Feb 20 11:01:29 ubuntu kernel: [15783.078537] wlan2: authenticated
>> Feb 20 11:01:29 ubuntu kernel: [15783.078540] wlan2: associate with AP 00:0e:8e:1d:f5:5b
>> Feb 20 11:01:29 ubuntu kernel: [15783.081874] wlan2: RX AssocResp from 00:0e:8e:1d:f5:5b (capab=0x411 status=0 aid=2)
>> Feb 20 11:01:29 ubuntu kernel: [15783.081878] wlan2: associated
>> Feb 20 11:01:29 ubuntu kernel: [15783.081889] ath9k: Choosing rate table for mode: 10
>>     
>                                                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 
>
> This is the bug, the AP advertises 20/40, but sets the operating channel to 20 Mhz,
> but ath9k chooses the wrong rate table since it doesn't honor the AP's current channel width.
>
> Sujith
>   
Hope this answers your question.

Steve

[ath9k-devel] AR9260 hang during connection to AP

[Sujith]

Steve Brown wrote:
> > This is the bug, the AP advertises 20/40, but sets the operating channel to 20 Mhz,
> > but ath9k chooses the wrong rate table since it doesn't honor the AP's current channel width.
> >

Can you try this patch ? This will apply on top of today's wireless-testing.
Thanks.

diff --git a/drivers/net/wireless/ath9k/rc.c b/drivers/net/wireless/ath9k/rc.c
index cf0559f..9ce99e1 100644
--- a/drivers/net/wireless/ath9k/rc.c
+++ b/drivers/net/wireless/ath9k/rc.c
@@ -1387,42 +1387,18 @@ static struct ath_rate_table *ath_choose_rate_table(struct ath_softc *sc,
 static void ath_rc_init(struct ath_softc *sc,
 			struct ath_rate_priv *ath_rc_priv,
 			struct ieee80211_supported_band *sband,
-			struct ieee80211_sta *sta)
+			struct ieee80211_sta *sta,
+			struct ath_rate_table *rate_table)
 {
-	struct ath_rate_table *rate_table = NULL;
 	struct ath_rateset *rateset = &ath_rc_priv->neg_rates;
 	u8 *ht_mcs = (u8 *)&ath_rc_priv->neg_ht_rates;
 	u8 i, j, k, hi = 0, hthi = 0;
-	struct ath_hw *ah = sc->sc_ah;
-
-	/* FIXME: Adhoc */
-	if ((sc->sc_ah->opmode == NL80211_IFTYPE_STATION) ||
-	    (sc->sc_ah->opmode == NL80211_IFTYPE_ADHOC)) {
-		bool is_cw_40 = sta->ht_cap.cap & IEEE80211_HT_CAP_SUP_WIDTH_20_40;
-		rate_table = ath_choose_rate_table(sc, sband->band,
-						   sta->ht_cap.ht_supported,
-						   is_cw_40);
-	} else if (sc->sc_ah->opmode == NL80211_IFTYPE_AP) {
-		/* cur_rate_table would be set on init through config() */
-		rate_table = sc->cur_rate_table;
-	}
 
 	if (!rate_table) {
 		DPRINTF(sc, ATH_DBG_FATAL, "Rate table not initialized\n");
 		return;
 	}
 
-	if (sta->ht_cap.ht_supported) {
-		ath_rc_priv->ht_cap = WLAN_RC_HT_FLAG;
-		if (sc->sc_ah->caps.tx_chainmask != 1 &&
-			ath9k_hw_getcapability(ah, ATH9K_CAP_DS, 0, NULL))
-			ath_rc_priv->ht_cap |= WLAN_RC_DS_FLAG;
-		if (sta->ht_cap.cap & IEEE80211_HT_CAP_SUP_WIDTH_20_40)
-			ath_rc_priv->ht_cap |= WLAN_RC_40_FLAG;
-		if (sta->ht_cap.cap & IEEE80211_HT_CAP_SGI_40)
-			ath_rc_priv->ht_cap |= WLAN_RC_SGI_FLAG;
-	}
-
 	/* Initial rate table size. Will change depending
 	 * on the working rate set */
 	ath_rc_priv->rate_table_size = RATE_TABLE_SIZE;
@@ -1442,7 +1418,7 @@ static void ath_rc_init(struct ath_softc *sc,
 			ath_rc_priv->valid_phy_rateidx[i][j] = 0;
 		ath_rc_priv->valid_phy_ratecnt[i] = 0;
 	}
-	ath_rc_priv->rc_phy_mode = (ath_rc_priv->ht_cap & WLAN_RC_40_FLAG);
+	ath_rc_priv->rc_phy_mode = ath_rc_priv->ht_cap & WLAN_RC_40_FLAG;
 
