Merge pull request #1 from evilsocket/master

pull request

README.md

@@ -2,11 +2,15 @@
 
 OpenSnitch is a GNU/Linux port of the Little Snitch application firewall.
 
+<p align="center">
+  <img src="https://raw.githubusercontent.com/evilsocket/opensnitch/master/screenshot.png" alt="OpenSnitch"/>
+</p>
+
 **Warning: This is still alpha quality software, don't rely on it (yet) for your computer security.**
 
-<center>
-  <img src="https://raw.githubusercontent.com/evilsocket/opensnitch/master/screenshot.png" alt="OpenSnitch"/>
-</center>
+## Requirements
+
+You'll need a GNU/Linux distribution with `iptables`, `NFQUEUE` and `ftrace` kernel support.
 
 ## Install
 
@@ -18,6 +22,33 @@ OpenSnitch is a GNU/Linux port of the Little Snitch application firewall.
 
     sudo opensnitch
 
+## Known Issues / Future Improvements
+
+Before opening an issue, keep in mind that the current implementation is just an experiment to see the doability of the project, future improvements of OpenSnitch will include:
+
+Split the project into `opensnitchd`, `opensnitch-ui` and `opensnitch-ruleman`:
+
+* `opensnitchd` will be a C++ daemon, running as root with the main logic. It'll fix [this](https://github.com/evilsocket/opensnitch/issues/28).
+* `opensnitch-ui` python (?) UI running as normal user, getting the daemon messages. Will fix [this](https://github.com/evilsocket/opensnitch/issues/20).
+* `opensnitch-ruleman` python (?) UI for rule editing.
+
+## How Does It Work
+
+OpenSnitch is an application level firewall, meaning then while running, it will detect and alert the user for every outgoing connection applications he's running are creating. This can be extremely **effective to detect and block unwanted connections** on your system that might be caused by a security breach, **causing data exfiltration to be much harder for an attacker**.
+In order to do that, OpenSnitch relies on `NFQUEUE`, an `iptables` target/extension which allows an userland software to intercept IP packets and either `ALLOW` or `DROP` them, once started it'll install the following iptables rules:
+
+    OUTPUT -t mangle -m conntrack --ctstate NEW -j NFQUEUE --queue-num 0 --queue-bypass 
+
+This will use `conntrack` iptables extension to pass all newly created connection packets to NFQUEUE number 0 (the one OpenSnitch is listening on), and then:
+
+    INPUT --protocol udp --sport 53 -j NFQUEUE --queue-num 0 --queue-bypass
+
+This will also redirect DNS queries to OpenSnitch, allowing the software to perform and IP -> hostname resolution without performing active DNS queries itself.
+
+Once a new connection is detected, the software relies on the `ftrace` kernel extension in order to track which PID (therefore which process) is creating the connection.
+
+If `ftrace` is not available for your kernel, OpenSnitch will fallback using the `/proc` filesystem, even if this method will also work, it's vulnerable to application path manipulation as [described in this issue](https://github.com/evilsocket/opensnitch/issues/12), therefore **it's highly suggested to run OpenSnitch on a ftrace enabled kernel**.
+
 ## TODOs
 
     grep -r TODO opensnitch | cut -d '#' -f 2 | sort -u

bin/opensnitch

@@ -1,4 +1,4 @@
-#!/usr/bin/python
+#!/usr/bin/env python2
 # This file is part of OpenSnitch.
 #
 # Copyright(c) 2017 Simone Margaritelli
@@ -61,5 +61,3 @@ def main():
     snitch.stop()
 
 main()
-
-

opensnitch/procmon.py

@@ -81,10 +81,11 @@ def is_ftrace_available():
         return False
 
     def get_app_name( self, pid ):
-        pid = int(pid)
-        with self.lock:
-            if pid in self.pids and 'filename' in self.pids[pid]:
-                return self.pids[pid]['filename']
+        if pid is not None:
+            pid = int(pid)
+            with self.lock:
+                if pid in self.pids and 'filename' in self.pids[pid]:
+                    return self.pids[pid]['filename']
 
         return None
 

opensnitch/snitch.py

@@ -35,7 +35,7 @@ class Snitch:
                        # Get connection packets
                        "OUTPUT -t mangle -m conntrack --ctstate NEW -j NFQUEUE --queue-num 0 --queue-bypass",
                        # Reject packets marked by OpenSnitch
-                       "OUTPUT --protocol tcp -m mark --mark 1 -j REJECT" )
+                       "OUTPUT --protocol tcp -m mark --mark 101285 -j REJECT" )
 
     # TODO: Support IPv6!
     def __init__( self ):
@@ -87,7 +87,7 @@ def pkt_callback(self,pkt):
         if verd == Rule.DROP:
             logging.info( "Dropping %s from %s" % ( conn, conn.get_app_name() ) )
             # mark this packet so iptables will drop it
-            pkt.set_mark(1)
+            pkt.set_mark(101285)
             pkt.drop()
         else:
             pkt.accept()