0.2.5 - cleaned up multicast propagation algorithm and factored it out into Multicaster.hpp and BloomFilter.hpp

(1) Probable fix for issue #7 and major cleanup of EthernetTap code with consolidation for all unix-like systems and specialization for different flavors only when needed.

(2) Refactor of Buffer<> to make its members private, and Packet to use Buffer's methods exclusively to access them. This improves clarity and means we're no longer lying about Buffer's role in the code's security posture.

(3) Add -fstack-protect to Makefile to bounds check stack variables.

AUTHORS.txt

@@ -1,6 +1,6 @@
 This file will track authors and contributors to ZeroTier's code base.
-People who contribute code that is accepted into the master (trunk) branch
-will be credited here.
+People who contribute new code or bug fixes that are accepted into
+the master (trunk) branch will be credited here.
 
 --
 

BUILDING.txt

@@ -2,6 +2,10 @@ Building ZeroTier One on different platforms:
 
 (See RUNNING.txt for what to do next.)
 
+Developers note: there is currently no management of dependencies on *nix
+platforms, so you should make clean ; make if you change a header. Will
+do this eventually.
+
 -- MacOS
 
 make -f Makefile.mac

Makefile.linux

@@ -6,7 +6,7 @@ ARCH=$(shell uname -m)
 DEFS=-DZT_ARCH="$(ARCH)" -DZT_OSNAME="linux" -DZT_TRACE
 
 # Uncomment for a release optimized build
-CFLAGS=-Wall -O6 -fno-unroll-loops -pthread $(INCLUDES) -DNDEBUG $(DEFS)
+CFLAGS=-Wall -O3 -fno-unroll-loops -fstack-protector -pthread $(INCLUDES) -DNDEBUG $(DEFS)
 STRIP=strip --strip-all
 
 # Uncomment for a debug build

Makefile.mac

@@ -5,7 +5,7 @@ INCLUDES=-Iext/bin/libcrypto/include -Iext/jsoncpp/include
 DEFS=-DZT_ARCH="x86_combined" -DZT_OSNAME="mac" -DZT_TRACE
 
 # Uncomment for a release optimized universal binary build
-CFLAGS=-arch i386 -arch x86_64 -Wall -O6 -ftree-vectorize -pthread -mmacosx-version-min=10.6 -DNDEBUG $(INCLUDES) $(DEFS)
+CFLAGS=-arch i386 -arch x86_64 -Wall -O3 -ftree-vectorize -fstack-protector -pthread -mmacosx-version-min=10.6 -DNDEBUG $(INCLUDES) $(DEFS)
 STRIP=strip
 
 # Uncomment for a debug build

README.md

@@ -5,7 +5,7 @@ ZeroTier One creates flat virtual Ethernet networks of almost unlimited size. [V
 
 This code is presently in **ALPHA** testing. That means that the protocol spec may change in incompatible ways, and it certainly has bugs. Testers should "git pull," rebuild, and restart fairly often. If things mysteriously stop working, do that.
 
-See BUILDING.txt and RUNNING.txt for instructions. It currently builds on Mac and Linux. A Windows port is coming soon. Nice packages/installers and auto-update is also coming when alpha transitions to beta.
+See BUILDING.txt and RUNNING.txt for instructions. It currently builds on Mac and Linux. A Windows port is coming soon. Nice packages/installers and auto-update are also coming when alpha transitions to beta.
 
 Note that this won't work if your firewall does not allow outbound UDP. It must allow UDP conversations on port 8993 at a minimum.
 
@@ -13,7 +13,7 @@ At present there is only one virtual LAN and you are dumped there by default. It
 
 **Security warning:** You read that right. ZeroTier One places your computer on an absolutely open global Ethernet party line. Please ensure that you are up to date on your OS patches and we recommend turning off unnecessary services. Also be sure that anything else you are sharing is password protected provided you don't want to share it: printers, iPhoto and iTunes shares, etc.
 
-ZeroTier One is licensed under the GNU General Public License version 3. Anyone wishing to embed this in a commercial product or create a derivative product should contact [ZeroTier Networks LLC](https://www.zerotier.com/) to obtain a commercial license.
+ZeroTier One is licensed under the GNU General Public License version 3. You are free to use, modify, or redistribute it under the terms of that license. If you would like to embed ZeroTier One in a closed source product or create a closed source derivative product, contact ZeroTier Networks LLC.
 
 Check out the [blog](http://blog.zerotier.com/) for announcements, in-depth articles, and related thoughts. There is also a [Google group](https://groups.google.com/forum/#!forum/zerotier-one-users) for questions and discussion.
 
