[ath9k-devel] AR9285: Possible intermittent connection drop-outs

[ath9k-devel] AR9285: Possible intermittent connection drop-outs

[Torquil Macdonald Sørensen]

Hi guys!

Essential, I'd like recommendations as to which ath9k debugging options I would 
want to use to effectively diagnose the following problem:

I'm having some problems with my AR9285 based wireless card on an Asus X5DID 
laptop, using Linux 2.6.35.2 on Debian Sid. I'm connecting to a D-Link DIR-615 
wireless router with the newest firmware.

The problem is: sometimes, even though Network Manager displays a good 
connection strength, I loose communication with the router and therefore also 
the internet. In this state, which is temporary, I cannot ping the router.

Typical scenario:

While I am browsing, suddenly Firefox is no longer able to receive web pages. I 
typically try to ping the router at 192.168.0.1, but get no reply or very slow 
replies and many lost pings.

I have a sneaking suspicion that my pinging efforts are able to "wake" the 
connection up to some degree, since often times web pages are able to load just 
after I start to ping.

The problem fixes itself, usually after a few minutes. But the problem will come 
and go quite randomly. Network Manager never notices any of this. It happily 
displays a good connection all the time.

Most of the time everything works fine. I'm quite certain that wireless in the 
Windows 7 OS on the same laptop works flawlessly, so its not a hardware or 
router problem. The other guy here uses Windows XP with a USB wireless card and 
never has similar problems.

I have recompiled Linux to get some debugging going, but there are lots of debug 
options and some of them result in a massive amount of information being logged. 
So I'd like to get some recommendation as to which debugging option I should try 
first, out of these:

http://wireless.kernel.org/en/users/Drivers/ath9k/debug

The settings on the router are:

Enable auto channel selection: on
Transmission rate: Best (auto)
WMM Enable: on (wireless QoS)

Security mode: WPA/WPA2 Wireless Security (enhanced)
Cipher type:	AUTO (TKIP/AES)
PSK/EAP:	PSK
Network Key:	******** (hidden)

Beacon interval: 100ms
RTS threshold: 2346
Fragmentation: 2346
DTIM interval: 1
Preamble type: Short
CTS mode: Auto
Wireless mode: 802.11 Mixed (g/b)

 From 'lspci':

04:00.0 Network controller: Atheros Communications Inc. AR9285 Wireless Network 
Adapter (PCI-Express) (rev 01)

[ath9k-devel] AR9285: Possible intermittent connection drop-outs

[Peter Stuge]

Torquil Macdonald S?rensen wrote:
> I'd like recommendations as to which ath9k debugging options I
> would want to use to effectively diagnose the following problem:

Don't hold your breath. It seems there are no resources for this kind
of debugging. Hopefully the driver will improve further over time.
(It has already improved a lot!)


//Peter

[ath9k-devel] AR9285: Possible intermittent connection drop-outs

[Alan]

On Wed, Aug 18, 2010 at 1:10 AM, Torquil Macdonald S?rensen
<torquil@gmail.com> wrote:
> I'm having some problems with my AR9285 based wireless card on an Asus X5DID
> laptop, using Linux 2.6.35.2 on Debian Sid. I'm connecting to a D-Link DIR-615
> wireless router with the newest firmware.

If you browse the mailing list archive you will see that me and others
have already reported this problem a long time ago, yet everything
remains the same.

I just got tired of waiting and exchanged the Atheros card with an
Intel 5100 and everything works fine now.
The Intel 5100 costs 10$ on ebay, and Intel's drivers for Linux work just fine.

[ath9k-devel] AR9285: Possible intermittent connection drop-outs

[Ben Gamari]

On Thu, 19 Aug 2010 02:08:51 +0900, Alan <lameventanas@gmail.com> wrote:
> I just got tired of waiting and exchanged the Atheros card with an
> Intel 5100 and everything works fine now.
> The Intel 5100 costs 10$ on ebay, and Intel's drivers for Linux work just fine.

Same situation here. Switched to a Intel 5300 and things are hunky dory
now.

