[ath9k-devel] Question about ath9k signal strength (AP mode)

[ath9k-devel] Question about ath9k signal strength (AP mode)

[Lars Hardy]

It seems that there is a difference in signal strength between the ath9k 
(latest openwrt-trunk with compat-wireless 08.07.2010) driver and the 
latest dd-wrt (madwifi N?) driver.
The dd-wrt driver has better signal strength and performance when a STA 
is some distance from the AP. (like 1 floor/2 rooms between)
I have tried with 2 different Atheros chipset in AP mode, the AR5416 and 
AR9280. dd-wrt gives a stronger signal with both chipsets compared with 
openwrt.

I know the ath9k is under development, so my question is if this is 
known by the development team and therefor will be worked on?

Lars

 

[ath9k-devel] Question about ath9k signal strength (AP mode)

[Peter Stuge]

Lars Hardy wrote:
> I have tried with 2 different Atheros chipset in AP mode, the AR5416
> and AR9280. dd-wrt gives a stronger signal with both chipsets
> compared with openwrt.
> 
> I know the ath9k is under development, so my question is if this is
> known by the development team and therefor will be worked on?

I continue to have problems with ath9k and development is slower than
I first expected since Atheros seems to have very limited focus on
support, and only for the very latest generation hardware. (I wish
they made this fact more clear.) Community resources are of course
also limited, but great improvements have been made to ath9k by
members of the OpenWRT community and I guess they will continue their
work.

Without looking much at the code, but following this list for half a
year, my gut feeling is that there is still too much major work
required on ath9k at this point for "fine tuning" such as signal
strength optimization to come up on the agenda in the near future..


//Peter

[ath9k-devel] Question about ath9k signal strength (AP mode)

[Felix Fietkau]

On 2010-07-11 3:37 AM, Peter Stuge wrote:
> Lars Hardy wrote:
>> I have tried with 2 different Atheros chipset in AP mode, the AR5416
>> and AR9280. dd-wrt gives a stronger signal with both chipsets
>> compared with openwrt.
>> 
>> I know the ath9k is under development, so my question is if this is
>> known by the development team and therefor will be worked on?
> 
> I continue to have problems with ath9k and development is slower than
> I first expected since Atheros seems to have very limited focus on
> support, and only for the very latest generation hardware. (I wish
> they made this fact more clear.) Community resources are of course
> also limited, but great improvements have been made to ath9k by
> members of the OpenWRT community and I guess they will continue their
> work.
> 
> Without looking much at the code, but following this list for half a
> year, my gut feeling is that there is still too much major work
> required on ath9k at this point for "fine tuning" such as signal
> strength optimization to come up on the agenda in the near future..
I disagree. The major work has been done and hardware support has mostly
caught up with Atheros' own codebase. What's missing is debugging work
and fine tuning - and other than that, maybe a few things that aren't
very hardware specific.

- Felix

[ath9k-devel] Question about ath9k signal strength (AP mode)

[RHS Linux User]

Hi All,

   Some VERY good questions.

   A *LOW* level interface would seem like a good beginning. i.e.- So
anyone can confirm what is really going on at the RF level and confirm
things are working as expected.

   Maybe that is a called a "debug" interface ??!!

   IMHO - there is a LOT of uncharted territory. Perhaps not even known
by the chip manufacturer?

   That said, *IF* we can get anything to work reliably we can probably
improve from there. Do "we" then own the intellectual property we have
created? And what DOES that mean ??

   Or does it mean that the manufacturer announces a "new" chip to further
confuse things?

   I continue to be amazed at the continuing and seeming "problems" and
the lack of any clear understanding or resolution of them.

   wiz
 

On Sun, 11 Jul 2010, Felix Fietkau wrote:

> On 2010-07-11 3:37 AM, Peter Stuge wrote:
> > Lars Hardy wrote:
> >> I have tried with 2 different Atheros chipset in AP mode, the AR5416
> >> and AR9280. dd-wrt gives a stronger signal with both chipsets
> >> compared with openwrt.
> >> 
> >> I know the ath9k is under development, so my question is if this is
> >> known by the development team and therefor will be worked on?
> > 
> > I continue to have problems with ath9k and development is slower than
> > I first expected since Atheros seems to have very limited focus on
> > support, and only for the very latest generation hardware. (I wish
> > they made this fact more clear.) Community resources are of course
> > also limited, but great improvements have been made to ath9k by
> > members of the OpenWRT community and I guess they will continue their
> > work.
> > 
> > Without looking much at the code, but following this list for half a
> > year, my gut feeling is that there is still too much major work
> > required on ath9k at this point for "fine tuning" such as signal
> > strength optimization to come up on the agenda in the near future..
> I disagree. The major work has been done and hardware support has mostly
> caught up with Atheros' own codebase. What's missing is debugging work
> and fine tuning - and other than that, maybe a few things that aren't
> very hardware specific.
> 
> - Felix
> _______________________________________________
> ath9k-devel mailing list
> ath9k-devel at lists.ath9k.org
> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
> 

[ath9k-devel] Question about ath9k signal strength (AP mode)

[Benoit Papillault]

Le 10/07/2010 18:12, Lars Hardy a ?crit :
> It seems that there is a difference in signal strength between the ath9k
> (latest openwrt-trunk with compat-wireless 08.07.2010) driver and the
> latest dd-wrt (madwifi N?) driver.
> The dd-wrt driver has better signal strength and performance when a STA
> is some distance from the AP. (like 1 floor/2 rooms between)
> I have tried with 2 different Atheros chipset in AP mode, the AR5416 and
> AR9280. dd-wrt gives a stronger signal with both chipsets compared with
> openwrt.
>
> I know the ath9k is under development, so my question is if this is
> known by the development team and therefor will be worked on?
>
> Lars

[ignoring other not very informative post...]

