[PATCH] mac80211: report operating frequency rather than current

[PATCH] mac80211: report operating frequency rather than current

[Johannes Berg]

It's not very helpful to see, in iwconfig, the current frequency
the card is tuned to if that frequency is currently somewhere
across the board because we're scanning. Since we keep track of
the frequency the user wants, display that instead.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
---
 net/mac80211/wext.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

--- wireless-testing.orig/net/mac80211/wext.c	2009-05-04 17:41:53.000000000 +0200
+++ wireless-testing/net/mac80211/wext.c	2009-05-04 17:42:02.000000000 +0200
@@ -185,7 +185,7 @@ static int ieee80211_ioctl_giwfreq(struc
 	if (sdata->vif.type == NL80211_IFTYPE_ADHOC)
 		return cfg80211_ibss_wext_giwfreq(dev, info, freq, extra);
 
-	freq->m = local->hw.conf.channel->center_freq;
+	freq->m = local->oper_channel->center_freq;
 	freq->e = 6;
 
 	return 0;

Re: [PATCH] mac80211: report operating frequency rather than current

[Richard Farina]

Johannes Berg wrote:
> It's not very helpful to see, in iwconfig, the current frequency
> the card is tuned to if that frequency is currently somewhere
> across the board because we're scanning. Since we keep track of
> the frequency the user wants, display that instead.
>
> Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
> ---
>  net/mac80211/wext.c |    2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> --- wireless-testing.orig/net/mac80211/wext.c	2009-05-04 17:41:53.000000000 +0200
> +++ wireless-testing/net/mac80211/wext.c	2009-05-04 17:42:02.000000000 +0200
> @@ -185,7 +185,7 @@ static int ieee80211_ioctl_giwfreq(struc
>  	if (sdata->vif.type == NL80211_IFTYPE_ADHOC)
>  		return cfg80211_ibss_wext_giwfreq(dev, info, freq, extra);
>  
> -	freq->m = local->hw.conf.channel->center_freq;
> +	freq->m = local->oper_channel->center_freq;
>  	freq->e = 6;
>  
>  	return 0;
>
>
>   
Nack.  People should see the hardware channel. If I set channel 6 and my 
wifi card is on channel 11 I need to see 11 not 6.  Doesn't make any 
sense to show some imaginary idealistic "I wish it was on channel 6" 
when the hardware is on channel 11.  Am I missing something here?

Thanks,
Rick Farina

> --
> To unsubscribe from this list: send the line "unsubscribe linux-wireless" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>
>   

Re: [PATCH] mac80211: report operating frequency rather than current

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1569 bytes --]

On Mon, 2009-05-04 at 13:50 -0400, Richard Farina wrote:
> Johannes Berg wrote:
> > It's not very helpful to see, in iwconfig, the current frequency
> > the card is tuned to if that frequency is currently somewhere
> > across the board because we're scanning. Since we keep track of
> > the frequency the user wants, display that instead.
> >
> > Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
> > ---
> >  net/mac80211/wext.c |    2 +-
> >  1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > --- wireless-testing.orig/net/mac80211/wext.c	2009-05-04 17:41:53.000000000 +0200
> > +++ wireless-testing/net/mac80211/wext.c	2009-05-04 17:42:02.000000000 +0200
> > @@ -185,7 +185,7 @@ static int ieee80211_ioctl_giwfreq(struc
> >  	if (sdata->vif.type == NL80211_IFTYPE_ADHOC)
> >  		return cfg80211_ibss_wext_giwfreq(dev, info, freq, extra);
> >  
> > -	freq->m = local->hw.conf.channel->center_freq;
> > +	freq->m = local->oper_channel->center_freq;
> >  	freq->e = 6;
> >  
> >  	return 0;
> >
> >
> >   
> Nack.  People should see the hardware channel. If I set channel 6 and my 
> wifi card is on channel 11 I need to see 11 not 6.  Doesn't make any 
> sense to show some imaginary idealistic "I wish it was on channel 6" 
> when the hardware is on channel 11.  Am I missing something here?