- Ben

[ath9k-devel] AR9285: Possible intermittent connection drop-outs

[Peter Stuge]

Alan wrote:
> The Intel 5100 costs 10$ on ebay, and Intel's drivers for Linux
> work just fine.

Seems there's little choice but ath9k for plain minipci cards though. :\


//Peter

[ath9k-devel] AR9285: Possible intermittent connection drop-outs

[Adrian Chadd]

So there's plenty of things you can vary given the current setup to
try and eliminate things.

* Can you buy one of those dlink units for testing? (ie, so your
testing doesn't interfere with others)
* Is it 11a, 11b, 11bg, 11n? 2.4 or 5ghz? It may be linked to the
underlying media you're using
* The OP is using WPA2. It may be related to crypto key management
* It may be related to other people using the unit, and/or spectrum
* It may be related to the channel you're using (eg, the AR5416 I have
here goes into a "stuck beacon" loop in hostap mode if I run it on
channel 11, but channels 1 and 6 are fine.)
* It may be occuring when the unit/client changes channel (if you're
roaming about, and/or the AP dynamically changes channel.)

Some of these overlap (eg if you change channel, you are changing how
the radio is setup but you're also changing how your AP behaves and
possibly changing the RF environment - what other clients/APs/RF noise
are on those channels.) So make up a diagram of the above, figure out
what internal/external factors you think can contribute, and then look
at varying one at a time.

As a developer (and as one who has debugged a few of these wireless
issues lately!) the worst thing you can be told is "it doesn't work
right." It may take some time, but narrowing down a specific,
controllable instance showing the deviation in behaviour is going to
help the ath9k developers get a potential clue where to look and what
debugging to turn on.

2c,

Adrian

On 19 August 2010 01:08, Alan <lameventanas@gmail.com> wrote:
> On Wed, Aug 18, 2010 at 1:10 AM, Torquil Macdonald S?rensen
> <torquil@gmail.com> wrote:
>> I'm having some problems with my AR9285 based wireless card on an Asus X5DID
>> laptop, using Linux 2.6.35.2 on Debian Sid. I'm connecting to a D-Link DIR-615
>> wireless router with the newest firmware.
>
> If you browse the mailing list archive you will see that me and others
> have already reported this problem a long time ago, yet everything
> remains the same.
>
> I just got tired of waiting and exchanged the Atheros card with an
> Intel 5100 and everything works fine now.
> The Intel 5100 costs 10$ on ebay, and Intel's drivers for Linux work just fine.
> _______________________________________________
> ath9k-devel mailing list
> ath9k-devel at lists.ath9k.org
> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
>

[ath9k-devel] AR9285: Possible intermittent connection drop-outs

[Torquil Macdonald Sørensen]

On 19/08/10 03:22, Adrian Chadd wrote:
> So there's plenty of things you can vary given the current setup to
> try and eliminate things.
>
> * Can you buy one of those dlink units for testing? (ie, so your
> testing doesn't interfere with others)
> * Is it 11a, 11b, 11bg, 11n? 2.4 or 5ghz? It may be linked to the
> underlying media you're using
> * The OP is using WPA2. It may be related to crypto key management
> * It may be related to other people using the unit, and/or spectrum
> * It may be related to the channel you're using (eg, the AR5416 I have
> here goes into a "stuck beacon" loop in hostap mode if I run it on
> channel 11, but channels 1 and 6 are fine.)
> * It may be occuring when the unit/client changes channel (if you're
> roaming about, and/or the AP dynamically changes channel.)
>
> Some of these overlap (eg if you change channel, you are changing how
> the radio is setup but you're also changing how your AP behaves and
> possibly changing the RF environment - what other clients/APs/RF noise
> are on those channels.) So make up a diagram of the above, figure out
> what internal/external factors you think can contribute, and then look
> at varying one at a time.
>
> As a developer (and as one who has debugged a few of these wireless
> issues lately!) the worst thing you can be told is "it doesn't work
> right." It may take some time, but narrowing down a specific,
> controllable instance showing the deviation in behaviour is going to
> help the ath9k developers get a potential clue where to look and what
> debugging to turn on.
>
> 2c,
>
> Adrian

Thanks Adrian! This is very useful information, I'll make sure to study it in 
detail. It is 2.432GHz. I have chosen "11b/g" on the router, intensionally 
disabling 11n since there are other bottlenecks in the system. I'm pretty sure 
that I'm using 11g.

