At 2013-04-10 18:35:12,"Banyan He" <banyan at rootong.com> wrote: >I remember I saw this behavior before. Mostly important thing is to find >out if the enc/dec goes well. I was having this problem that triggers >kernel to send out the icmp. > enc/dec ?? You mean the content of ip header, which has been added in my program while using raw socket?? How to fix or hack it, is there some way to do research?? I feel the package is sent to the upper GRE protocol while been intercepted with raw socket. Is there one way to prevent this?? Thanks. >------------ >Banyan He >Blog: http://www.rootong.com >Email: banyan at rootong.com > >On 4/10/2013 5:27 PM, yihect wrote: >> >> Hi, all: >> >> I'm playing with Gre Tunnel programming in centos 6. >> >> I wrote my program in these steps: >> a, open /dev/net/tun device and ioctl with TUNSETIFF, get fd to tun device, say "tun0"; >> b, create one socket and use it to set tun0's ip address, namyly SIOCSIFADDR and SIOCSIFDSTADDR; and then turn it up; >> c, create one raw socket to send things which got from the tun0 device to other machine, and to receive packages which should be feeded into the tun0 device; >> >> Finally, I ran the program in two machines, say MA and MB, synchronously, and set route info in MA and MB too. The last environment is like this: >> >> 10.0.3.1/24 192.168.0.31/24 192.168.0.124/24 10.0.4.1/24 >> ---------- MA ------------------------------------------- MB ---------- >> tun0 eth0 eth0 tun0 >> >> When I ping MB in MA, and tcpdump eth0 in MA, I get: >> >> 15:06:39.712390 IP 192.168.0.31 > 192.168.0.124: GREv0, length 88: IP 10.0.3.1 > 10.0.4.1: ICMP echo request, id 28797, seq 1, length 64 >> 15:06:39.713232 IP 192.168.0.124 > 192.168.0.31: ICMP 192.168.0.124 protocol 47 port 2048 unreachable, length 116 >> 15:06:39.713667 IP 192.168.0.124 > 192.168.0.31: GREv0, length 88: IP 10.0.4.1 > 10.0.3.1: ICMP echo reply, id 28797, seq 1, length 64 >> 15:06:39.713754 IP 192.168.0.31 > 192.168.0.124: ICMP 192.168.0.31 protocol 47 port 2048 unreachable, length 116 >> >> The 1st and 3rd lines is easy to understand, but what about the 2nd and 4th lines??? Is the raw package been sent up into gre protocol while caught by my program?? >> >> I have seen ICMP_PORT_UNREACH stuff been placed in ip_gre.c kernel code, but say it is "Impossible event". So, what's wrong, have I missed something in my program? Any suggestions are appreciated, thanks. >> >> >> >> >> _______________________________________________ >> CentOS mailing list >> CentOS at centos.org >> http://lists.centos.org/mailman/listinfo/centos >