Kernel oops in ath9k

Kernel oops in ath9k

[Douglass, Woodrow]

[-- Attachment #1: Type: text/plain, Size: 720 bytes --]

Hi all,

I'm not sure if this is the right place to post this, but I may have a
bug report for the ath9k driver.

I'm running a single board computer with a 600Mhz celeron processor. On
the minipci bus I have a Ubiquiti SR71-12 802.11n card. Also attached is
a shell script with the commands I run to cause the oops. Note that the
first time the ifconfig line in this script is run, a "SIOCSIFFLAGS:
Input/Output Error" is triggered, the kernel oops happens about 5
seconds after the ifconfig line is run a second time.

If this is not the right place to post this, could someone please point
me in the right direction? Thanks.

And thanks again for donating your time to such a great project!

-Woody

[-- Attachment #2: oops.txt --]
[-- Type: text/plain, Size: 3573 bytes --]

[  292.816058] wlan0: Creating new IBSS network, BSSID 66:4a:0b:f1:cf:96
[  293.001493] BUG: unable to handle kernel NULL pointer dereference at 00000018
[  293.001658] IP: [<debccac1>] ath_beacon_setup+0x59/0xea [ath9k]
[  293.001804] *pde = 00000000 
[  293.001901] Oops: 0000 [#1] SMP 
[  293.002032] last sysfs file: /sys/class/firmware/0000:01:07.0/loading
[  293.002105] Modules linked in: loop arc4 ecb ath9k mac80211 led_class ath cfg80211 rfkill parport_pc parport snd_intel8x0 button processor snd_ac97_codec ac97_bus snd_pcm snd_timer snd soundcore i2c_i801 snd_page_alloc serio_raw shpchp i2c_core pci_hotplug intel_agp rng_core agpgart pcspkr joydev evdev usbhid hid ext3 jbd mbcache sd_mod piix ide_pci_generic ide_core ata_generic ata_piix libata scsi_mod ehci_hcd uhci_hcd e100 mii floppy usbcore nls_base thermal fan thermal_sys [last unloaded: scsi_wait_scan]
[  293.004323] 
[  293.004386] Pid: 1454, comm: phy0 Not tainted (2.6.31.6 #1)  
[  293.004459] EIP: 0060:[<debccac1>] EFLAGS: 00010246 CPU: 0
[  293.004558] EIP is at ath_beacon_setup+0x59/0xea [ath9k]
[  293.004629] EAX: 00000000 EBX: dc4d5000 ECX: dc81a974 EDX: 00000022
[  293.004703] ESI: dc81a974 EDI: dc7360c0 EBP: dc4c8000 ESP: dc8bdd04
[  293.004777]  DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
[  293.004848] Process phy0 (pid: 1454, ti=dc8bc000 task=dc1e3ae0 task.ti=dc8bc000)
[  293.004848] Stack:
[  293.004848]  dc81a974 c1013a49 00019000 dc4c8000 c1003996 00019000 00019000 00008240
[  293.004848] <0> dc4c8000 00000087 00000000 deadbeef 0000007b 0000007b 000000d8 00000000
[  293.004848] <0> c82dbeef dc4c8000 04001000 00000000 debbe8da dc187000 dc81a974 dc4c8000
[  293.004848] Call Trace:
[  293.004848]  [<c1013a49>] ? smp_apic_timer_interrupt+0x6c/0x76
[  293.004848]  [<c1003996>] ? apic_timer_interrupt+0x2a/0x30
[  293.004848]  [<debbe8da>] ? ath9k_hw_set_interrupts+0x1df/0x25d [ath9k]
[  293.004848]  [<debccd1f>] ? ath_beacon_config+0x1cd/0x362 [ath9k]
[  293.004848]  [<debcd117>] ? ath_beacon_alloc+0x1be/0x215 [ath9k]
[  293.004848]  [<debce2f6>] ? ath9k_bss_info_changed+0x15d/0x31f [ath9k]
[  293.004848]  [<debce199>] ? ath9k_bss_info_changed+0x0/0x31f [ath9k]
[  293.004848]  [<dea26fb5>] ? ieee80211_bss_info_change_notify+0x10d/0x11d [mac80211]
[  293.004848]  [<dea2b8d8>] ? __ieee80211_sta_join_ibss+0x334/0x3a6 [mac80211]
[  293.004848]  [<c117152d>] ? extract_entropy+0x3e/0x84
[  293.004848]  [<dea2bc66>] ? ieee80211_sta_find_ibss+0x31c/0x3b0 [mac80211]
[  293.004848]  [<c1021e2c>] ? update_curr+0x66/0x187
[  293.004848]  [<c10224a7>] ? dequeue_entity+0x10/0x11b
[  293.004848]  [<dea2c2c2>] ? ieee80211_ibss_work+0x291/0x34f [mac80211]
[  293.004848]  [<c1219da2>] ? schedule+0x811/0x87f
[  293.004848]  [<c10287f0>] ? finish_task_switch+0x4a/0xca
[  293.004848]  [<c103c4fb>] ? worker_thread+0x13e/0x1c0
[  293.004848]  [<dea2c031>] ? ieee80211_ibss_work+0x0/0x34f [mac80211]
[  293.004848]  [<c103fa56>] ? autoremove_wake_function+0x0/0x2d
[  293.004848]  [<c103c3bd>] ? worker_thread+0x0/0x1c0
[  293.004848]  [<c103f7b3>] ? kthread+0x6e/0x73
[  293.004848]  [<c103f745>] ? kthread+0x0/0x73
[  293.004848]  [<c1003b77>] ? kernel_thread_helper+0x7/0x10
[  293.004848] Code: 18 ba 22 00 00 00 89 03 eb 0b c7 03 00 00 00 00 ba 02 00 00 00 8b 41 1c 89 43 04 8b 0c 24 8b 81 50 07 00 00 f6 81 e4 00 00 00 20 <8b> 70 18 74 05 8a 48 19 09 ce 8b 4f 50 89 e8 52 89 da 6a 00 6a 
[  293.004848] EIP: [<debccac1>] ath_beacon_setup+0x59/0xea [ath9k] SS:ESP 0068:dc8bdd04
[  293.004848] CR2: 0000000000000018
[  293.010890] ---[ end trace 71bd7de86da68ba7 ]---

