Been a long day, but I think I finally see the problem.
If you have an ath9k AP interface running with hostapd, and you
run: ip link set vap0 down; ip link set vap0 up;
then it will disable beaconing.
One reason is that the ieee80211_do_open calls the
ieee80211_bss_info_change_notify before it sets the RUNNING
flag, so it would disable beaconing. A second is that
it doesn't set the BEACON_CHANGED flag anyway, so even if you
hack things to set RUNNING first, it still doesn't work right.
I'm not sure where the problem actually lies: Should ath9k
start beaconing automatically on VAP interface add? Is
it up to hostapd to detect the ifdown/ifup and re-set everything
up properly? Or maybe it's just a very bad idea to bounce
a VAP interface with 'ip link set'?
Thanks,
Ben
--
Ben Greear <[email protected]>
Candela Technologies Inc http://www.candelatech.com
El 14/01/2011 10:12, Jouni Malinen escribi?:
> On Thu, Jan 13, 2011 at 11:13:40PM -0800, Ben Greear wrote:
>> If you have an ath9k AP interface running with hostapd, and you
>> run: ip link set vap0 down; ip link set vap0 up;
>> then it will disable beaconing.
>
> Yes. So what?
>
>> I'm not sure where the problem actually lies: Should ath9k
>> start beaconing automatically on VAP interface add? Is
>> it up to hostapd to detect the ifdown/ifup and re-set everything
>> up properly? Or maybe it's just a very bad idea to bounce
>> a VAP interface with 'ip link set'?
>
> I would say it is up to whoever decided to set the interface down to
> restart hostapd.. There is a proposed patch for hostapd to do this
> automatically, but I'm yet to fully understand why one would set the
> interface down in the first place.
I usually do it to influence routing from within quagga. Quagga have no
provisions to restart another daemons on interface state change. This is just an
example, I'm sure there are others valid use cases, so I really think that
hostapd should restart beaconing/whatever in this case.
Regards,
Jorge
--
==============================================================
Jorge Boncompte - Ingenieria y Gestion de RED
DTI2 - Desarrollo de la Tecnologia de las Comunicaciones
--------------------------------------------------------------
C/ Abogado Enriquez Barrios, 5 14004 CORDOBA (SPAIN)
Tlf: +34 957 761395 / FAX: +34 957 450380
==============================================================
- There is only so much duct tape you can put on something
before it just becomes a giant ball of duct tape.
==============================================================
On Thu, 2011-01-13 at 23:13 -0800, Ben Greear wrote:
> Been a long day, but I think I finally see the problem.
>
> If you have an ath9k AP interface running with hostapd, and you
> run: ip link set vap0 down; ip link set vap0 up;
> then it will disable beaconing.
>
> One reason is that the ieee80211_do_open calls the
> ieee80211_bss_info_change_notify before it sets the RUNNING
> flag, so it would disable beaconing. A second is that
> it doesn't set the BEACON_CHANGED flag anyway, so even if you
> hack things to set RUNNING first, it still doesn't work right.
>
> I'm not sure where the problem actually lies: Should ath9k
> start beaconing automatically on VAP interface add? Is
> it up to hostapd to detect the ifdown/ifup and re-set everything
> up properly? Or maybe it's just a very bad idea to bounce
> a VAP interface with 'ip link set'?
Why would that be a good idea? I mean ... why would you want to do
that?!
johannes
On Fri, Jan 14, 2011 at 8:22 AM, Ben Greear <[email protected]> wrote:
> On 01/14/2011 01:12 AM, Jouni Malinen wrote:
>>
>> On Thu, Jan 13, 2011 at 11:13:40PM -0800, Ben Greear wrote:
>>>
>>> If you have an ath9k AP interface running with hostapd, and you
>>> run: ?ip link set vap0 down; ip link set vap0 up;
>>> then it will disable beaconing.
>>
>> Yes. So what?