Did you measure the signal strength on the STA side? What tool did you 
use? iw / iwconfig ?

Is there a way you can bisect between OpenWrt and dd-wrt? I have AR5416, 
AR9160 and AR9280 myself, so I can test if you tell me what version you 
used.

Regards,
Benoit

[ath9k-devel] Question about ath9k signal strength (AP mode)

[Lars Hardy]

Benoit Papillault skrev:
> Le 10/07/2010 18:12, Lars Hardy a ?crit :
>> It seems that there is a difference in signal strength between the ath9k
>> (latest openwrt-trunk with compat-wireless 08.07.2010) driver and the
>> latest dd-wrt (madwifi N?) driver.
>> The dd-wrt driver has better signal strength and performance when a STA
>> is some distance from the AP. (like 1 floor/2 rooms between)
>> I have tried with 2 different Atheros chipset in AP mode, the AR5416 and
>> AR9280. dd-wrt gives a stronger signal with both chipsets compared with
>> openwrt.
>>
>> I know the ath9k is under development, so my question is if this is
>> known by the development team and therefor will be worked on?
>>
>> Lars
>
> [ignoring other not very informative post...]
>
> Did you measure the signal strength on the STA side? What tool did you 
> use? iw / iwconfig ?
>
This was from the client side, with an AR5416 MAC/BB Rev:2 AR5133 in 
HT40 - as AP.
Tested a bit further and It seems that the signal is only weaker when 
ath9k use the 5 Ghz band. (pretty much the same as dd-wrt in 2.4 Ghz)
iw reported: Router with dd-wrt: signal: -72.00 dBm  Router with 
OpenWrt: signal: -80.00 dBm
(from this distance the STA could not connect when using OpenWrt but 
connected when dd-wrt was on the router)
STA is an AR5418 AR5133 card, using the same/latest compat-wireless as 
OpenWrt.

> Is there a way you can bisect between OpenWrt and dd-wrt? I have 
> AR5416, AR9160 and AR9280 myself, so I can test if you tell me what 
> version you used.
>
I'm sorry but I have no idea how to do that. I only used the latest 
dd-wrt binary: 
ftp://ftp.dd-wrt.com/others/eko/BrainSlayer-V24-preSP2/06-30-10-r14684/
> Regards,
> Benoit

[ath9k-devel] Question about ath9k signal strength (AP mode)

[Peter Stuge]

Benoit Papillault wrote:
> > I know the ath9k is under development, so my question is if this is
> > known by the development team and therefor will be worked on?
> 
> [ignoring other not very informative post...]

Maybe not informative to you, but I wanted to prepare Lars a little
for what to expect from this mailing list. I know that Felix and
others are doing great work on ath9k and I'm very happy that he is
more optimistic about the driver than I :) but in my experience,
ath9k just does not give a very robust wifi experience and
development speed is moderate, supposedly because of limited
resources.

Note that I am not really complaining much about that, only
observing and informing.


> Is there a way you can bisect between OpenWrt and dd-wrt? I have
> AR5416, AR9160 and AR9280 myself, so I can test if you tell me what
> version you used.

dd-wrt was using madwifi according to the original post. No go for
the bisect.


//Peter

[ath9k-devel] Question about ath9k signal strength (AP mode)

[Benoit Papillault]

Le 11/07/2010 13:51, Peter Stuge a ?crit :
> Benoit Papillault wrote:
>>> I know the ath9k is under development, so my question is if this is
>>> known by the development team and therefor will be worked on?
>>
>> [ignoring other not very informative post...]
>
> Maybe not informative to you, but I wanted to prepare Lars a little
> for what to expect from this mailing list. I know that Felix and
> others are doing great work on ath9k and I'm very happy that he is
> more optimistic about the driver than I :) but in my experience,
> ath9k just does not give a very robust wifi experience and
> development speed is moderate, supposedly because of limited
> resources.
>
> Note that I am not really complaining much about that, only
> observing and informing.
>
>
>> Is there a way you can bisect between OpenWrt and dd-wrt? I have
>> AR5416, AR9160 and AR9280 myself, so I can test if you tell me what
>> version you used.
>
> dd-wrt was using madwifi according to the original post. No go for
> the bisect.

At least, we have a known working binary and a known not-working binary, 
so we can study the difference. If there are using different drivers, so 
we could try to narrow the difference between register settings.