I don't seem to have experienced the problem (at least not as noticeable) in the 
last couple of days. Lately I have started to compile my own kernels, and in the 
process turned off a lot of unecessary kernel compile options.

Btw, I know that there are no kernel log messages when the problems occurs. 
That's partly why i have been compiling my own kernels, in order to activate the 
ath9k debugging.

Torquil

[ath9k-devel] AR9285: Possible intermittent connection drop-outs

[Adrian Chadd]

Good work!

Make sure you test the old kernels which exhibited the issue. See if
the problem returns. If it does, it's likely something in the
recompiling (version, compiler, options, etc - not in that order!) has
something to do with it. If it doesn't, it may be environmental.


Adrian

On 20 August 2010 07:43, Torquil Macdonald S?rensen <torquil@gmail.com> wrote:
> On 19/08/10 03:22, Adrian Chadd wrote:
>>
>> So there's plenty of things you can vary given the current setup to
>> try and eliminate things.
>>
>> * Can you buy one of those dlink units for testing? (ie, so your
>> testing doesn't interfere with others)
>> * Is it 11a, 11b, 11bg, 11n? 2.4 or 5ghz? It may be linked to the
>> underlying media you're using
>> * The OP is using WPA2. It may be related to crypto key management
>> * It may be related to other people using the unit, and/or spectrum
>> * It may be related to the channel you're using (eg, the AR5416 I have
>> here goes into a "stuck beacon" loop in hostap mode if I run it on
>> channel 11, but channels 1 and 6 are fine.)
>> * It may be occuring when the unit/client changes channel (if you're
>> roaming about, and/or the AP dynamically changes channel.)
>>
>> Some of these overlap (eg if you change channel, you are changing how
>> the radio is setup but you're also changing how your AP behaves and
>> possibly changing the RF environment - what other clients/APs/RF noise
>> are on those channels.) So make up a diagram of the above, figure out
>> what internal/external factors you think can contribute, and then look
>> at varying one at a time.
>>
>> As a developer (and as one who has debugged a few of these wireless
>> issues lately!) the worst thing you can be told is "it doesn't work
>> right." It may take some time, but narrowing down a specific,
>> controllable instance showing the deviation in behaviour is going to
>> help the ath9k developers get a potential clue where to look and what
>> debugging to turn on.
>>
>> 2c,
>>
>> Adrian
>
> Thanks Adrian! This is very useful information, I'll make sure to study it
> in detail. It is 2.432GHz. I have chosen "11b/g" on the router,
> intensionally disabling 11n since there are other bottlenecks in the system.
> I'm pretty sure that I'm using 11g.
>
> I don't seem to have experienced the problem (at least not as noticeable) in
> the last couple of days. Lately I have started to compile my own kernels,
> and in the process turned off a lot of unecessary kernel compile options.
>
> Btw, I know that there are no kernel log messages when the problems occurs.
> That's partly why i have been compiling my own kernels, in order to activate
> the ath9k debugging.
>
> Torquil
>
>

[ath9k-devel] AR9285: Possible intermittent connection drop-outs

[Torquil Macdonald Sørensen]

On 19/08/10 15:15, Mohammed Shafi wrote:
> Hi Alan ,
>
>                       Can you please provide the dmesg and kern.log when
> the driver hits the issue.
>
> with thanks,
> shafi

Hi Shafi (and others)!

Sorry if everybody/somebody gets this twice. The network problems I'm trying to 
debug are occurring at this very moment :-)

I have some more information now before I try newer drivers. I have looked at 
the "ATH_DBG_ANI" debugging of the ath9k driver. I.e. I have enabled debugging 
by running:

echo "0x00000040" > /sys/kernel/debug/ath9k/phy0/debug

Now, I noticed some error messages that start to occur at the same time as my 
connection problems.

