iwl4965 vs b43 time drift?

[Johannes Berg <johannes@sipsolutions.net>]

[-- Attachment #1: Type: text/plain, Size: 1353 bytes --]

Hi,

I'm debugging a timing problem with b43, can you tell me exactly when
the timestamp iwl4965 shows in radiotap is taken? Is it the first bit of
the preamble?

When associated to my b43 AP, I'd expect the difference between the
beacon/proberesp timestamp and the MAC timestamp of the iwl4965 monitor
mode to be 192us for the MAC header and another 192us for the PLCP
preamble/PLCP header, a total of 384 us.

However, when I have just associated, the difference starts out at 383us
and decreases, down to ~370us and then varies between 370-380us, afaict
never reaching 384us.

The strange thing is that it doesn't seem to recover. No matter which of
the timers is actually wrong (I guess I could find out by comparing with
the system timestamp and hoping that the delay through mac80211 into
wireshark is constant) shouldn't it recover the time?

Another problem is that when I enable powersave on the iwl4965, I see no
beacons at all. And oddly, the difference between the timestamp in the
probe responses and the iwl4965 MAC timestamp starts varying wildly, up
to almost 5000us.

I guess that b43 has a timing problem otherwise we'd be seeing totally
bad behaviour of iwl4965 on stock APs. I'm just trying to understand the
problem better right now. Do you know when iwl4965 will resync the
timer?

Thanks,
johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 828 bytes --]