Bug#362745: [pkg-wpa-devel] Bug#362745: wpasupplicant: Managed mode leaves wpa_supplicant running if ifup fails

[Kel Modderman]

Felix Homann wrote:
> Package: wpasupplicant
> Version: 0.4.8-3
> Severity: normal
>
>
> If ifup fails on an interface configured to use wpasupplicant's "managed
> mode" a wpa_supplicant process is left running in the background.
>   

And the device is still marked as up in ifstate, so wpa_supplicant has 
every right to be active.
> The same happens if you interrupt ifup by pressing CTRL-C before it has 
> finished.
>   

This is *not* the same. I think this is the real flaw; ifupdown cannot 
send a "down" signal on a user interrupt.

> This has a major effect if you're trying to set up managed mode: If the first 
> attempt fails the second attempt will fail, too.
>   

Only when you cancel the first attempt manually, IMO.

> How to reproduce this behavior:
>
> Break a working "managed mode" setup by commenting out the wpa-passphrase 
> line:
>
> # /etc/network/interfaces:
> iface wlan inet dhcp
>         wpa-driver wext
>         wpa-ssid yourssid
>         wpa-key-mgmt WPA-PSK
> #        wpa-passphrase Iwonttellyou
>         wpa-proto WPA
>
> ifup the interface and wait until ifup gives in. You will still have a 
> wpa_supplicant running in the background
>   

If you ifup the device, and do not cancel the operation manually, the 
device is marked up regardless of whether or not the inet method was 
successful.

For example:

iface ath0 inet dhcp
    wpa-ssid MadWifiAP
    #wpa-passphrase foo

# ifup ath0
Internet Systems Consortium DHCP Client V3.0.3
Copyright 2004-2005 Internet Systems Consortium.
All rights reserved.
For info, please visit http://www.isc.org/products/DHCP

wifi1: unknown hardware address type 801
wifi0: unknown hardware address type 801
wifi1: unknown hardware address type 801
wifi0: unknown hardware address type 801
Listening on LPF/ath0/00:09:5b:94:51:82
Sending on   LPF/ath0/00:09:5b:94:51:82
Sending on   Socket/fallback
DHCPDISCOVER on ath0 to 255.255.255.255 port 67 interval 3
DHCPDISCOVER on ath0 to 255.255.255.255 port 67 interval 4
DHCPDISCOVER on ath0 to 255.255.255.255 port 67 interval 7
DHCPDISCOVER on ath0 to 255.255.255.255 port 67 interval 9
DHCPDISCOVER on ath0 to 255.255.255.255 port 67 interval 20
DHCPDISCOVER on ath0 to 255.255.255.255 port 67 interval 18
No DHCPOFFERS received.
No working leases in persistent database - sleeping.
# ifup ath0
ifup: interface ath0 already configured
# ifup ath0
ifup: interface ath0 already configured
# ifdown ath0
Internet Systems Consortium DHCP Client V3.0.3
Copyright 2004-2005 Internet Systems Consortium.
All rights reserved.
For info, please visit http://www.isc.org/products/DHCP

wifi1: unknown hardware address type 801
wifi0: unknown hardware address type 801
wifi1: unknown hardware address type 801
wifi0: unknown hardware address type 801
Listening on LPF/ath0/00:09:5b:94:51:82
Sending on   LPF/ath0/00:09:5b:94:51:82
Sending on   Socket/fallback
DHCPRELEASE on ath0 to 192.168.0.1 port 67
send_packet: Network is unreachable
send_packet: please consult README file regarding broadcast address.


The first attempt failed, yet the device was still marked up, and 
ifupdown recorded the state. The wpa_supplicant daemon has every right 
to keep running, in the chance that the inet method will eventually succeed.

Therefore, you must ifdown the iface before attempting to configure it a 
second time.

If you get impatient and kill the process manually, the state is not 
recorded as up, and we hit the problem you describe. We can work around 
this by killing these orphans next time we up the iface, but a better 
solution would be for  ifupdown to send a down signal when it is 
interrupted by the user.

> Now, uncomment the wpa-passphrase line and try to ifup wlan again. It will 
> fail.
>   

Only if the first attempt was cut short by manual termination. The issue 
here is that we are spawning daemons, these are not sent any "die" 
signal when ifupdown is terminated manually.

Anyway, thanks for bringing this up, it is an interesting situation.

Please correct my statements and examples if they are wrong, I have only 
just begun investigating it.

Thanks, Kel.