Date: Wed, 06 Dec 2000 09:03:01 -0000 (GMT) From: Duncan Barclay <dmlb@dmlb.org> To: Greg Lehey <grog@lemis.com> Cc: Chris Yeoh <cyeoh@linuxcare.com.au>, mobile@FreeBSD.ORG, Wesley Morgan <morganw@chemicals.tacorp.com>, Warner Losh <imp@village.org> Subject: Re: ray committed Message-ID: <XFMail.001206090301.dmlb@computer.my.domain> In-Reply-To: <20001206154244.J20481@wantadilla.lemis.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi Warner,
On 06-Dec-00 Greg Lehey wrote:
> On Tuesday, 5 December 2000 at 22:09:18 -0700, Warner Losh wrote:
>> In message <20001206152332.A21187@wantadilla.lemis.com> Greg Lehey writes:
>>> I discussed this with Chris Yeoh at Linuxcare last week. He has no
>>> problems with packet loss; in fact, he considers them better in that
>>> respect than the Orinoco cards.
>>
>> Interesting... Is that with the FreeBSd drivers, or the Linux ones?
>> I'm definitely seeing major problems with the two cards here and none
>> with the ZoomAir + Orinoco Gold. I wonder what is different between
>> him and I.
Rememeber that the Orinoco and Webgear cards use very different RF techniques
that give rise to different range/interference rejection etc. Multipath
is handled very differently, do you have a lot of other buildings within say
100m?
Also, the Webgear cards seems to be not very well made. I've had a lot
of reports of "bad" cards (two myself) out of the box - i.e. cards that
are deaf.
> These are the Linux drivers, so that's an obvious difference. He also
> reports that the current Linux drivers are broken, so he's using an
> older version.
We run a number of cards at home now, and yes there is some packet loss but
not much, I just tried 12MB of backup from a Win98 box with no packets lost. The
other machine is about 50ft away, on another floor in my house.
Try playing with the RTS_THRESH and Fragmentation parameters. These will
both chop up the ethernet packets a bit.
What does
# raycontrol -i ray0 -o
report for the clear channel noise level? I get 55 (not sure if thats hex).
Using the -C option will print out the signal level and antenna cache.
# raycontrol -i ray0 -C
Slot 0: 00:00:8f:48:e4:04 7c,7c,75,71,6a,82,7d,78 0000000000000000
...
The hex digits with commas are the signal level. Higher is better. The binary
string is what antenna was picked. It should be reasonably constant.
I've not seen any evidence that packet loss is the driver fault (well I did
write it!), what might be happening is poor parameter choices for the
stuff raycontrol dumps out - the manuals were unclear and the different
exisiting drivers (Linux/NetBSD) were all inconsistent. If you look in the
header files, you'll see what choices I had. The entries marked Symbionics were
from people at work that develop 802.11 protocol stacks/hardware and should be
the "best" from understanding the standard - this is not the same as "best" for
a particular installation.
Duncan
---
________________________________________________________________________
Duncan Barclay | God smiles upon the little children,
dmlb@dmlb.org | the alcoholics, and the permanently stoned.
________________________________________________________________________
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-mobile" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?XFMail.001206090301.dmlb>