[-- Attachment #3: wireless.sh --]
[-- Type: application/octet-stream, Size: 114 bytes --]

#!/bin/bash

iwconfig wlan0 mode ad-hoc channel 1 essid atk-test
ifconfig wlan0 192.168.0.1 netmask 255.255.255.0

Re: Kernel oops in ath9k

[Pavel Roskin]

On Thu, 2009-11-19 at 09:54 -0500, Douglass, Woodrow wrote:
> Hi all,
> 
> I'm not sure if this is the right place to post this, but I may have a
> bug report for the ath9k driver.

You may want to post to ath9k-devel@lists.ath9k.org, as the problem
seems to be specific to the Atheros driver.

> I'm running a single board computer with a 600Mhz celeron processor. On
> the minipci bus I have a Ubiquiti SR71-12 802.11n card.

It would be helpful if you also specified the exact version of the
kernel.

>  Also attached is
> a shell script with the commands I run to cause the oops. Note that the
> first time the ifconfig line in this script is run, a "SIOCSIFFLAGS:
> Input/Output Error" is triggered, the kernel oops happens about 5
> seconds after the ifconfig line is run a second time.

I can reproduce "Input/Output Error" with SR71-12, but not the oops with
the current linux-wireless.

I know that it's caused by returning false in this code in file
drivers/net/wireless/ath/ath9k/calib.c, function ath9k_hw_init_cal():

                /* Poll for offset calibration complete */
                if (!ath9k_hw_wait(ah, AR_PHY_AGC_CONTROL, AR_PHY_AGC_CONTROL_CAL,
                                   0, AH_WAIT_TIMEOUT)) {
                        ath_print(common, ATH_DBG_CALIBRATE,
                                  "offset calibration failed to "
                                  "complete in 1ms; noisy environment?\n");
                        return false;
                }

With full debugging (modprobe ath9k debug=0xffffffff), the kernel
messages are (with some context):

ath: Reset TXQ, inactive queue: 7
ath: Reset TX queue: 8
ath: tx ok 0x0 err 0x0 desc 0x10f eol 0x10f urn 0x0
ath: Reset TX queue: 9
ath: tx ok 0x0 err 0x0 desc 0x10f eol 0x10f urn 0x0
ath: ah->misc_mode 0x4
ath: timeout (100000 us) on reg 0x9860: 0xdeadbeef & 0x00000001 != 0x00000000
ath: offset calibration failed to complete in 1ms; noisy environment?
ath: Unable to reset hardware; reset status -5 (freq 2412 MHz)

I have access to various Ubiquiti cards, but SR71-12 is the only one
that fails.  I know that it works in Windows XP with the Ubiquiti driver
for other SR71 cards (the inf file should be tweaked to recognize the
revision of SR71-12).  That makes me think the problem is not in the
hardware and not in the noisy environment.

ath9k identifies the card as:

phy0: Atheros AR9280 Rev:2 mem=0xffffc90010da0000, irq=18

# lspci -vnn -s 03:01.0
03:01.0 Network controller [0280]: Atheros Communications Inc. AR922X
Wireless Network Adapter [168c:0029] (rev 01)
        Subsystem: Device [0777:4002]
        Flags: bus master, 66MHz, medium devsel, latency 168, IRQ 18
        Memory at fe3f0000 (32-bit, non-prefetchable) [size=64K]
        Capabilities: [44] Power Management version 2
        Kernel driver in use: ath9k
        Kernel modules: ath9k

