Re: HEADSUP: new ath+hal

Yamamoto Shigeru wrote:
> Hi, sam,
> 
> I install 20060130 current to same note PC to compare 'current 20060130' and
> 'current 20060222'.
> After installing, I watch control message with
>  'tcpdump -n -i ath0 -y IEEE802_11'.
> 
> When using '20060130 current', I can watch "Probe request" from my note PC.
> But I can't watch "Probe request" when using '20060222 current'.
> 
> I add 'printf()' to codes and I check when 'IEEE80211_CHAN_PASSIVE' is set.
> I found ath_getchannels() at _at_src/sys/dev/ath/if_ath.c sets
> 'IEEE80211_CHAN_PASSIVE', because 'CHANNEL_PASSIVE' in channelFlags is set.

FWIW if you do:

ifconfig ath0 list chan

you will see which channels are marked for passive scan (they have a '*' 
appended--though it appears this is not explained in the man page). 
Also doing

wlandebug scan

should show information like this.  Otherwise adding debug; e.g.

wlandebug scan+debug

will show additional info including when probe request frames are sent.

It should not be necessary to add printf's to the kernel to diagnose 
stuff like this (at least that is the intent).

> 
> I think ath driver send "Probe request" if ignore 'CHANNEL_PASSIVE' in
> channelFlags.
> So I add a code to ignore 'CHANNEL_PASSIVE' and to unset
> 'IEEE80211_CHAN_PASSIVE'.
> I test this test code at "broadcast SSID in beacon", and my note PC can
> associate an AP.
> 
> It seems me current ath hal returns 'CHANNEL_PASSIVE' for all channels.
> In _at_src/sys/contrib/dev/ath/ah.h, comment for 'CHANNEL_PASSIVE' is "Only
> passive scan allowed in the channel".
> Does it mean my note PC can't send a probe message?

I answered this in previous mail but to add to it; you should be aware 
that regulatory agencies are very sticky about code that violates 
passive scan requirements.  What I want to do is make sure the new hal 
is doing the right thing for the regdomain you are using (hence my 
request for the info).  The rules in Japan have changed significantly 
since the 0.9.14.9 hal was released so it is not surprising you are 
seeing different behaviour with the new code.

	Sam