It seems that every other driver is not handling TX power settings 
properly and I think this is quite a complex subject. For instance, I've 
seen that most hardware are using calibration data for TX power and 
there are calibration data for each TX rate as well, and such 
calibration might depends upon temperature. Atheros hardware also has 
per descriptor specific registers regarding TX power which can be 
enabled or disabled if I remember correctly.

Moreover, datasheet show different maximum TX power for each rate, which 
sounds a bit strange. If someone could shed some light on all this 
complex stuff?

Regards,
Benoit

[ath9k-devel] ath9k in Android

[Jónatan Muñoz]

Hi all,

Could anyone tell me if ath9k is present in the kernel of the new version of
Android OS 2.2?

Any help would be highly appreciated.

Regards, 

John 

[ath9k-devel] ath9k in Android

[Luis R. Rodriguez]

On Thu, Jul 15, 2010 at 10:00 AM, J?natan Mu?oz
<jonatan.munoz@alu.umh.es> wrote:
> Hi all,
>
> Could anyone tell me if ath9k is present in the kernel of the new version of
> Android OS 2.2?

It certainly is part of the Linux kernel they use. ath9k is upstream
and Android 2.2 (aka Froyo) comes with the 2.6.32 kernel. Whether they
enabled it is another question, look for the kernel configuration and
see if CONFIG_ATH9K=m or CONFIG_ATH9K=y

  Luis

[ath9k-devel] ath9k in Android

[Jónatan Muñoz]

It is set as CONFIG_ATH9K=m

Thanks for reply

John

-----Mensaje original-----
De: Luis R. Rodriguez [mailto:mcgrof at gmail.com] 
Enviado el: jueves, 15 de julio de 2010 19:54
Para: J?natan Mu?oz
CC: ath9k-devel at lists.ath9k.org
Asunto: Re: [ath9k-devel] ath9k in Android

On Thu, Jul 15, 2010 at 10:00 AM, J?natan Mu?oz
<jonatan.munoz@alu.umh.es> wrote:
> Hi all,
>
> Could anyone tell me if ath9k is present in the kernel of the new version of
> Android OS 2.2?

It certainly is part of the Linux kernel they use. ath9k is upstream
and Android 2.2 (aka Froyo) comes with the 2.6.32 kernel. Whether they
enabled it is another question, look for the kernel configuration and
see if CONFIG_ATH9K=m or CONFIG_ATH9K=y

  Luis

[ath9k-devel] ath9k in Android

[Luis R. Rodriguez]

On Wed, Jul 21, 2010 at 2:55 AM, J?natan Mu?oz <jonatan.munoz@alu.umh.es> wrote:
> It is set as CONFIG_ATH9K=m

I should note if you want to upgrade the 802.11 module for a specific
Android release you can just compile compat-wireless. For that you can
either use a bleeding edge linux-next.git based release [1] or wait
for a stable release [2].

[1] http://wireless.kernel.org/en/users/Download/
[2] http://wireless.kernel.org/en/users/Download/stable/

  Luis

[ath9k-devel] ath9k in Android

[Jónatan Muñoz]

Thanks Luis,

Yes, I want to use an USB wireless interface (AR9170) connect to a mobile phone, which runs Android 2.2. The kernel version is 2.6.32 so I assumed that ath9k is on the kernel. 
Android runs in a embedded device with ARM architecture, I think to compile only specific modules. Is it enough to upgrade ath9k.ko? or is it also necessary to upgrade ar9170usb.ko or more modules?

Thanks for your help.

John

-----Mensaje original-----
De: Luis R. Rodriguez [mailto:mcgrof at gmail.com] 
Enviado el: mi?rcoles, 21 de julio de 2010 18:32
Para: J?natan Mu?oz
CC: ath9k-devel at lists.ath9k.org
Asunto: Re: [ath9k-devel] ath9k in Android

On Wed, Jul 21, 2010 at 2:55 AM, J?natan Mu?oz <jonatan.munoz@alu.umh.es> wrote:
> It is set as CONFIG_ATH9K=m

I should note if you want to upgrade the 802.11 module for a specific
Android release you can just compile compat-wireless. For that you can
either use a bleeding edge linux-next.git based release [1] or wait
for a stable release [2].

[1] http://wireless.kernel.org/en/users/Download/
[2] http://wireless.kernel.org/en/users/Download/stable/

  Luis

[ath9k-devel] ath9k in Android

[Luis R. Rodriguez]

On Wed, Jul 21, 2010 at 9:55 AM, J?natan Mu?oz <jonatan.munoz@alu.umh.es> wrote:
> Thanks Luis,
>
> Yes, I want to use an USB wireless interface (AR9170) connect to a mobile phone, which runs Android 2.2. The kernel version is 2.6.32 so I assumed that ath9k is on the kernel.
> Android runs in a embedded device with ARM architecture, I think to compile only specific modules. Is it enough to upgrade ath9k.ko? or is it also necessary to upgrade ar9170usb.ko or more modules?
>

You can use ./scripts/driver-select ar9170 to enable only compilation
of what ar9170 needs.

  Luis