[ath9k-devel] Hardware Timestamp Question

[ath9k-devel] Hardware Timestamp Question

[Diego Bernardes]

Hi,

I have one question about hardware timestamp in Atheros chipsets.

[ath9k-devel] Hardware Timestamp Question

[Sergey Ryazanov]

Hi Diego,

2014-02-18 18:45 GMT+04:00 Diego Bernardes <di3go.bernardes@gmail.com>:
> The question is, how accurated this timestamp is? The precision is in
> nanoseconds, but what accuracy i can expect?
>
The source of time information for timestamps is TSF timer which is
64-bit timer with microsecond precision and accuracy not worse than
0.01% according to IEEE 802.11-2012 section 10.1.3.7

Seems that all atheros chipsets support rx timestamp generation in
hardware and driver responsible for processing of such information, in
USB-dongle case there are also level of firmware which should carry
timestamp to the driver.

> The other question is, what is the difference between the ath9k_htc and
> carl9170 drivers? Both support the AR9287 (from the debian wiki).
>
I don't know the actual differences, try to find answer in
linux-wireless wiki at [1].

1. http://wireless.kernel.org/en/users/Drivers

--
BR,
Sergey

[ath9k-devel] Hardware Timestamp Question

[Adrian Chadd]

Hi,

All of the chips do a hardware RX timestamp of some sort.

The default timestamp is the TSF timestamp. It's based on the TSF,
which gets nudged by the AP.

The AR9287 is the first chip with a second TSF counter, designed to be
free-running and not be locked into the AP broadcast TSF. Ie, if you
wanted to be an AP and a STA, you'd have one TSF for the AP and one
TSF for the STA association. But I don't know how it all works in too
much depth and I'm not sure how the relevant TSF gets stamped to which
RX frame.

There's also the Azimuth mode timestamp that is in a debug mode
somewhere; I don't know (yet!) much about how that works.


-a


On 18 February 2014 06:45, Diego Bernardes <di3go.bernardes@gmail.com> wrote:
> Hi,
>
> I have one question about hardware timestamp in Atheros chipsets.
> From a post in Wireshark community, the chipset AR9287 can do hardware
> timestamp.
>
> The question is, how accurated this timestamp is? The precision is in
> nanoseconds, but what accuracy i can expect?
>
> My main need is the implement a good precision TDOA algorithm.
>
> I don't know jack about driver development, but is something that interests
> me. If the accuracy is low, can i improve it in driver or firmware?
> If its possible, i can even trade reception packet speed for a better
> accuracy in timestamp.
>
> The other question is, what is the difference between the ath9k_htc and
> carl9170 drivers? Both support the AR9287 (from the debian wiki).
>
> Thanks for all
>
> Links:
>
> Wireshark community:
> http://ask.wireshark.org/questions/28683/recommended-wireless-adapter-usb-with-linux-wireshark-that-reports-mactime-in-radiotap-header
>
> Debian Wiki ath9k_htc: https://wiki.debian.org/ath9k_htc
>
> Debian Wiki carl9170: https://wiki.debian.org/carl9170
>
> _______________________________________________
> ath9k-devel mailing list
> ath9k-devel at lists.ath9k.org
> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
>

[ath9k-devel] Hardware Timestamp Question

[Sujith Manoharan]

Adrian Chadd wrote:
> The AR9287 is the first chip with a second TSF counter, designed to be
> free-running and not be locked into the AP broadcast TSF. Ie, if you
> wanted to be an AP and a STA, you'd have one TSF for the AP and one
> TSF for the STA association. But I don't know how it all works in too
> much depth and I'm not sure how the relevant TSF gets stamped to which
> RX frame.

Are you sure ? I think AR9485 is the first chip with 2 TSF counters.

Sujith

[ath9k-devel] Hardware Timestamp Question

[Adrian Chadd]

Look at TSF2 in the source.

I have the "how does tsf2 in Kiwi (AR9287) work" documentation here somewhere...


-a


On 18 February 2014 18:25, Sujith Manoharan <sujith@msujith.org> wrote:
> Adrian Chadd wrote:
>> The AR9287 is the first chip with a second TSF counter, designed to be
>> free-running and not be locked into the AP broadcast TSF. Ie, if you
>> wanted to be an AP and a STA, you'd have one TSF for the AP and one
>> TSF for the STA association. But I don't know how it all works in too
>> much depth and I'm not sure how the relevant TSF gets stamped to which
>> RX frame.
>
> Are you sure ? I think AR9485 is the first chip with 2 TSF counters.
>
> Sujith

[ath9k-devel] Hardware Timestamp Question

[Oleksij Rempel]

Am 18.02.2014 15:45, schrieb Diego Bernardes:
> Hi,
> 
> I have one question about hardware timestamp in Atheros chipsets.
> From a post in Wireshark community, the chipset AR9287 can do hardware
> timestamp.
> 
> The question is, how accurated this timestamp is? The precision is in
> nanoseconds, but what accuracy i can expect?
> 
> My main need is the implement a good precision TDOA algorithm. 
> 
> I don't know jack about driver development, but is something
> that interests me. If the accuracy is low, can i improve it in driver or
> firmware? 
> If its possible, i can even trade reception packet speed for a better
> accuracy in timestamp.
> 
> The other question is, what is the difference between the ath9k_htc and
> carl9170 drivers? Both support the AR9287 (from the debian wiki).

carl9170 firmware works on ar9170 CPU and so far i know there is no
ar9170+AR9287 devices.

Major difference between this two drivers: ath9k_htc shares bigger part
of RX and some other hw related code with ath9k. I hope it will be more
then this :)
carl9170 firmware was rewritten from scratch.
ath9k_htc firmware is not rewritten :)

> Thanks for all
> 
> Links:
> 
> Wireshark community:
> http://ask.wireshark.org/questions/28683/recommended-wireless-adapter-usb-with-linux-wireshark-that-reports-mactime-in-radiotap-header
> 
> Debian Wiki ath9k_htc: https://wiki.debian.org/ath9k_htc
> 
> Debian Wiki carl9170: https://wiki.debian.org/carl9170
> 
> 
> _______________________________________________
> ath9k-devel mailing list
> ath9k-devel at lists.ath9k.org
> https://lists.ath9k.org/mailman/listinfo/ath9k-devel
> 


-- 
Regards,
Oleksij

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 278 bytes
Desc: OpenPGP digital signature
Url : http://lists.ath9k.org/pipermail/ath9k-devel/attachments/20140219/82b79571/attachment.pgp