Hi Johannes,
On Fri, Jan 4, 2013 at 5:05 AM, Johannes Berg <johannes@xxxxxxxxxxxxxxxx> wrote:
> On Wed, 2013-01-02 at 14:55 -0800, Thomas Pedersen wrote:
>
>> @@ -359,6 +359,8 @@ struct mac80211_hwsim_data {
>> /* difference between this hw's clock and the real clock, in usecs */
>> s64 tsf_offset;
>> s64 bcn_delta;
>> + /* absolute beacon transmission time. Used to cover up "tx" delay. */
>> + u64 abs_bcn_tstamp;
>
> This can't work if you have multiple interfaces, as far as I can tell.
In practice, I think it'll be fine. abs_bcn_tstamp is only relevant
from the beacon timestamp to the rx mactime code, of which you have a
single (non-preemptible) path per hwsim hw. Per-vif beaconing (was) is
already not supported.
>> + /*
>> + * Account for delay from filling in the timestamp to mactime.
>> + */
>> + now = mac80211_hwsim_get_tsf_raw();
>> + if (ieee80211_is_beacon(hdr->frame_control) ||
>> + ieee80211_is_probe_resp(hdr->frame_control))
>> + now -= now - data->abs_bcn_tstamp;
>
> Uhh.
> now -= now - data->abs_bcn_tstamp;
> <=> now = now - now + data->abs_bcn_tstamp;
> <=> now = dat->abs_bcn_tstamp;
Oops. I guess that factors down to what I was really trying to do. No
wonder the test passed :P
> No? But it seems to me that you're trying to timestamp based on theT
> transmitter? This is weird anyway?
The global hwsim tsf is just ktime, so all interfaces have the same
reference. This might cause the beacon mactime to show as earlier than
just-received non-beacons, but it's only on the order of ~10 us, and
I'd rather have that uncertainty there than Toffset.
Will you take this patch after a little refactoring?
Thomas
--
To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
