Possible bug in vpnc-connect script

David Woodhouse dwmw2 at infradead.org
Tue Feb 23 13:15:04 EST 2010


On Fri, 2010-01-22 at 00:32 +0100, Bernhard Schmidt wrote:
> 
> we just stumbled about a weird bug in OpenConnect in conjunction with 
> the vpnc-script from 
> http://git.infradead.org/users/dwmw2/vpnc-scripts.git. Possibly caused 
> by a misconfiguration on our side, but I think it could be solved pretty 
> easily in vpnc-script
> 
> Our ASA is at the IP address x.x.254.40. It returns
> 
> X-CSTP-Address: x.x.48.207
> X-CSTP-Netmask: 255.255.0.0
> 
> i.e. a wrong netmask. What happens now is do_connect() in vpnc-script 
> invokes do_ifconfig(), which configures a route to 
> x.x.48.207/255.255.0.0 = x.x.0.0/16 to the tun interface. After that 
> do_connect() invokes set_vpngateway_route(), which retrieves the route 
> for x.x.254.40 (at this moment already pointing to tun0) and sets a 
> hostroute. And we have a very pretty endless recursion.
> 
> Proposed solution: reorder do_ifconfig() and set_vpngateway_route(), 
> since setting the hostroute to the gateway should always operate on the 
> previous routingtable.
> 
> Very basic git patch attached.

Looks sensible to me. I've applied it to the git repository; thanks.

-- 
David Woodhouse                            Open Source Technology Centre
David.Woodhouse at intel.com                              Intel Corporation




More information about the openconnect-devel mailing list