Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 9 Aug 2011 02:26:54 +0800
From:      Adrian Chadd <adrian@freebsd.org>
To:        Lev Serebryakov <lev@serebryakov.spb.ru>
Cc:        freebsd-wireless@freebsd.org
Subject:   Re: Bad host-vased AP performace -- and qouple of questions about FreeBSD WiFi stack tuning
Message-ID:  <CAJ-VmonrRTRrvR9PWexxVzRzEGS9AW8%2BtZLZqUN1aZT9Uvoqqg@mail.gmail.com>
In-Reply-To: <1832976164.20110808222108@serebryakov.spb.ru>
References:  <1151152206.20110808220720@serebryakov.spb.ru> <CAJ-VmomMh4uvyJGHHkGsi5_C4p3ZJjRwadmstRVe4BjY76QzoA@mail.gmail.com> <1832976164.20110808222108@serebryakov.spb.ru>

next in thread | previous in thread | raw e-mail | index | archive | help
2011/8/9 Lev Serebryakov <lev@serebryakov.spb.ru>:
> Hello, Adrian.
> You wrote 8 =C1=D7=C7=D5=D3=D4=C1 2011 =C7., 22:11:03:
>
>> Can you verify when "bad" and "good" performance occured? Can you
>> narrow it down to FreeBSD versions in particular?

> =9A I'm afraid, that will be difficult. I could try 8.2-RELEASE and
> =9Aeven 8.0-RELEASE (but I'll need to build NanoBSD images for them).
> =9AI'm sure, that 7-STABLE gives me 3.2MiB/s regularly, but noteboks
> =9Awere changed from these times. I'll try to build 7-STABLE NanoBSD
> =9Aimage, but it will need some more effort, as config files changes from
> =9Athese times :)

Hi,

That's definitely required knowledge. 8.0-RELEASE is a good starting point.
I'm afraid it's going to be difficult for me to narrow down what's
changed without you doing a bit of legwork to find which kernel
version(s) introduced the performance regression.
Once we narrow that down, I'll fix it. :)

Please keep everything else the same though, including the laptop. :)

>> 2) not sure, it's late and I'm not knee-deep in the beacon frame
>> format. But I bet it's ok?
> =9AIt is "not accurate enough" :) Many APs, seen in my apartments,
> =9Aannounce all speeds up to 54Mb :) But, of course, It is practically Ok=
 till
> =9Aclients can connect on higher speeds.

Hm, that's odd. The 11b rates should show up  (1 -> 11), the rest are
11g rates. I can't think of why the AP wouldn't advertise all 11g
rates, but then, I haven't gone digging in that code lately.

>> 3) txpower is in dBm units, based on what the card is supposed to be
>> putting out of the antenna socket, before antennas
> =9AOk, so 30 means 1:1 (according to dBm definition) and means "full
> =9Apower".

Well, some cards "lie". You program in x dBm, you get x+constant dBm.
That's how so high powered cards work - you program in some lower
value that fits inside regulatory limits (say, 23dBm) and you then get
(say) a +6 dBm boost no matter what you program in. So a 23 dBm
txpower would equal 29dBm. 0dBm txpower would equal 6dBm. etc, etc.

The only real way to know is a spectrum analyser. :)


Adrian



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAJ-VmonrRTRrvR9PWexxVzRzEGS9AW8%2BtZLZqUN1aZT9Uvoqqg>