>>
>>> I'm not sure where the problem actually lies: ?Should ath9k
>>> start beaconing automatically on VAP interface add? ?Is
>>> it up to hostapd to detect the ifdown/ifup and re-set everything
>>> up properly? ?Or maybe it's just a very bad idea to bounce
>>> a VAP interface with 'ip link set'?
>>
>> I would say it is up to whoever decided to set the interface down to
>> restart hostapd.. There is a proposed patch for hostapd to do this
>> automatically, but I'm yet to fully understand why one would set the
>> interface down in the first place.
>
> My tool automatically downs and ups interfaces when IPs configuration
> changes to clear IP & routing information. ?It's not so difficult to
> restart hostapd or even not bounce the VAP to begin with.
>
> But, it was quite difficult to figure out
> that the VAP was just *mostly* working. ?It seems that everything but
> beaconing was working, or at least 70 or so virtual station devices
> could associate and pass traffic, though they did complain about
> not receiving beacons.
>
> I think that we should either change hostapd to fix this up automatically,
> or have hostapd exit on error when this happens, or have some other very
> obvious clue that the VAP is not fully functional.
>
> Thanks,
> Ben
OpenWrt brings the AP interface down after launching hostapd to set
the MAC address every time during network restart, and beaconing
becomes disabled then, too. I created this ticket 6 weeks ago:
https://dev.openwrt.org/ticket/8362
My conclusion was that bringing the interface down with hostapd up is
a terrible thing to do. Some sort of fix beyond "don't do that" would
be really nice, though. It would be great if hostapd would detect
that the interface went down, and reinitialize when the interface
comes back up.
On Thu, Jan 13, 2011 at 11:13:40PM -0800, Ben Greear wrote:
> If you have an ath9k AP interface running with hostapd, and you
> run: ip link set vap0 down; ip link set vap0 up;
> then it will disable beaconing.
Yes. So what?
> I'm not sure where the problem actually lies: Should ath9k
> start beaconing automatically on VAP interface add? Is
> it up to hostapd to detect the ifdown/ifup and re-set everything
> up properly? Or maybe it's just a very bad idea to bounce
> a VAP interface with 'ip link set'?
I would say it is up to whoever decided to set the interface down to
restart hostapd.. There is a proposed patch for hostapd to do this
automatically, but I'm yet to fully understand why one would set the
interface down in the first place.
--
Jouni Malinen PGP id EFC895FA
On 01/14/2011 01:12 AM, Jouni Malinen wrote:
> On Thu, Jan 13, 2011 at 11:13:40PM -0800, Ben Greear wrote:
>> If you have an ath9k AP interface running with hostapd, and you
>> run: ip link set vap0 down; ip link set vap0 up;
>> then it will disable beaconing.
>
> Yes. So what?
>
>> I'm not sure where the problem actually lies: Should ath9k
>> start beaconing automatically on VAP interface add? Is
>> it up to hostapd to detect the ifdown/ifup and re-set everything
>> up properly? Or maybe it's just a very bad idea to bounce
>> a VAP interface with 'ip link set'?
>
> I would say it is up to whoever decided to set the interface down to
> restart hostapd.. There is a proposed patch for hostapd to do this
> automatically, but I'm yet to fully understand why one would set the
> interface down in the first place.
My tool automatically downs and ups interfaces when IPs configuration
changes to clear IP & routing information. It's not so difficult to
restart hostapd or even not bounce the VAP to begin with.
But, it was quite difficult to figure out
that the VAP was just *mostly* working. It seems that everything but
beaconing was working, or at least 70 or so virtual station devices
could associate and pass traffic, though they did complain about
not receiving beacons.
I think that we should either change hostapd to fix this up automatically,
or have hostapd exit on error when this happens, or have some other very
obvious clue that the VAP is not fully functional.
Thanks,
Ben
--
Ben Greear <[email protected]>
Candela Technologies Inc http://www.candelatech.com