Umm, if _you_ set channel 11 then you _will_ see channel 11. It's just
that when it's scanning and happens to be on channel 132 instead of 11,
while you previously set 11, you will _after_ this patch see 11, not
132.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH] mac80211: report operating frequency rather than current

[Richard Farina]

Johannes Berg wrote:
> On Mon, 2009-05-04 at 13:50 -0400, Richard Farina wrote:
>   
>> Johannes Berg wrote:
>>     
>>> It's not very helpful to see, in iwconfig, the current frequency
>>> the card is tuned to if that frequency is currently somewhere
>>> across the board because we're scanning. Since we keep track of
>>> the frequency the user wants, display that instead.
>>>
>>> Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
>>> ---
>>>  net/mac80211/wext.c |    2 +-
>>>  1 file changed, 1 insertion(+), 1 deletion(-)
>>>
>>> --- wireless-testing.orig/net/mac80211/wext.c	2009-05-04 17:41:53.000000000 +0200
>>> +++ wireless-testing/net/mac80211/wext.c	2009-05-04 17:42:02.000000000 +0200
>>> @@ -185,7 +185,7 @@ static int ieee80211_ioctl_giwfreq(struc
>>>  	if (sdata->vif.type == NL80211_IFTYPE_ADHOC)
>>>  		return cfg80211_ibss_wext_giwfreq(dev, info, freq, extra);
>>>  
>>> -	freq->m = local->hw.conf.channel->center_freq;
>>> +	freq->m = local->oper_channel->center_freq;
>>>  	freq->e = 6;
>>>  
>>>  	return 0;
>>>
>>>
>>>   
>>>       
>> Nack.  People should see the hardware channel. If I set channel 6 and my 
>> wifi card is on channel 11 I need to see 11 not 6.  Doesn't make any 
>> sense to show some imaginary idealistic "I wish it was on channel 6" 
>> when the hardware is on channel 11.  Am I missing something here?
>>     
>
> Umm, if _you_ set channel 11 then you _will_ see channel 11. It's just
> that when it's scanning and happens to be on channel 132 instead of 11,
> while you previously set 11, you will _after_ this patch see 11, not
> 132.
>
>   
Yes, and at what point does it seem like a good idea to hide the channel 
the wifi card is on?  If I set channel 11 and it is scanning instead of 
locked on channel 11 then I should see the current channel the hardware 
is on.  This seems like an aweful idea to me, granted, it may help a few 
people that don't understand how scanning works, but hiding the truth is 
never a good idea. NACK.

-Rick Farina

> johannes
>   

Re: [PATCH] mac80211: report operating frequency rather than current

[John W. Linville]

On Mon, May 04, 2009 at 02:11:52PM -0400, Richard Farina wrote:
> Johannes Berg wrote:

>> Umm, if _you_ set channel 11 then you _will_ see channel 11. It's just
>> that when it's scanning and happens to be on channel 132 instead of 11,
>> while you previously set 11, you will _after_ this patch see 11, not
>> 132.
>>
>>   
> Yes, and at what point does it seem like a good idea to hide the channel  
> the wifi card is on?  If I set channel 11 and it is scanning instead of  
> locked on channel 11 then I should see the current channel the hardware  
> is on.  This seems like an aweful idea to me, granted, it may help a few  
> people that don't understand how scanning works, but hiding the truth is  
> never a good idea. NACK.

I can see what you mean, but I think showing seemingly random
fluctuations in channel assignments is at best distracting.  Don't you
agree that most people are more interested in seeing the configuration
state than the transient state of the hardware?

John
-- 
John W. Linville		Someday the world will need a hero, and you
linville@tuxdriver.com			might be all we have.  Be ready.

Re: [PATCH] mac80211: report operating frequency rather than current

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1059 bytes --]

On Mon, 2009-05-04 at 14:11 -0400, Richard Farina wrote:

> Yes, and at what point does it seem like a good idea to hide the channel 
> the wifi card is on?  If I set channel 11 and it is scanning instead of 
> locked on channel 11 then I should see the current channel the hardware 
> is on.  This seems like an aweful idea to me, granted, it may help a few 
> people that don't understand how scanning works, but hiding the truth is 
> never a good idea. NACK.

It's not really about hiding the truth. Besides, many many devices are
_already_ "hiding the truth" (to use your words) because they do
scanning in firmware and you never know what channel they're on (e.g.
all Intel wireless hardware).

I think it's completely pointless for us to reply to the question "what
channel am I on" with the answer "currently the receiver is tuned to
132", when we can say "you're associated to a network on channel 11".