-- 
Regards,
Pavel Roskin

RE: Kernel oops in ath9k

[Douglass, Woodrow]

Thanks for the quick reply Pavel! I've seen this Oops with kernel
2.6.31.6, as well as with the  latest linux-wireless that I downloaded
on the 18th of november. Also, it's interesting that the card is
identified as having a AR9280 chipset, the website for the card
(http://www.ubnt.com/products/sr7112.php) says the card has an AR9220
chipset. Could this be causing the problem?

I'm going to repost this to the ath9k mailing list you mentioned.

Thanks again,
Woody

> -----Original Message-----
> From: Pavel Roskin [mailto:proski@gnu.org] 
> Sent: Thursday, November 19, 2009 10:25 PM
> To: Douglass, Woodrow
> Cc: linux-wireless@vger.kernel.org
> Subject: Re: Kernel oops in ath9k
> 
> On Thu, 2009-11-19 at 09:54 -0500, Douglass, Woodrow wrote:
> > Hi all,
> > 
> > I'm not sure if this is the right place to post this, but I 
> may have a 
> > bug report for the ath9k driver.
> 
> You may want to post to ath9k-devel@lists.ath9k.org, as the 
> problem seems to be specific to the Atheros driver.
> 
> > I'm running a single board computer with a 600Mhz celeron 
> processor. 
> > On the minipci bus I have a Ubiquiti SR71-12 802.11n card.
> 
> It would be helpful if you also specified the exact version 
> of the kernel.
> 
> >  Also attached is
> > a shell script with the commands I run to cause the oops. Note that 
> > the first time the ifconfig line in this script is run, a 
> "SIOCSIFFLAGS:
> > Input/Output Error" is triggered, the kernel oops happens about 5 
> > seconds after the ifconfig line is run a second time.
> 
> I can reproduce "Input/Output Error" with SR71-12, but not 
> the oops with the current linux-wireless.
> 
> I know that it's caused by returning false in this code in 
> file drivers/net/wireless/ath/ath9k/calib.c, function 
> ath9k_hw_init_cal():
> 
>                 /* Poll for offset calibration complete */
>                 if (!ath9k_hw_wait(ah, AR_PHY_AGC_CONTROL, 
> AR_PHY_AGC_CONTROL_CAL,
>                                    0, AH_WAIT_TIMEOUT)) {
>                         ath_print(common, ATH_DBG_CALIBRATE,
>                                   "offset calibration failed to "
>                                   "complete in 1ms; noisy 
> environment?\n");
>                         return false;
>                 }
> 
> With full debugging (modprobe ath9k debug=0xffffffff), the 
> kernel messages are (with some context):
> 
> ath: Reset TXQ, inactive queue: 7
> ath: Reset TX queue: 8
> ath: tx ok 0x0 err 0x0 desc 0x10f eol 0x10f urn 0x0
> ath: Reset TX queue: 9
> ath: tx ok 0x0 err 0x0 desc 0x10f eol 0x10f urn 0x0
> ath: ah->misc_mode 0x4
> ath: timeout (100000 us) on reg 0x9860: 0xdeadbeef & 
> 0x00000001 != 0x00000000
> ath: offset calibration failed to complete in 1ms; noisy environment?
> ath: Unable to reset hardware; reset status -5 (freq 2412 MHz)
> 
> I have access to various Ubiquiti cards, but SR71-12 is the 
> only one that fails.  I know that it works in Windows XP with 
> the Ubiquiti driver for other SR71 cards (the inf file should 
> be tweaked to recognize the revision of SR71-12).  That makes 
> me think the problem is not in the hardware and not in the 
> noisy environment.
> 
> ath9k identifies the card as:
> 
> phy0: Atheros AR9280 Rev:2 mem=0xffffc90010da0000, irq=18
> 
> # lspci -vnn -s 03:01.0
> 03:01.0 Network controller [0280]: Atheros Communications 
> Inc. AR922X Wireless Network Adapter [168c:0029] (rev 01)
>         Subsystem: Device [0777:4002]
>         Flags: bus master, 66MHz, medium devsel, latency 168, IRQ 18
>         Memory at fe3f0000 (32-bit, non-prefetchable) [size=64K]
>         Capabilities: [44] Power Management version 2
>         Kernel driver in use: ath9k
>         Kernel modules: ath9k
> 
> --
> Regards,
> Pavel Roskin
> 

Re: Kernel oops in ath9k

[Luis R. Rodriguez]

On Fri, Nov 20, 2009 at 5:56 AM, Douglass, Woodrow
<Woodrow.Douglass@qinetiq-na.com> wrote:
> Thanks for the quick reply Pavel! I've seen this Oops with kernel
> 2.6.31.6, as well as with the  latest linux-wireless that I downloaded
> on the 18th of november. Also, it's interesting that the card is
> identified as having a AR9280 chipset, the website for the card
> (http://www.ubnt.com/products/sr7112.php) says the card has an AR9220
> chipset. Could this be causing the problem?

