[WBEL-users] cipe on x86_64

[Tanel Kindsigo]

Has anybody tried to run Cipe RPM package provided with distribution for 
x86_64 architecture? I didn't get it to work, though on some other 
machine, x86_32 it works for me.



If anyone's interested, here's deeper problem description:


I'm using WhiteboxLinux on 2 
machines, one is x86_32, other is x86_64 architecture, 
kernel-2.4.21-15.0.4.EL. I use cipe-1.4.5-16 original rpm packages for 
both of them.

Tried to connect both machines with RedHat 7.3 router(cipe-1.4.5-9). 
I used same configuration for both Whitebox machines. x86_32 worked well, 
x86_64 did not. The connection was established, but if I tried to ping RH7.3 
machine (193.40.x.x) from my Whitebox (192.168.1.213 behind NAT), then 
tcpdump -i eth0 on WhiteBox showed that my ping requests were transferred 
to the destination and replies even came back, but 
tcpdump -i cipcb0 showed only icmp requests, not replies.


Kernel IP routing table contains:
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
192.168.150.2   0.0.0.0         255.255.255.255 UH    0      0        0 cipcb0

ifconfig cipcb0 shows some errors:

cipcb0    Link encap:IPIP Tunnel  HWaddr
          inet addr:192.168.150.1  P-t-P:192.168.150.2  Mask:255.255.255.255
          UP POINTOPOINT NOTRAILERS RUNNING NOARP  MTU:1442  Metric:1
          RX packets:0 errors:49 dropped:0 overruns:0 frame:49
          TX packets:42 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:100
          RX bytes:0 (0.0 b)  TX bytes:5544 (5.4 Kb)


/etc/sysconfig/network-scripts/ifcfg-cipcb0 contains:

DEVICE=cipcb0
ONBOOT=yes
USERCTL=yes

# PORT specifies which UDP port is used by CIPE process
# for the local endpoint.  Make sure this passes through possible
# firewalls.
ME=192.168.1.213
MYPORT=6666

# PEER specifies the _real_ address of the remote CIPE endpoint.
# 0.0.0.0 means it's allocated dynamically (only one end can set it this,
# though).  You must specifiy the peer port address on this line
# in the form peer.host.example.com:port
PEER=193.40.xx.xx:6666

# PTPADDR specifies the _virtual_ address for the remote end of the
# CIPE tunnel.  Usually it should be an address reserved for local
# networks so that it won't mess up routing to "real" addresses.
PTPADDR=192.168.150.2

# IPADDR specifies the _virtual_ address for the local end of the
# CIPE tunnel.  Usually it should be an address reserved for local
# networks so that it won't mess up routing to "real" addresses.
IPADDR=192.168.150.1




Then I switched cipcb module to debug mode 4, then 16 and 128 and tried to 
reload the tunnel couple of times, dmesg shows:


cipcb0: sending 132 from 192.168.1.213:6666 to 193.40.xxx.xxx:6666
cipcb0: cipe_xmit len=84(84)
route dev=eth0 flags=0 type=1
pmtu=1452 dmtu=1500 size=84
cipcb0: sending 132 from 192.168.1.213:6666 to 193.40.xxx.xxx:6666
cipcb0: cipe_xmit len=84(84)
route dev=eth0 flags=0 type=1
pmtu=1452 dmtu=1500 size=84
cipcb0: sending 132 from 192.168.1.213:6666 to 193.40.xxx.xxx:6666
cipcb0: cipe_xmit len=84(84)
route dev=eth0 flags=0 type=1
pmtu=1452 dmtu=1500 size=84
cipcb0: sending 132 from 192.168.1.213:6666 to 193.40.xxx.xxx:6666
divert: no divert_blk to free, cipcb0 not ethernet
cipcb: CIPE driver vers 1.4.5 (c) Olaf Titz 1996-2000, 100 channels, 
debug=16
divert: not allocating divert_blk for non-ethernet device cipcb0
cipcb0: setkey 1
cipcb0: fake kxc block typ=1
cipcb0: real sendmsg len 128 text=68c0...
cipcb0: setkey 10
cipcb0: fake kxc block typ=1
cipcb0: real sendmsg len 72 text=11ea...
cipcb0: setkey 10
eth0: Promiscuous mode enabled.
device eth0 entered promiscuous mode
device eth0 left promiscuous mode
cipcb0: fake kxc block typ=1
cipcb0: real sendmsg len 304 text=1699...
cipcb0: setkey 10
cipcb0: skb_recv_datagram: -512
cipcb0: real sendmsg len 136 text=6088...
divert: no divert_blk to free, cipcb0 not ethernet
cipcb: CIPE driver vers 1.4.5 (c) Olaf Titz 1996-2000, 100 channels, 
debug=128
divert: not allocating divert_blk for non-ethernet device cipcb0