Re: Upgrading to r297291 LAGG(4) stops working.

Please add an entry to src/UPDATING to explain the change needed.

Also, my question remains, if wlan now knows how to set the MAC
address, do we still need to set the MAC address manually and
can we go back to using lagg just like it was 2 non-wireless
connections (in the case that the underlying wireless device
_can_ change the MAC address).

In the case that the underlying wireless device cannot change
the MAC address, then the examples should change the MAC
address of the wired device to match that of the wireless
instead of the other way around (or recommend doing it this
way all the time).


On Fri, 4 Sep 2015, Daniel Eischen wrote:

> On Fri, 4 Sep 2015, Sergey Kandaurov wrote:
>
>> On 1 September 2015 at 04:47, John Baldwin <jhb_at_freebsd.org> wrote:
>>> On Monday, August 31, 2015 09:58:45 AM Adrian Chadd wrote:
>>>> Hi,
>>>> 
>>>> +glebius, as he recently messed around with the wifi stack and his
>>>> changes may have broken how mac addresses are assigned to the
>>>> hardware.
>>> 
>>> Glebius did break this, though not because of what you say.  It's broken
>>> because the 'ifconfig_ath0' line that sets the mac address no longer
>>> does anything because 'ath0' is no longer an interface (and so that
>>> line is now ignored, plus it wouldn't work if it were passed to ifconfig
>>> now anyway).
>>> 
>>> At the very least the Handbook section on this needs to be updated to give
>>> working instructions for both HEAD and stable branches.
>> 
>> What about this change?
>
> [ ... ]
>
>>     <programlisting>ifconfig_bge0="up"
>> -ifconfig_<replaceable>iwn0</replaceable>="<replaceable>ether
>> 00:21:70:da:ae:37</replaceable>"
>> wlans_<replaceable>iwn0</replaceable>="wlan0"
>> ifconfig_wlan0="WPA"
>> +ifconfig_wlan0_alias0="ether 00:21:70:da:ae:3"
>> cloned_interfaces="<literal>lagg<replaceable>0</replaceable></literal>"
>> ifconfig_<literal>lagg<replaceable>0</replaceable></literal>="laggproto
>> failover laggport bge0 laggport wlan0 DHCP"</programlisting>
>>       </example>
>
> Is this in human readable form anywhere?  Seems like it's missing
> something (the ifconfig_iwn0="up"?).
>
> So to go from this:
>
>  ifconfig_iwn0="ether 00:11:22:33:44:55"
>  wlans_iwn0=wlan0
>  ifconfig_wlan0="ssid MYSSID WPA"
>  ifconfig_re0="up"
>  cloned_interfaces="lagg0"
>  ifconfig_lagg0="laggproto failover laggport re0 laggport wlan0"
>  ifconfig_lagg0_alias0="inet 10.0.0.7 netmask 0xffffff00"
>
> It would now be:
>
>  ifconfig_iwn0="up"
>  wlans_iwn0=wlan0
>  ifconfig_wlan0="ssid MYSSID WPA"
>  ifconfig_wlan0_alias0="ether 00:11:22:33:44:55"
>  ifconfig_re0="up"
>  cloned_interfaces="lagg0"
>  ifconfig_lagg0="laggproto failover laggport re0 laggport wlan0"
>  ifconfig_lagg0_alias0="inet 10.0.0.7 netmask 0xffffff00"
>
> Is that correct?
>
> Do we even need the ifconfig_wlan0_alias0 to set the MAC address?
> If wlan0 (the cloned device under lagg0) knows how to set the MAC
> address, then shouldn't lagg be able to tell wlan0 to set the MAC
> address?  The problem before was that wlan didn't know how to
> set the MAC address, so it was ignored when lagg tried to set it.

-- 
DE