Nah, your card is just PCI version of the AR9280. We just need to
distinguish that on the print on ath9k load. So PCI-express is AR9280
and mini-pci AR9220.

> I'm going to repost this to the ath9k mailing list you mentioned.

linux-wireless is fine too though no need to repost there. I just
tested your script with AR5416 and it worked fine, can you please try
the latest compat-wireless, either latest stable or bleeding edge?

http://wireless.kernel.org/en/users/Download/stable
http://wireless.kernel.org/en/users/Download

  Luis

RE: Kernel oops in ath9k

[Douglass, Woodrow]

Ok I compiled compat-wireless-2.6.32-rc7.tar.bz2, ran the script, and now the kernel outright crashed rather then just oopsing. The terminal is all locked up, is there any way I can post the crash dump here? I'm sorry, but I'm kind of new at this.

Thanks again,
-Woody
 

> -----Original Message-----
> From: Luis R. Rodriguez [mailto:mcgrof@gmail.com] 
> Sent: Friday, November 20, 2009 9:54 AM
> To: Douglass, Woodrow
> Cc: Pavel Roskin; linux-wireless@vger.kernel.org
> Subject: Re: Kernel oops in ath9k
> 
> On Fri, Nov 20, 2009 at 5:56 AM, Douglass, Woodrow 
> <Woodrow.Douglass@qinetiq-na.com> wrote:
> > Thanks for the quick reply Pavel! I've seen this Oops with kernel 
> > 2.6.31.6, as well as with the  latest linux-wireless that I 
> downloaded 
> > on the 18th of november. Also, it's interesting that the card is 
> > identified as having a AR9280 chipset, the website for the card
> > (http://www.ubnt.com/products/sr7112.php) says the card has 
> an AR9220 
> > chipset. Could this be causing the problem?
> 
> Nah, your card is just PCI version of the AR9280. We just 
> need to distinguish that on the print on ath9k load. So 
> PCI-express is AR9280 and mini-pci AR9220.
> 
> > I'm going to repost this to the ath9k mailing list you mentioned.
> 
> linux-wireless is fine too though no need to repost there. I 
> just tested your script with AR5416 and it worked fine, can 
> you please try the latest compat-wireless, either latest 
> stable or bleeding edge?
> 
> http://wireless.kernel.org/en/users/Download/stable
> http://wireless.kernel.org/en/users/Download
> 
>   Luis
> 

Re: Kernel oops in ath9k

[Luis R. Rodriguez]

On Fri, Nov 20, 2009 at 8:13 AM, Douglass, Woodrow
<Woodrow.Douglass@qinetiq-na.com> wrote:
> Ok I compiled compat-wireless-2.6.32-rc7.tar.bz2, ran the script, and now the
> kernel outright crashed rather then just oopsing. The terminal is all locked up,
> is there any way I can post the crash dump here?

Are you asking for help in trying to get the dump or for where to post
the dump? If the later then please use a pastebin or upload your file
somwhere.

  Luis

RE: Kernel oops in ath9k

[Douglass, Woodrow]

http://img406.imageshack.us/img406/910/kernelcrash.jpg

After running the script I sent earlier, this is what happens with the
latest released compat-wireless. Sorry if the image is a little blurry
at the top, it's the best camera I could get on short notice.

Thanks again
-Woody

> -----Original Message-----
> From: Luis R. Rodriguez [mailto:mcgrof@gmail.com] 
> Sent: Friday, November 20, 2009 11:40 AM
> To: Douglass, Woodrow
> Cc: Pavel Roskin; linux-wireless@vger.kernel.org
> Subject: Re: Kernel oops in ath9k
> 
> On Fri, Nov 20, 2009 at 8:13 AM, Douglass, Woodrow 
> <Woodrow.Douglass@qinetiq-na.com> wrote:
> > Ok I compiled compat-wireless-2.6.32-rc7.tar.bz2, ran the 
> script, and 
> > now the kernel outright crashed rather then just oopsing. 
> The terminal 
> > is all locked up, is there any way I can post the crash dump here?
> 
> Are you asking for help in trying to get the dump or for 
> where to post the dump? If the later then please use a 
> pastebin or upload your file somwhere.
> 
>   Luis
>