Re: Question regarding Madwifi signal strength

["Darren Albers" <dalbers gmail com>]

On 6/8/07, Dan Williams <dcbw redhat com> wrote:
> On Sat, 2007-06-09 at 00:11 +1200, Simon Geard wrote:
> > On Fri, 2007-06-08 at 23:10 +1200, Simon Geard wrote:
> > > On Thu, 2007-06-07 at 16:25 -0400, Darren Albers wrote:
> > > > Do the above changes seem like they resolved these issues or maybe not
> > > > affect it at all?   I can dig out my atheros card (it's in the attic!)
> > > > and find out if nobody knows but I figured it would save me the hassle
> > > > of climbing up there if someone already has checked.
> > >
> > > I use an Atheros-based PCI card, so I'll try updating the drivers, see
> > > if the signal strength is still an issue. As of whatever version I'm
> > > running right now (SVN from about a month or two ago), NM reports about
> > > 15% while the Gnome network applet shows about 60%. What would you
> > > expect me to see on new drivers?
> >
> > Ok, running the latest madwifi driver from SVN now. Doesn't seem any
> > different from previous versions - I'm seeing roughly the same signals
> > in NM as before. Is the following output from iwlist of any interest to
> > you?
> >
> >     00:17:9A:12:31:72 : Quality=49/70  Signal level=-46 dBm  Noise level=-95 dBm
>
>                                   ^^^^ This looks wrong
>
> Quality is supposed to be a linear value between 0 and max_qual.qual
> inclusive.  49/70 = 70%; and with a signal of -46dBm and a noise of -95
> dBm, that seems bogus.  It should be higher than that.  -46 dBm is
> _really good_.
>
> >     00:14:7C:AD:39:92 : Quality=120/70  Signal level=-256 dBm  Noise level=-95 dBm
>
>                                   ^^^^ This is wrong
>
> A value of 120 is clearly outside the bounds of 0, 70.
>
> madwifi is free to use dBm or RSSI in the 'level' and 'noise' fields,
> which it apparently does just fine (except for the -256 dBm signal level
> in the above result).  But if the driver specifies quality, NM will use
> that first, which it should do.
>
> Dan

Thanks Dan!  That is too bad...  Though in retrospect I really want to
avoid any non-free components (Like the madwifi HAL since dadwifi
looks to be awhile off) so I should stick with the IPW3945 with the
new iwlwifi driver instead.