You'll be able to see log extracts below, but first I'll mention here the 
following recurring error messages:

"ath: invalid cmd 2"

and

"ath: cycleCount=-1593549770, ofdmPhyErrCount=3, cckPhyErrCount=60"

I.e. the last one shows that some kind of error has occured 60 times at that 
point in the log. I have no idea what it means, only that these seem like 
errors. So I just need you ath9k gurus to interpret them for me. Also, it seems 
a bit strange to me that what seems like a "counting variable" is negative 
(cycleCount), but I really know nothing about this.

When the connection gets back into a healthy state, the "invalid cmd: 2" no 
longer occurs, cycleCount is positive again, and the PhyErrCount goes back to 0.

Here is some log extracts showing more details when the problem occurs:

.
.
.
Aug 21 14:02:53 asus kernel: ath: ANI parameters:
Aug 21 14:02:53 asus kernel: ath: noiseImmunityLevel=0, spurImmunityLevel=0, 
ofdmWeakSigDetectOff=1
Aug 21 14:02:53 asus kernel: ath: cckWeakSigThreshold=0, firstepLevel=1, 
listenTime=183
Aug 21 14:02:53 asus kernel: ath: cycleCount=-1602349766, ofdmPhyErrCount=1, 
cckPhyErrCount=62
Aug 21 14:02:53 asus kernel:
Aug 21 14:02:53 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:02:53 asus kernel: ath:  calibrate chan 2432/c0 nf: -117
Aug 21 14:02:53 asus kernel: ath: invalid cmd 2
Aug 21 14:02:53 asus kernel: ath: ANI parameters:
Aug 21 14:02:53 asus kernel: ath: noiseImmunityLevel=0, spurImmunityLevel=0, 
ofdmWeakSigDetectOff=1
Aug 21 14:02:53 asus kernel: ath: cckWeakSigThreshold=0, firstepLevel=2, 
listenTime=188
Aug 21 14:02:53 asus kernel: ath: cycleCount=-1593549770, ofdmPhyErrCount=3, 
cckPhyErrCount=60
.
.
.

and

.
.
.
Aug 21 14:02:53 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:02:54 asus kernel: ath: invalid cmd 2
Aug 21 14:02:54 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:02:54 asus kernel: ath: invalid cmd 2
Aug 21 14:02:54 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:02:54 asus kernel: ath: invalid cmd 2
Aug 21 14:02:54 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:02:54 asus kernel: ath: shortcal @4299689952
Aug 21 14:02:54 asus kernel: ath: invalid cmd 2
Aug 21 14:02:54 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:02:54 asus kernel: ath:  calibrate chan 2432/c0 nf: -117
Aug 21 14:02:54 asus kernel: ath: invalid cmd 2
Aug 21 14:02:54 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:02:55 asus kernel: ath: invalid cmd 2
Aug 21 14:02:55 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:02:55 asus kernel: ath: invalid cmd 2
Aug 21 14:02:55 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:02:55 asus kernel: ath: invalid cmd 2
Aug 21 14:02:55 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:02:55 asus kernel: ath: shortcal @4299690952
Aug 21 14:02:55 asus kernel: ath: invalid cmd 2
Aug 21 14:02:55 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:02:55 asus kernel: ath:  calibrate chan 2432/c0 nf: -117
Aug 21 14:02:55 asus kernel: ath: invalid cmd 2
Aug 21 14:02:55 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
.
.
.

Anything interesting here?

I'll now compile linux 2.6.35.3 and use that instead. Or perhaps I would need to 
go to 2.6.36-rc1 or 2.6.36-rc1-git3 to get a more up-to-date ath9k driver? I'm 
just looking at what is available for download at the moment on www.kernel.org


On the other hand, when the connection is back to normal I seem to only get this 
stuff:

.
.
.
Aug 21 14:34:50 asus kernel: ath:  calibrate chan 2432/c0 nf: -117
Aug 21 14:34:51 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:34:51 asus kernel: ath: shortcal @4301606655
Aug 21 14:34:51 asus kernel: ath:  calibrate chan 2432/c0 nf: -117
Aug 21 14:34:51 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:34:52 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:34:52 asus kernel: ath: shortcal @4301607655
Aug 21 14:34:52 asus kernel: ath:  calibrate chan 2432/c0 nf: -117
Aug 21 14:34:52 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:34:53 asus kernel: ath: shortcal @4301608655
Aug 21 14:34:53 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:34:53 asus kernel: ath:  calibrate chan 2432/c0 nf: -117
Aug 21 14:34:54 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:34:54 asus kernel: ath: shortcal @4301609655
Aug 21 14:34:54 asus kernel: ath:  calibrate chan 2432/c0 nf: -117
Aug 21 14:34:54 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:34:55 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:34:55 asus kernel: ath: shortcal @4301610655
Aug 21 14:34:55 asus kernel: ath:  calibrate chan 2432/c0 nf: -117
Aug 21 14:34:55 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:34:56 asus kernel: ath: shortcal @4301611655
Aug 21 14:34:56 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:34:56 asus kernel: ath:  calibrate chan 2432/c0 nf: -117
Aug 21 14:34:57 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:34:57 asus kernel: ath: shortcal @4301612655
Aug 21 14:34:57 asus kernel: ath:  calibrate chan 2432/c0 nf: -117
Aug 21 14:34:57 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
Aug 21 14:34:58 asus kernel: ath: Writing ofdmbase=12582412   cckbase=12582712
.
.

without any "invalid cmd 2" messages.

Thanks!
Torquil S?rensen

[ath9k-devel] AR9285: Possible intermittent connection drop-outs

[Torquil Macdonald Sørensen]

On 23/08/10 14:05, Mohammed Shafi wrote:
> Hi Torquil,
>
>                  Thanks for your ath9k debug messages .I tested the
> ath9k driver in 2.6.35.2
>
> with D-LINK DIR655  router  in ubunt 10.04 , but could not    reproduce
> the issue .I got  all
>
> of  your   similar debugging  messages(invalid cmd 2  several times
> ,cycle count
>
> ,ofdm,cck etc   under healthy connection( which you had  suspected for
> the connection loss ).
>
>                I think those were mainly  (modulation schemes ofdm ,cck ).
>
> while cmd is related to ANI information (enum ath9k_ani_cmd in ani.h).
>
>
> I had got these things under healthy connection ,
>
>   6464.669170] ath: Writing ofdmbase=12582412   cckbase=12582712
> [ 6464.728750] ath: invalid cmd 2
>
>   Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264144] ath: ANI parameters:
> 1994 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264149] ath:
> noiseImmunityLevel=0, spurImmunityLevel=7, ofdmWeakSigDetectOff=0
> 1995 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264156] ath:
> cckWeakSigThreshold=0, firstepLevel=2, listenTime=78
> 1996 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264163] ath:
> cycleCount=1452171193, ofdmPhyErrCount=436, cckPhyErrCount=14
> 1997 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264169]
> 1998 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264176] ath: ANI
> parameters:
> 1999 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264181] ath:
> noiseImmunityLevel=0, spurImmunityLevel=7, ofdmWeakSigDetectOff=0
> 2000 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264188] ath:
> cckWeakSigThreshold=0, firstepLevel=2, listenTime=78
> 2001 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264195] ath:
> cycleCount=1452171193, ofdmPhyErrCount=436, cckPhyErrCount=14
>
>
>
>                         I hope that enabling ATH_DBG_FATAL will give us
> some some clear idea.
>
> If you obtain the ath9k debug messages with ATH_DBG_FATAL please do send
> as soon
>
> as you get it.
>
>
> with thanks and regards,
>
> shafi

Thanks for your effort! Your results are interesting. I will check the FATAL DBG 
option too see if it gives me some more interesting info.

I will also determine for sure if the error messages I mentioned earlier only 
appear when I have connection problems. I'll probably make a simple script that 
will parse the ath9k log and pick out only those messages so I can see exactly 
when they occur.

I will get back to you when I have more info.