After all, we don't unset the SSID/BSSID when scanning either, even
though we're currently not talking to the AP.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH] mac80211: report operating frequency rather than current

[Richard Farina]

Johannes Berg wrote:
> On Mon, 2009-05-04 at 14:11 -0400, Richard Farina wrote:
>
>   
>> Yes, and at what point does it seem like a good idea to hide the channel 
>> the wifi card is on?  If I set channel 11 and it is scanning instead of 
>> locked on channel 11 then I should see the current channel the hardware 
>> is on.  This seems like an aweful idea to me, granted, it may help a few 
>> people that don't understand how scanning works, but hiding the truth is 
>> never a good idea. NACK.
>>     
>
> It's not really about hiding the truth. Besides, many many devices are
> _already_ "hiding the truth" (to use your words) because they do
> scanning in firmware and you never know what channel they're on (e.g.
> all Intel wireless hardware).
>
> I think it's completely pointless for us to reply to the question "what
> channel am I on" with the answer "currently the receiver is tuned to
> 132", when we can say "you're associated to a network on channel 11".
>
>   
And if the card claims to be associated but ends up stuck on another 
channel?  I can't see a good reason for this at all but if no one agrees 
with me then it is well without your power to simply overrule me.

I think it is an awful idea, and intel being broken isn't an excuse to 
break everything else.  If the card isn't on the right channel because 
of some random failure this patch ensures there is no way to 
troubleshoot.  If you want to display current channel that is the 
hardware channel, if you want to display "Channel the AP I'm associated 
to is on" then call it something other than "Current Channel"

My 0.02.

Thanks,
Rick Farina

> After all, we don't unset the SSID/BSSID when scanning either, even
> though we're currently not talking to the AP.
>
> johannes
>   

Re: [PATCH] mac80211: report operating frequency rather than current

[Luis R. Rodriguez]

On Mon, May 4, 2009 at 11:30 AM, Richard Farina <sidhayn@gmail.com> wro=
te:
> Johannes Berg wrote:
>>
>> On Mon, 2009-05-04 at 14:11 -0400, Richard Farina wrote:
>>
>>
>>>
>>> Yes, and at what point does it seem like a good idea to hide the ch=
annel
>>> the wifi card is on? =C2=A0If I set channel 11 and it is scanning i=
nstead of
>>> locked on channel 11 then I should see the current channel the hard=
ware is
>>> on. =C2=A0This seems like an aweful idea to me, granted, it may hel=
p a few people
>>> that don't understand how scanning works, but hiding the truth is n=
ever a
>>> good idea. NACK.
>>>
>>
>> It's not really about hiding the truth. Besides, many many devices a=
re
>> _already_ "hiding the truth" (to use your words) because they do
>> scanning in firmware and you never know what channel they're on (e.g=
=2E
>> all Intel wireless hardware).
>>
>> I think it's completely pointless for us to reply to the question "w=
hat
>> channel am I on" with the answer "currently the receiver is tuned to
>> 132", when we can say "you're associated to a network on channel 11"=
=2E
>>
>>
>
> And if the card claims to be associated but ends up stuck on another
> channel? =C2=A0I can't see a good reason for this at all but if no on=
e agrees
> with me then it is well without your power to simply overrule me.
>
> I think it is an awful idea, and intel being broken isn't an excuse t=
o break
> everything else. =C2=A0If the card isn't on the right channel because=
 of some
> random failure this patch ensures there is no way to troubleshoot. =C2=
=A0If you
> want to display current channel that is the hardware channel, if you =
want to
> display "Channel the AP I'm associated to is on" then call it somethi=
ng
> other than "Current Channel"
>
> My 0.02.

Why not show both through iw and for wext just use whatever wext did
for most devices?

  Luis
--
To unsubscribe from this list: send the line "unsubscribe linux-wireles=
s" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] mac80211: report operating frequency rather than current

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 1356 bytes --]

On Mon, 2009-05-04 at 14:30 -0400, Richard Farina wrote:

> And if the card claims to be associated but ends up stuck on another 
> channel?