 	/* Set stream capability */
 	ath_rc_priv->single_stream = (ath_rc_priv->ht_cap & WLAN_RC_DS_FLAG) ? 0 : 1;
@@ -1487,9 +1463,34 @@ static void ath_rc_init(struct ath_softc *sc,
 	ath_rc_sort_validrates(rate_table, ath_rc_priv);
 	ath_rc_priv->rate_max_phy = ath_rc_priv->valid_rate_index[k-4];
 	sc->cur_rate_table = rate_table;
+
+	DPRINTF(sc, ATH_DBG_CONFIG, "RC Initialized with capabilities: 0x%x\n",
+		ath_rc_priv->ht_cap);
 }
 
-/* Rate Control callbacks */
+static u8 ath_rc_build_ht_caps(struct ath_softc *sc, bool is_ht, bool is_cw40,
+			       bool is_sgi40)
+{
+	u8 caps = 0;
+
+	if (is_ht) {
+		caps = WLAN_RC_HT_FLAG;
+		if (sc->sc_ah->caps.tx_chainmask != 1 &&
+		    ath9k_hw_getcapability(sc->sc_ah, ATH9K_CAP_DS, 0, NULL))
+			caps |= WLAN_RC_DS_FLAG;
+		if (is_cw40)
+			caps |= WLAN_RC_40_FLAG;
+		if (is_sgi40)
+			caps |= WLAN_RC_SGI_FLAG;
+	}
+
+	return caps;
+}
+
+/***********************************/
+/* mac80211 Rate Control callbacks */
+/***********************************/
+
 static void ath_tx_status(void *priv, struct ieee80211_supported_band *sband,
 			  struct ieee80211_sta *sta, void *priv_sta,
 			  struct sk_buff *skb)
@@ -1585,6 +1586,8 @@ static void ath_rate_init(void *priv, struct ieee80211_supported_band *sband,
 {
 	struct ath_softc *sc = priv;
 	struct ath_rate_priv *ath_rc_priv = priv_sta;
+	struct ath_rate_table *rate_table = NULL;
+	bool is_cw40, is_sgi40;
 	int i, j = 0;
 
 	for (i = 0; i < sband->n_bitrates; i++) {
@@ -1606,7 +1609,63 @@ static void ath_rate_init(void *priv, struct ieee80211_supported_band *sband,
 		ath_rc_priv->neg_ht_rates.rs_nrates = j;
 	}
 
-	ath_rc_init(sc, priv_sta, sband, sta);
+	is_cw40 = sta->ht_cap.cap & IEEE80211_HT_CAP_SUP_WIDTH_20_40;
+	is_sgi40 = sta->ht_cap.cap & IEEE80211_HT_CAP_SGI_40;
+
+	/* Choose rate table first */
+
+	if ((sc->sc_ah->opmode == NL80211_IFTYPE_STATION) ||
+	    (sc->sc_ah->opmode == NL80211_IFTYPE_ADHOC)) {
+		rate_table = ath_choose_rate_table(sc, sband->band,
+						   sta->ht_cap.ht_supported,
+						   is_cw40);
+	} else if (sc->sc_ah->opmode == NL80211_IFTYPE_AP) {
+		/* cur_rate_table would be set on init through config() */
+		rate_table = sc->cur_rate_table;
+	}
+
+	ath_rc_priv->ht_cap = ath_rc_build_ht_caps(sc, sta->ht_cap.ht_supported,
+						   is_cw40, is_sgi40);
+	ath_rc_init(sc, priv_sta, sband, sta, rate_table);
+}
+
+static void ath_rate_update(void *priv, struct ieee80211_supported_band *sband,
+			    struct ieee80211_sta *sta, void *priv_sta,
+			    u32 changed)
+{
+	struct ath_softc *sc = priv;
+	struct ath_rate_priv *ath_rc_priv = priv_sta;
+	struct ath_rate_table *rate_table = NULL;
+	bool oper_cw40 = false, oper_sgi40;
+	bool local_cw40 = ((ath_rc_priv->ht_cap & WLAN_RC_40_FLAG) ? 1 : 0);
+	bool local_sgi40 = ((ath_rc_priv->ht_cap & WLAN_RC_SGI_FLAG) ? 1 : 0);
+
+	/* FIXME: Handle AP mode later when we support CWM */
+
+	if (changed & IEEE80211_RC_HT_CHANGED) {
+		if (sc->sc_ah->opmode != NL80211_IFTYPE_STATION)
+			return;
+
+		if (sc->hw->conf.channel_type == NL80211_CHAN_HT40MINUS ||
+		    sc->hw->conf.channel_type == NL80211_CHAN_HT40PLUS)
+			oper_cw40 = true;
+
+		oper_sgi40 = sta->ht_cap.cap & IEEE80211_HT_CAP_SGI_40;
+
+		if ((local_cw40 != oper_cw40) || (local_sgi40 != oper_sgi40)) {
+			rate_table = ath_choose_rate_table(sc, sband->band,
+						   sta->ht_cap.ht_supported,
+						   oper_cw40);
+			ath_rc_priv->ht_cap = ath_rc_build_ht_caps(sc,
+						   sta->ht_cap.ht_supported,
+						   oper_cw40, oper_sgi40);
+			ath_rc_init(sc, priv_sta, sband, sta, rate_table);
+
+			DPRINTF(sc, ATH_DBG_CONFIG,
+				"Operating HT Bandwidth changed to: %d\n",
+				sc->hw->conf.channel_type);
+		}
+	}
 }
 
