Hello everyone, I sent this to the wrong email address first, so I hope this one gets through.
I've been trying to run a simple firewall with the ip_pipeline example application but without success so far. After some initial setup issues, I was able to successfully run the firewall and add specific rules to it. But the problem is, if I have understood the parameters correctly, my rules don't seem to have any effect on accepting or rejecting packets through the firewall. On my current tests, I have 2 different hosts (10.0.0.1 and 10.10.0.1), connected to a central host that runs the firewall. This central host should be responsible for blocking/accepting packets between these two hosts, but I don't seem to get any packets through. I configure two links for the ip_pipeline application (config file below) and I check the connections (and they work) before starting the application. This is the config file I'm using: [PIPELINE0] type = MASTER core = 0 [PIPELINE1] type = FIREWALL core = 1 pktq_in = RXQ0.0 RXQ1.0 pktq_out = TXQ0.0 TXQ1.0 When the application is running, I add these two rules: p 1 firewall add ipv4 1 10.0.0.0 24 10.10.0.0 24 0 65535 0 65535 0 0 1 p 1 firewall add ipv4 2 10.10.0.0 24 10.0.0.0 24 0 65535 0 65535 0 0 0 The first, as far as I understood the parameters, should accept packets coming from 10.0.0.0/24 and bound to 10.10.0.0/24, for any TCP port (0-65535), both for destination and source, for any protocol (protocol mask = 0) and forward them to link 1, which is connected to the host 10.10.0.1. The second is similar to the first, but on the other direction 10.10.0.0/24 -> 10.0.0.0/24. When I try pinging a host from the other, I can see the packets being received by the application (command "link ls" shows # of RX packets increasing gradually) but none of them gets transmitted to the specified link. The only situation in which I can get packets to the other side, is when I allow all traffic through the "p 1 firewall add default <link>" command. I wondered if the packets were being modified somewhere, so I looked for the place in the code where the packets were actually being matched against the rules so I could print them and check. While looking through the code, I saw that the dpdk ACL data structure was being used to implement the rules, and after reading the DPDK Programmer's Guide chapter regarding the acl library (Chapter 22), I saw that the function rte_acl_classify should be called to match the packets against the rules, but I couldn't find this function being called anywhere in the code or in the libraries being used by the application. So my question is: Am I missing anything in my configuration in order to successfully run the firewall? Did I interpret the rule parameters correctly? Or, although I think this is less likely, is the firewall failing to match the packets against the rules? I've been working on this for a few days and I can't seem to understand what's wrong. I appreciate your kind help. Thank you. -- Matheus Salgueiro Castanho
