Scan wifi card on linux when connected-

[hong zhang]

For VOIP, maximum end to end delay should be less than
150-200ms; 150ms is better. This, to guarantee the
good quality of the sound to be transmitted. But
Tmobile assigns 60ms in roaming part and rest of 100ms
will be saved for other area.

Maybe progressive scan should be introduced in dadwifi
driver some time.

--- Dan Williams <dcbw at redhat.com> wrote:

> On Wed, 2007-05-30 at 13:55 -0700, hong zhang wrote:
> > Your suggestion should be called progressive scan
> not
> > normal scan. Some driver supports it. But I do not
> > think madwifi supports progressive scan yet. For
> QoS
> > feature, voice data interval is not more than
> 60ms.
> > Otherwise, voice will be paused and quality is
> bad. To
> > support WMM, client needs progressive scan to find
> APs
> > near it so that it can roam if need. But normal
> scan
> > would not work for QoS when connected. Timing is
> also
> > an issue here too. 
> > 
> > Should AP know when client is scanning for normal
> > scan?
> > 
> > Madwifif provides background scan. I believe it
> scans
> > when client connected.
> > 
> > There are 3 kinds of scan. Normal scan, background
> > scan and progressive scan. Madwifi currently
> supports
> > first two.
> 
> That's in the driver.  iwlist and userspace tools
> don't care what type
> of scan the driver does (except for active/passive
> scan) and probably
> shouldn't have to care.
> 
> But user tools certainly expect to be able to get
> full scans whenever
> they request a scan, and the driver needs to provide
> that.  Doesn't
> matter whether the driver does a "normal" scan,
> background scan, or
> progressive scan, whatever.  The driver should be
> smart enough that it
> can provide scan results on request, and update
> those scan results in a
> manner which is conducive to whatever userspace is
> doing at the time as
> well.
> 
> For example, if you're doing a VOIP call at the
> time, and some other
> userspace process requests a scan, then the driver
> might be able to scan
> one or two channels then jump back to the current
> channel for another
> dwell time, then one or two others, etc.  The ITU
> recommends that
> conversational voice applications have a maximum
> delay of about 250ms,
> not 60ms, so a dwell time of 100 - 200ms is quite
> within that range.
> 
> Dan
> 
> > 
> > --- Dan Williams <dcbw at redhat.com> wrote:
> > 
> > > On Wed, 2007-05-30 at 12:44 -0700, hong zhang
> wrote:
> > > > when driver scans, it will scan all channels
> as
> > > > default. Driver supports about 32 channels on
> > > a/b/g.
> > > > Default scan time on each channel is 100ms. It
> > > takes
> > > > at least 3 seconds to complete scan job. If
> client
> > > is
> > > > associated with AP, for example at channel 6.
> If
> > > scan
> > > > is allowed, driver will leave channel 6 when
> it
> > > scan
> > > > on other channels. Unless AP konws driver is
> > > canning
> > > > and AP buffers packets for more than 3
> seconds.
> > > > Otherwise, packets will be lost.
> > > > 
> > > > So I think normal scan is not allowed when
> > > connected.
> > > 
> > > So that's you answer then, as to why it's not
> > > returning scan results
> > > when it's connected.
> > > 
> > > If the driver is smart, it will send the AP a
> power
> > > save enter frame, do
> > > the scanning, then tell the AP it's woken up
> again,
> > > then get all the
> > > buffered packets from the AP.
> > > 
> > > Or, the driver could scan 2 or 3 channels, then
> hop
> > > back to the
> > > associated channel for a small amount of time to
> > > talk to the AP, then do
> > > the next set of channels.
> > > 
> > > The driver should be keeping a list of scan
> results
> > > from previous scans
> > > to return when iwlist gets called, and aging
> those
> > > scan results to
> > > delete each BSS when a certain amount of time
> has
> > > elapsed since the last
> > > beacon has been seen from that BSS.
> > > 
> > > In any case, if the driver doesn't allow scans
> while
> > > connected, that
> > > driver is dumb and broken and should be made
> > > smarter.
> > > 
> > > Dan
> > > 
> > > > --- Dan Williams <dcbw at redhat.com> wrote:
> > > > 
> > > > > On Wed, 2007-05-30 at 10:31 -0700, hong
> zhang
> > > wrote:
> > > > > > Magda,
> > > > > > 
> > > > > > When client is already associated with AP,
> > > iwlist
> > > > > will
> > > > > > not do scan (normal scan). You can
> > > disaasociate
> > > > > with
> > > > > > the AP and run iwlist again. See any
> > > difference.
> > > > > To 
> > > > > > verify what I said. You can put printk()
> on
> > > > > > ieee80211_notify_scan_done(struct
> ieee80211vap
> > > > > *vap)
> > > > > > to prove scan is done when iwlist is
> executed.
> > > > > 
> > > > > Sounds like a driver bug; if you look in the
> > > source
> > > > > for the SIOCSIWSCAN
> > > > > handler, does it dis-allow a scan when
> > > connected?
> > > > > 
> > > > > Dan
> > > > > 
> > > > > > 
> > > > > > ----henry
> > > > > > --- magda chelly <sandra26899 at yahoo.fr>
> wrote:
> > > > > > 
> > > > > > > Hello,
> > > > > > > 
> > > > > > > I'm trying to scan a wifi card on debian
> > > > > > > distribution,
> > > > > > > (armel:nokia n800), but with iwlist
> scan, it
> > > > > doesn't
> > > > > > > show me all the networks, the connected
> one,
> > > > > does
> > > > > > > not
> > > > > > > appear and others are missing.
> > > > > > > Can someone help me please.
> > > > > > > 
> > > > > > > Thanks,
> > > > > > > Magda,
> > > > > > > 
> > > > > > > 
> > > > > > >      
> > > > > > >
> > > > > >
> > > > >
> > > >
> > >
> >
>
_____________________________________________________________________________
> > > > > > > 
> > > > > > > Ne gardez plus qu'une seule adresse mail
> !
> > > > > Copiez
> > > > > > > vos mails vers Yahoo! Mail 
> > > > > > >
> 
=== message truncated ===



       
____________________________________________________________________________________
Got a little couch potato? 
Check out fun summer activities for kids.
http://search.yahoo.com/search?fr=oni_on_mail&p=summer+activities+for+kids&cs=bz