Then your network breaks anyway? And you see messages in the log that it
failed to switch channel? We're talking about mac80211 devices :)

> I can't see a good reason for this at all but if no one agrees 
> with me then it is well without your power to simply overrule me.
> 
> I think it is an awful idea, and intel being broken isn't an excuse to 
> break everything else.  If the card isn't on the right channel because 
> of some random failure this patch ensures there is no way to 
> troubleshoot.  If you want to display current channel that is the 
> hardware channel, if you want to display "Channel the AP I'm associated 
> to is on" then call it something other than "Current Channel"

It's not just Intel... All fullmac devices, ie. everything not using
mac80211, shows the same behaviour.

And trying to say "current channel" is "hardware channel" runs into the
nice pitfall that, unfortunately, wext is so undefined that you can't
really say either. I think the channel it's trying to use makes a whole
lot more sense because otherwise a user who just randomly does
"iwconfig" will be mightily confused if for example wpa_supplicant had
just requested a scan.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH] mac80211: report operating frequency rather than current

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 478 bytes --]

On Mon, 2009-05-04 at 11:33 -0700, Luis R. Rodriguez wrote:

> Why not show both through iw and for wext just use whatever wext did
> for most devices?

Where "most devices" counts as number of drivers (I think all
pre-mac80211 drivers behave like _after_ my patch), or number of
deployed devices (piles of Intel devices, I guess), or...?

Also keep in mind that for many devices we simply cannot provide the
real channel the radio is currently tuned to.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH] mac80211: report operating frequency rather than current

[Fabio Rossi]

On Monday 04 May 2009, John W. Linville wrote:

> On Mon, May 04, 2009 at 02:11:52PM -0400, Richard Farina wrote:
> > Johannes Berg wrote:
> >> Umm, if _you_ set channel 11 then you _will_ see channel 11. It's just
> >> that when it's scanning and happens to be on channel 132 instead of 11,
> >> while you previously set 11, you will _after_ this patch see 11, not
> >> 132.
> >
> > Yes, and at what point does it seem like a good idea to hide the channel
> > the wifi card is on?  If I set channel 11 and it is scanning instead of
> > locked on channel 11 then I should see the current channel the hardware
> > is on.  This seems like an aweful idea to me, granted, it may help a few
> > people that don't understand how scanning works, but hiding the truth is
> > never a good idea. NACK.
>
> I can see what you mean, but I think showing seemingly random
> fluctuations in channel assignments is at best distracting.  Don't you
> agree that most people are more interested in seeing the configuration
> state than the transient state of the hardware?

I want to report my experience. There was a bug in the ath5k driver with a 
hanging result during scanning mode. I was able to discover the problem 
thanks to the iwconfig output.

In alternative, is it possible to add a status flag so that iwconfig from user 
space can report scanning mode is operative?

Fabio

Re: [PATCH] mac80211: report operating frequency rather than current

[Luis R. Rodriguez]

On Mon, May 4, 2009 at 11:53 AM, Fabio Rossi <rossi.f@inwind.it> wrote:
> On Monday 04 May 2009, John W. Linville wrote:
>
>> On Mon, May 04, 2009 at 02:11:52PM -0400, Richard Farina wrote:
>> > Johannes Berg wrote:
>> >> Umm, if _you_ set channel 11 then you _will_ see channel 11. It's=
 just
>> >> that when it's scanning and happens to be on channel 132 instead =
of 11,
>> >> while you previously set 11, you will _after_ this patch see 11, =
not
>> >> 132.
>> >
>> > Yes, and at what point does it seem like a good idea to hide the c=
hannel
>> > the wifi card is on? =C2=A0If I set channel 11 and it is scanning =
instead of
>> > locked on channel 11 then I should see the current channel the har=
dware
>> > is on. =C2=A0This seems like an aweful idea to me, granted, it may=
 help a few
>> > people that don't understand how scanning works, but hiding the tr=
uth is
>> > never a good idea. NACK.
>>
>> I can see what you mean, but I think showing seemingly random
>> fluctuations in channel assignments is at best distracting. =C2=A0Do=
n't you
>> agree that most people are more interested in seeing the configurati=
on
>> state than the transient state of the hardware?
>
> I want to report my experience. There was a bug in the ath5k driver w=
ith a
> hanging result during scanning mode. I was able to discover the probl=
em
> thanks to the iwconfig output.
>
> In alternative, is it possible to add a status flag so that iwconfig =
from user
> space can report scanning mode is operative?

iw event

Or the old way : iwevent

  Luis
--
To unsubscribe from this list: send the line "unsubscribe linux-wireles=
s" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH] mac80211: report operating frequency rather than current

