On Sun, Oct 24, 2010 at 3:19 PM, Jan Engelhardt <jengelh@xxxxxxxxxx> wrote: > On Sunday 2010-10-24 14:34, Kfir Lavi wrote: > >>On Sun, Oct 24, 2010 at 1:31 PM, Jan Engelhardt <jengelh@xxxxxxxxxx> wrote: >>> On Sunday 2010-10-24 10:54, Kfir Lavi wrote: >>> >>>>I will need to move Ethernet, ARP and IP packets. >>>>What are my options regarding moving the packet modified with new >>>>custom header from one process to another process? >>> >>> http://en.wikipedia.org/wiki/Inter-process_communication lists ways of >>> transferring data between programs. >>> >> >>Thanks, >>There is no mention in this page about Netlink. > > Netlink belongs to the class of sockets. > >>As I'm getting packets from the kernel via netlink, I would think it's >>a good choice to continue using netlink. >>I'm trying now to create a small example code using libnl. >>What is your take on using netlink? > > For user-to-user communications, the code would be simple enough to > not warrant the use of libnl. Neither is a nlmsghdr structure needed. > > My take is that I would not normally use Netlink for user-to-user > communications. For one, you would need to know the PID of the other > process, which is not always feasible to determine. Compare with a named > PF_LOCAL socket where both processes only need to know a rendezvous > point. Even TCP socket numbers are more predictable - and controllable > even - than a random PID. Also note that due to Netlink's use of PIDs, > you cannot access other PID namespaces. > I'm not sure I have a problem with the PID's, because both processes are in my hand, and I write the code for them. Does this changes your answer? I'm having a look on PF_LOCAL, to see if it suits my needs. Regards, Kfir -- To unsubscribe from this list: send the line "unsubscribe netfilter-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
