[Swan] _updown.klips and ip command return code causing conn failure

[Paul Wouters]

I noticed this failure:

from _updown.klips:

if ! ip -o route get ${PLUTO_MY_SOURCEIP%/*} | grep ipsec0; then
      cmd=change

However, on a machine with

eth0      Link encap:Ethernet  HWaddr 00:01:49:00:C2:9B
           inet addr:192.168.10.2  Bcast:192.168.10.255 Mask:255.255.255.0
eth1      Link encap:Ethernet  HWaddr 00:01:49:00:C2:9C
           inet addr:192.168.100.2  Bcast:192.168.100.255 Mask:255.255.255.0

ipsec0    Link encap:Ethernet  HWaddr 00:01:49:00:C2:9B
           inet addr:192.168.10.2  Mask:255.255.255.255
ipsec1    Link encap:Ethernet  HWaddr 00:01:49:00:C2:9C
           inet addr:192.168.100.2  Mask:255.255.255.255

I got:

"aaa" #2: up-client output: /usr/libexec/ipsec/_updown.klips: changesource "ip route change 192.168.20.0/24 dev ipsec0 src 192.168.100.2" failed (RTNETLINK answers: No such file or directory)

checking manually:

# ip -o route get 192.168.100.2
local 192.168.100.2 dev lo  src 192.168.100.2 \
cache <local>

# ip -o route get 111.111.111.111
111.111.111.111 via 192.168.100.1 dev eth1  src
192.168.100.2 \    cache

(note none of those say "ipsec" in the output.

# export PLUTO_MY_SOURCEIP=192.168.100.2
# ip -o route get ${PLUTO_MY_SOURCEIP%/*} | grep ipsec0
# echo $?
1

I then tested using:

# if test -n "`ip -o route get ${PLUTO_MY_SOURCEIP%/*} | grep ipsec0`"; then echo change; fi
#

Which seems a better way to do things.

I'll commit this change if no one objects once the tree is unfrozen,

Paul