[PATCH] Setting quality in scan results

[Pavel Roskin]

On Wed, 28 Apr 2004, Jouni Malinen wrote:

> On Wed, Apr 21, 2004 at 12:57:46PM -0400, Pavel Roskin wrote:
>
> > Scan results show random "quality" because it's never set.  The firmware
> > only provides quality as part of HFA384X_RID_COMMSQUALITY request.  For
> > scan results, only signal and noise are provided.  I think "quality" is
> > not really applicable for a single probe response.
>
> Random?? It is cleared to zero in __prism2_translate_scan() (memset).

Indeed.  My error.  Still, 0 would be wrong.

> > However, iwlist requires "quality" is scan results, so we should calculate
> > it somehow.  I suggest to subtract noise from signal.  Since both are in
> > decibels, the result should have a monotone dependency on the signal to
> > noise ratio.
>
> Is this field documented somewhere? Why is it required? I don't see much
> point in reporting signal-noise separately when both of the values are
> available anyway..

"man iwconfig", from "Link quality": "This is an aggregate value, and
depends totally on the driver and hardware."

I'm discussing the issue with Jean in private.  My understanding is that
the "updated" field is supposed to be a bitmask that would indicate
whether quality, signal and noise are valid.  Unfortunately, it was never
documented.

As is stands now, "updated" is ignored and "qual" is used always.  It may
change soon.  You may want to wait until the issue is clarified.  But if
you apply my patch, please set "updated" to 7, not 0 as I wrote.  It may
be used in the future Wireless Tools.

-- 
Regards,
Pavel Roskin