[prev in list] [next in list] [prev in thread] [next in thread] 

List:       orinoco-devel
Subject:    Re: [Orinoco-devel] Re: Orinoco + NetworkManager = No signal
From:       Pavel Roskin <proski () gnu ! org>
Date:       2006-05-05 3:38:45
Message-ID: 1146800325.18716.59.camel () dv
[Download RAW message or body]

Hello, Mark!

On Thu, 2006-05-04 at 15:31 +0100, Mark Hellman wrote:
> Pavel Roskin wrote:
> 
> > I'm not going to check userspace software other than wireless tools for
> > compatibility with the driver.  If wireless tools can work with the
> > driver, then other software should also be able to.  If you think
> > otherwise, I'd like to hear detailed arguments.
> 
> Hello Pavel,
> 
> Please mind that I am not involved in NetworkManager development, nor I am
> aware of its internals. I just wanted to let orinoco developers know about
> a problem that afects orinoco cards with NetworkManager. It doesn't affect
> the Intel or other native drivers.

That's an important piece of information.  I believe NetworkManager
wants to to get quality for the scan data.

The Orinoco driver doesn't have that data from the chipset, but it has
the signal and noise levels.  The driver sets IW_QUAL_QUAL_INVALID to
indicate that the quality data is not provided.

The Intel's drivers use the common wireless code in
net/ieee80211/ieee80211_wx.c, which has a bizarre looking formula for
calculating the quality (look for number 62).  I'm afraid it's Intel
specific.

Quick search on Google found this article:
http://www.ces.clemson.edu/linux/signal_quality.shtml

"This model is based on a least squares fit of four data points
collected using the Windows XP driver".

So, the driver is supposed to use an algebraic expression to approximate
the formula (probably something very simple-minded) used in Windows XP!

I'm not very enthusiastic about having such monstrosities in the kernel
code.  The driver becomes responsible for the correct number of bars in
the user interface.  Besides, the driver doesn't really know what is the
best signal possible for the card.  There are subtle hardware details
that the driver cannot check.  -50 dBm may be very good for one card and
pretty bad for another.  I think the noise data is often unreliable, and
so is the SNR.

Cannot the userspace program calculate the quality data based on the dBm
values the driver provides?  By all means, emulate Windows XP or
whatever, approximate it with any functions, but please don't expect the
driver to do it for you.  It gives as much as it can already.

-- 
Regards,
Pavel Roskin



-------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
Orinoco-devel mailing list
Orinoco-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/orinoco-devel
[prev in list] [next in list] [prev in thread] [next in thread]