question about packet time stamping

[Jouni Malinen]

On Thu, Oct 30, 2003 at 04:21:51PM -0800, Vlasios Tsiatsis wrote:

> I noticed that the driver maintains information about
> the received packet
> time (field "time" in struct hfa384x_rx_frame in
> hostap_wlan.h) in respect
> with the network interface.
> Is this timestamp converted somewhere in the code to
> refer to the host
> computer local time ?

No, this field is only passed as is to a wireless sniffer when the
driver is in monitor mode with full sniffer headers being used.

> Also is there a way to get the timestamp of the packet
> when it actually
> leaves the interface and goes into the air ? i.e.
> something like a TX
> timestamp ? Or otherwise does the prism chip generates
> a TX END interrupt
> as a UART does ?

Prism2 f/w does not provide the exact time of the packet transmission,
but you can request an interrupt for both TX success and failure. If you
want to use this, please note that Host AP driver is by default only
requesting interrupt for failed TX frames and very limited set of
(non-data) frames get TX OK event.

> I am NOT interested in the time that the packet is
> passed from the driver
> to the network interface because it takes some time
> (due to medium access
> delays) for the packet to go into the air.

I don't think there is any guarantee in getting the TX event immediately
after the frame is sent out, but I would assume it would at least be
generated quite close to this. Anyway, you would not get microsecond
accuracy for this in the same way as for RX packets.

-- 
Jouni Malinen                                            PGP id EFC895FA