[Kalle Valo]

"John W. Linville" <linville@tuxdriver.com> writes:

>> Yes, and at what point does it seem like a good idea to hide the channel  
>> the wifi card is on?  If I set channel 11 and it is scanning instead of  
>> locked on channel 11 then I should see the current channel the hardware  
>> is on.  This seems like an aweful idea to me, granted, it may help a few  
>> people that don't understand how scanning works, but hiding the truth is  
>> never a good idea. NACK.
>
> I can see what you mean, but I think showing seemingly random
> fluctuations in channel assignments is at best distracting.  Don't you
> agree that most people are more interested in seeing the configuration
> state than the transient state of the hardware?

I agree, it's very distracting. And while doing mac80211 software scan
the device is on a channel approximately 30 ms with queues stopped, so
no data is transfered at that time. I don't see any benefit from
reporting this value to user space: "Hey, I'm on channel 5 now but by
the time to you read this I will be on channel 6 already."

If people want to follow how mac80211 changes the frequency during scan,
the proper way is to add debug messages for op_config() calls. I saw
that Johannes was already working on that.

-- 
Kalle Valo

Re: [PATCH] mac80211: report operating frequency rather than current

[Richard Farina]

Kalle Valo wrote:
> "John W. Linville" <linville@tuxdriver.com> writes:
>
>   
>>> Yes, and at what point does it seem like a good idea to hide the channel  
>>> the wifi card is on?  If I set channel 11 and it is scanning instead of  
>>> locked on channel 11 then I should see the current channel the hardware  
>>> is on.  This seems like an aweful idea to me, granted, it may help a few  
>>> people that don't understand how scanning works, but hiding the truth is  
>>> never a good idea. NACK.
>>>       
>> I can see what you mean, but I think showing seemingly random
>> fluctuations in channel assignments is at best distracting.  Don't you
>> agree that most people are more interested in seeing the configuration
>> state than the transient state of the hardware?
>>     
>
> I agree, it's very distracting. And while doing mac80211 software scan
> the device is on a channel approximately 30 ms with queues stopped, so
> no data is transfered at that time. I don't see any benefit from
> reporting this value to user space: "Hey, I'm on channel 5 now but by
> the time to you read this I will be on channel 6 already."
>
> If people want to follow how mac80211 changes the frequency during scan,
> the proper way is to add debug messages for op_config() calls. I saw
> that Johannes was already working on that.
>
>   
To be honest, I would be perfectly happy with this patch if something 
like a scanning "flag" were to be show in the output if that is what is 
happening.  I don't so much need to see the current channel (when in sta 
mode at least) as I need to know that the card is scanning if it is not 
currently on the channel the AP is on.  I'd be very happy to simply see 
it say "scanning" next to the channel or some such when this occurs.

thanks,
Rick Farina

Re: [PATCH] mac80211: report operating frequency rather than current

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 572 bytes --]

On Tue, 2009-05-05 at 14:32 -0400, Richard Farina wrote:

> To be honest, I would be perfectly happy with this patch if something 
> like a scanning "flag" were to be show in the output if that is what is 
> happening.  I don't so much need to see the current channel (when in sta 
> mode at least) as I need to know that the card is scanning if it is not 
> currently on the channel the AP is on.  I'd be very happy to simply see 
> it say "scanning" next to the channel or some such when this occurs.

Except that's totally not possible with wext.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]

Re: [PATCH] mac80211: report operating frequency rather than current

[Johannes Berg]

[-- Attachment #1: Type: text/plain, Size: 787 bytes --]

On Tue, 2009-05-05 at 20:49 +0200, Johannes Berg wrote:
> On Tue, 2009-05-05 at 14:32 -0400, Richard Farina wrote:
> 
> > To be honest, I would be perfectly happy with this patch if something 
> > like a scanning "flag" were to be show in the output if that is what is 
> > happening.  I don't so much need to see the current channel (when in sta 
> > mode at least) as I need to know that the card is scanning if it is not 
> > currently on the channel the AP is on.  I'd be very happy to simply see 
> > it say "scanning" next to the channel or some such when this occurs.
> 
> Except that's totally not possible with wext.

Also, again, it is impossible in practise to do that for all cards since
fullmac cards might be bg scanning without telling the host.

johannes

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]