 static void *ath_rate_alloc(struct ieee80211_hw *hw, struct dentry *debugfsdir)
@@ -1650,6 +1709,7 @@ static struct rate_control_ops ath_rate_ops = {
 	.tx_status = ath_tx_status,
 	.get_rate = ath_get_rate,
 	.rate_init = ath_rate_init,
+	.rate_update = ath_rate_update,
 	.alloc = ath_rate_alloc,
 	.free = ath_rate_free,
 	.alloc_sta = ath_rate_alloc_sta,

[ath9k-devel] AR9260 hang during connection to AP

[Steve Brown]

Sujith wrote:
> Steve Brown wrote:
>   
>>> This is the bug, the AP advertises 20/40, but sets the operating channel to 20 Mhz,
>>> but ath9k chooses the wrong rate table since it doesn't honor the AP's current channel width.
>>>
>>>       
>
> Can you try this patch ? This will apply on top of today's wireless-testing.
> Thanks.
>
>   
Thanks for the patch. I applied it to today's wireless-testing.  It 
didn't solve the problem with the assert at rc.c:745.

I put the printk's back in and the attached is what I got.

Steve

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: ath9k-hang.log
Url: http://lists.ath9k.org/pipermail/ath9k-devel/attachments/20090224/83accf8f/attachment-0001.txt 

[ath9k-devel] AR9260 hang during connection to AP

[Sujith]

Steve Brown wrote:
> hostapd.conf
> channel=6
> ht_capab=[HT40-][SHORT-GI-40][DSSS_CCK-40]
> 
> iw phy phy1 info
> Wiphy phy1
>         Band 1:
>                 HT capabilities: 0x104e
>                          * 20/40 MHz operation
>                          * SM PS disabled
>                          * 40 MHz short GI
>                          * max A-MSDU len 3839
>                          * DSSS/CCK 40 MHz
>                 HT A-MPDU factor: 0x0003 (65535 bytes)
>                 HT A-MPDU density: 0x0006 (8 usec)
>                 HT MCS set: ff ff 00 00 00 00 00 00 00 00 00 00 01 00 00 00
> 

Ah, are you using ath9k as the AP ?
Can you paste the hostapd log ? (hostapd -dd)

[ath9k-devel] AR9260 hang during connection to AP

[Steve Brown]

Sujith wrote:
> Steve Brown wrote:
>   
>> hostapd.conf
>> channel=6
>> ht_capab=[HT40-][SHORT-GI-40][DSSS_CCK-40]
>>
>> iw phy phy1 info
>> Wiphy phy1
>>         Band 1:
>>                 HT capabilities: 0x104e
>>                          * 20/40 MHz operation
>>                          * SM PS disabled
>>                          * 40 MHz short GI
>>                          * max A-MSDU len 3839
>>                          * DSSS/CCK 40 MHz
>>                 HT A-MPDU factor: 0x0003 (65535 bytes)
>>                 HT A-MPDU density: 0x0006 (8 usec)
>>                 HT MCS set: ff ff 00 00 00 00 00 00 00 00 00 00 01 00 00 00
>>
>>     
>
> Ah, are you using ath9k as the AP ?
>   
Yes, I'm using the same card for the ap. Also, I have no problem w/ g 
connections. I moved to channel 11.

hostapd is today's git.

ath9k still has printk's instead of the assert.
> Can you paste the hostapd log ? (hostapd -dd)
>   
attached.
> From your other mail, the chosen rate table is completely wrong,
> am not sure if this is related to AP mode in ath9k.
>
> Sujith
>   
Steve

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: ht.log
Url: http://lists.ath9k.org/pipermail/ath9k-devel/attachments/20090225/8837c2cc/attachment.txt