Hi Johnny,

quoted text
> I've noticed during some recent testing that the scan results
> delivered from mad-wifi are fairly non-deterministic. I'm using the
> latest stable release, but I've tested
> other versions with similar results.

I have noticed the same thing. Although Madwifi interfaces seem to do a 
large active scan when they first come up, I think there are too many 
channels, 11 in 2.4 GHz and ~19-24 in the 5 GHz band. So I think 
sometimes APs are missed.

quoted text
> Essentially, if you perform a few scans in a row you will get
> different results, with some networks dropping and coming back on the
> list.> Also, if you perform a scan, unplug an AP, and do another scan, the
> unplugged AP is likely to still show up. It also seems  fairly common
> to get no scan results back, even though performing a scan a second later
> will return 5 or 6 AP's.  I understand out-of-range or turned off AP's
> lingering in
> the scan cache, but I'm not sure what would cause in-range ones to
> disappear and reappear. I also don't understand why no results are
> returned occasionally.

I have noticed these things as well.

quoted text
> While this isn't much of a problem in static environments, I'd like to
> be able to get an accurate view of the surrounding networks without
> resorting to
> channel hopping and parsing packets pulled off a monitor mode
> interface. Bringing the interface down and up between scans seems to
> improve reliability overall (it definetly
> seems to decrease the no-scan-results case) but it still problematic.
>
> Does anyone know why this might be or what I can do to force more
> aggressive scanning?

I usually just do a "ifconfig ath0 down" then "ifconfig ath0 up". It is 
far from perfect but it does seem to force the driver to rescan.

quoted text
> I haven't verified this behavior in ath5k yet.
>
> Thanks for an suggestions. It's possible this is a problem in iwtools
> and not the driver itself. If nobody has any suggestions I'll start
> calling the ioctl's myself
> to get a better handle on debugging.

I am not one of the developers of madwifi or wireless-toosls but, I 
have looked through some of the code in iwtools. You probably want to 
look at iwlist.c and iwlib.h. When sifting through this code it seems 
that although wireless-tools does try to control scanning, for many 
drivers, madwifi included, iwlist simply grabs the scanning cache from 
the driver.

I am tackling some similar issues at the moment so if you have any joy, 
let me know :)

Have a good day

Dave

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Madwifi-devel mailing list
Madwifi-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/madwifi-devel