ATH9k_HTC mactime vs. system timestamp (timeval)

[Adrian Chadd]

Hm, does the firmware expose the TX completion timestamps at all? I
don't know. :(



-a

On 19 March 2015 at 04:54, Steger, Marco <Marco.Steger at v2c2.at> wrote:
> Hi,
>
> thank you for your help. This is very interesting and I will take this into account.
>
>  On additional question: I used the ath9k_htc_rx_status struct to get the receive time (I found in a header file that rx_status->mactime ts contains the time when the beginning of the data frame is received). The ath9k_htc_tx_ctl struct seems to be the equivalent on the sender side. I will try to use this to get the time when a frame is sent out.
> Can one of you give me a hint how to do that?!
>
> Best regards,
> Marco
>
>>-----Ursprüngliche Nachricht-----
>>Von: adrian.chadd at gmail.com [mailto:adrian.chadd at gmail.com] Im Auftrag von
>>Adrian Chadd
>>Gesendet: Mittwoch, 18. März 2015 17:05
>>An: Steger, Marco
>>Cc: ath9k_htc_fw at lists.infradead.org
>>Betreff: Re: ATH9k_HTC mactime vs. system timestamp (timeval)
>>
>>Hi,
>>
>>They're different because the MAC timestamp is based on whatever clock the
>>MAC is synched to.
>>
>>In hostap mode it'll be itself, and no, there's no requirement for it to be synched
>>against the system timestamp.
>>
>>In station mode it'll sync the MAC timestamp against the hostap timestamp.
>>
>>The 802.11 timestamps are not supposed to be synched against the system
>>timestamp - they're designed for timekeeping inside a BSS to synchronise things
>>like power save, etc.
>>
>>
>>
>>-a
>>
>>
>>On 18 March 2015 at 01:27, Steger, Marco <Marco.Steger at v2c2.at> wrote:
>>> Hi all,
>>>
>>> I have a small question about the mactime timestamp. I want to get an
>>accurate timestamp for the point in time when the packet is received (driver/hw).
>>>
>>> Therefore, I took a look on the ath9k_rx_prepare.c file. In the function
>>ath9k_rx_prepare the mactime is set:
>>> rx_status->mactime = be64_to_cpu(rxstatus->rs_tstamp);
>>>
>>> When I print this timestamp and compare it with the Debian/linux system
>>timestamp (timeval) I recognized that:
>>> * both are (should be) 64bit values
>>> * but they are different:
>>>  [ 2141.033905] TS: timeval-ts [1426665454085102] mactime-ts
>>> [688414257]  [ 2141.034290] TS: timeval-ts [1426665454085613]
>>> mactime-ts [688414675]  [ 2141.034697] TS: timeval-ts
>>> [1426665454086024] mactime-ts [688415074]  [ 2141.035071] TS:
>>> timeval-ts [1426665454086420] mactime-ts [688415456]  [ 2141.035442]
>>> TS: timeval-ts [1426665454086797] mactime-ts [688415827]  [
>>> 2141.035817] TS: timeval-ts [1426665454087170] mactime-ts [688416203]
>>> [ 2141.036173] TS: timeval-ts [1426665454087544] mactime-ts
>>> [688416569]  [ 2141.036566] TS: timeval-ts [1426665454087915]
>>> mactime-ts [688416953]  [ 2141.036943] TS: timeval-ts
>>> [1426665454088301] mactime-ts [688417342]  [ 2141.037345] TS:
>>> timeval-ts [1426665454088682] mactime-ts [688417745]
>>>
>>> So I think there must/should be an easy way to get a system ts out of the
>>mactime ts but I don't know how at the moment.
>>>
>>> Would be great if you can help me here!
>>>
>>> Best regards,
>>> Marco
>>>
>>>
>>> Dipl.-Ing. Marco Steger
>>> Junior Researcher
>>> Area Electrics/Electronics & Software (Area E)
>>>
>>> VIRTUAL VEHICLE
>>> Kompetenzzentrum - Das virtuelle Fahrzeug Forschungs-GmbH
>>>
>>> Inffeldgasse 21/A
>>> 8010  Graz  /  Austria
>>> Tel:        +43 316 873 9009
>>> E-Mail:     marco.steger at v2c2.at
>>> Web:        www.v2c2.at <http://www.v2c2.at/>
>>>
>>> Firmenname: Kompetenzzentrum - Das virtuelle Fahrzeug,
>>> Forschungsgesellschaft mbH
>>> Rechtsform: Gesellschaft mit beschränkter Haftung
>>> Firmensitz: A-8010 Graz, Inffeldgasse 21/A
>>> Firmenbuchnummer: 224755y
>>> Firmenbuchgericht: Landesgericht fürZRS Graz
>>>
>>>
>>>
>>> _______________________________________________
>>> ath9k_htc_fw mailing list
>>> ath9k_htc_fw at lists.infradead.org
>>> http://lists.infradead.org/mailman/listinfo/ath9k_htc_fw
> _______________________________________________
> ath9k_htc_fw mailing list
> ath9k_htc_fw at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/ath9k_htc_fw