Thanks again for your help,
Torquil S?rensen

[ath9k-devel] AR9285: Possible intermittent connection drop-outs

[Torquil Macdonald Sørensen]

On 23/08/10 14:05, Mohammed Shafi wrote:
> Hi Torquil,
>
>                  Thanks for your ath9k debug messages .I tested the
> ath9k driver in 2.6.35.2
>
> with D-LINK DIR655  router  in ubunt 10.04 , but could not    reproduce
> the issue .I got  all
>
> of  your   similar debugging  messages(invalid cmd 2  several times
> ,cycle count
>
> ,ofdm,cck etc   under healthy connection( which you had  suspected for
> the connection loss ).
>
>                I think those were mainly  (modulation schemes ofdm ,cck ).
>
> while cmd is related to ANI information (enum ath9k_ani_cmd in ani.h).
>
>
> I had got these things under healthy connection ,
>
>   6464.669170] ath: Writing ofdmbase=12582412   cckbase=12582712
> [ 6464.728750] ath: invalid cmd 2
>
>   Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264144] ath: ANI parameters:
> 1994 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264149] ath:
> noiseImmunityLevel=0, spurImmunityLevel=7, ofdmWeakSigDetectOff=0
> 1995 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264156] ath:
> cckWeakSigThreshold=0, firstepLevel=2, listenTime=78
> 1996 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264163] ath:
> cycleCount=1452171193, ofdmPhyErrCount=436, cckPhyErrCount=14
> 1997 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264169]
> 1998 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264176] ath: ANI
> parameters:
> 1999 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264181] ath:
> noiseImmunityLevel=0, spurImmunityLevel=7, ofdmWeakSigDetectOff=0
> 2000 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264188] ath:
> cckWeakSigThreshold=0, firstepLevel=2, listenTime=78
> 2001 Aug 23 16:35:51 atheros-laptop kernel: [ 6345.264195] ath:
> cycleCount=1452171193, ofdmPhyErrCount=436, cckPhyErrCount=14
>
>
>
>                         I hope that enabling ATH_DBG_FATAL will give us
> some some clear idea.
>
> If you obtain the ath9k debug messages with ATH_DBG_FATAL please do send
> as soon
>
> as you get it.

Unfortunately, I get no debug output from the FATAL debugging when the problems 
come and go. I also discovered that the same problem persists even when using no 
encryption/authentication on the network.

In addition, this is an area with little to no other wireless networks, and 
problems persists even when I am the only user of the network.

Using Linux 2.6.35.4 at the moment.

Sometimes I get:

ath: tx hung, resetting the chip
ath: DMA failed to stop in 10 ms AR_CR=0x00000024 AR_DIAG_SW=0x00000020

but I suspect that it is not related to my problems.

Best regards
Torquil S?rensen

[ath9k-devel] AR9285: Possible intermittent connection drop-outs

[Dan Borlovan]

Just got a DIR-615 router and a free ticket into the "mee to" vagon. DIR 
running 4.11 configured as AP only

AR9285 (eeepc 1005p)

Ubuntu 10.10 kernel 2.6.35-22, with and without compat-wireless-2.6.35-1

If the AP is set to 20/40MHz I connect at 150Mbps but every few minutes 
I have long intervals in which no packet goes by. Link stays up, nothing 
in dmesg. No matter if wlan0 powersaving is on or of. Long term packet 
loss with the AP around 11-14%. A few times ifconfig wlan0 showed 1Mbps.

BUT if I switch ap to 20MHz only I connect at (only) 65Mbps with no 
problems.

Under windows 7 no problems so far.

This is one of the "pauses"

64 bytes from 10.0.0.14: icmp_req=74 ttl=64 time=0.979 ms
64 bytes from 10.0.0.14: icmp_req=75 ttl=64 time=1.40 ms
64 bytes from 10.0.0.14: icmp_req=76 ttl=64 time=5.29 ms
<here comes the pause>
64 bytes from 10.0.0.14: icmp_req=168 ttl=64 time=44.4 ms
64 bytes from 10.0.0.14: icmp_req=169 ttl=64 time=1.06 ms
64 bytes from 10.0.0.14: icmp_req=170 ttl=64 time=1.00 ms

Is this by any chance fixed in newer kernel/compat-wireless or for now 
there is no hope for me?

-- 
Dan Borlovan
Datagroup-Int