@@ -21,16 +21,19 @@ Check out the [blog](http://blog.zerotier.com/) for announcements, in-depth arti
 **FAQ**
 
 **Q:** What can I do with this?  
-**A:** For starters, try opening iTunes if you have it installed. If others are also online and sharing their collections, you might see them. If you have any games that run over a LAN (except those that require IPX), try those. What else can you think of to do on a completely flat, open network? Collaborative software development? Remote debugging? Transferring files using simple drive shares? Sharing your desktop printer to someone on another continent? Use your imagination.
+**A:** For starters, try opening iTunes if you have it installed. If others are also online and sharing their collections, you might see them. If you have any games that run over a LAN (except those that require IPX), try those. What else can you think of to do on a completely flat, open network? Games? Collaborative software development? Remote debugging? Transferring files using simple drive shares? Sharing your desktop printer to someone on another continent? Use your imagination.
 
 **Q:** Why do I get an IP address in the 27.0.0.0 or 28.0.0.0 range? And why does a lookup claim these addresses belong to the U.S. Department of Defense?  
-**A:** Short answer: because IPv4 needs to die. Long answer: the Earth network assigns IPv4 IPs from these ranges. They do in fact belong to the DOD, but they are *not* routed to the open Internet. The DOD owns them but uses them internally for private networks. As a result, there is nothing *technically* wrong with "bogarting" these for our own private network. It's considered bad practice, but if you want a private address space in IPv4 that is unlikely to overlap other private address spaces (like 10/8 and 192.168/16), it's the only way. [Cellular carriers](http://www.androidcentral.com/sprint-internet-dept-defense-and-you)  and [cable companies](http://www.dslreports/forum/r25679029-Why-is-my-first-hop-to-a-DoD-assigned-IP-address-) frequently do the same thing.
+**A:** Short answer: because IPv4 needs to die. Long answer: the Earth network assigns IPv4 IPs from these ranges. They do in fact belong to the DOD, but they are *not* routed to the open Internet. The DOD owns them but uses them internally for private networks. As a result, there is nothing *technically* wrong with "bogarting" these for our own private network. It's considered bad practice, but if you want a private address space in IPv4 that is unlikely to overlap other private address spaces (like 10/8 and 192.168/16), it's the only way. [Cellular carriers](http://www.androidcentral.com/sprint-internet-dept-defense-and-you)  and [cable companies](http://www.dslreports.com/forum/r25679029-Why-is-my-first-hop-to-a-DoD-assigned-IP-address-) frequently do the same thing.
 
 **Q:** Is IPv6 supported?  
 **A:** Yes. IPv6 link-local addresses (those in the fe80::/10 block) are auto-assigned and should work fine. No other IPv6 addresses are assigned *yet*, but there are plans to do interesting things in this area in the future.
 
+**Q:** I don't want a giant Ethernet party line. Can I leave it and create private LANs instead?  
+**A:** Yes, soon. A GUI to configure such things is in development. But for now there's only Earth.
+
 **Q:** Are you going to charge for this?  
-**A:** Public networks will remain free, but we intend to charge for private virtual LANs. ZeroTier has other ideas too, but they're top secret for the moment.
+**A:** Public virtual LANs will remain free. We intend to charge for private networks in some way, but the exact model is TBD. Other cloud-supported paid features are also TBD.
 
 **Q:** What's a supernode?  
 **A:** Supernodes are nodes run by ZeroTier Networks that orindary users use to find one another and communicate until/unless they can perform NAT traversal and connect directly. They run the exact same software as everyone else. The only thing that really makes a supernode special is that it's designated as such.
@@ -38,14 +41,20 @@ Check out the [blog](http://blog.zerotier.com/) for announcements, in-depth arti
 **Q:** Can I run a supernode?  
 **A:** No, not at the moment, and there would be no benefit to doing so.
 
+**Q:** Will my local firewall rules apply to ZeroTier One traffic?  
+**A:** ZeroTier creates a virtual Ethernet tap device (zt# on Mac and Linux) that emulates a wired Ethernet port. If your firewall applies to *all* network ports, it will filter traffic through this port as well. If it applies only to the primary interface, it may not. See your OS's firewall documentation, as different OSes and flavors thereof have slightly different configurations in this regard.
+
 **Q:** Can you see my traffic? What about other users? Can you sniff the LAN?  
 **A:** No. All unicast (direct computer to computer) traffic is encrypted end-to-end (even if it's being relayed), and the ZeroTier virtual LAN behaves like a LAN with a secure enterprise-grade switch that does not allow unicast conversations to be sniffed. Multicast and broadcast traffic will of course be seen by many recipients, but that's the idea.
 
-**Q:** You say "almost unlimited size." Isn't multicast and broadcast traffic eventually going to be too much? What happens then?  
-**A:** ZeroTier One uses an algorithm called *implicit social switching*. The overall maximum number of recipients for a multicast is limited, so if there are too many listeners to a given multicast address then obviously not everyone will receive every message. So who does? Social switching causes multicasts to propagate in the direction of peers to whom you have recently communicated. As a result, multicasts tend to propagate along lines of association. The people most likely to get your service announcements are those with whom you frequently connect.
+**Q:** What about privacy? Does this hide my location on the network?  
+**A:** ZeroTier is not a connection anonymizer. Other than encryption, it doesn't do anything special to hide your identity or network location. If you want strong privacy protection there are already very advanced tools like [Tor](https://www.torproject.org) for that, and this isn't trying to duplicate their functionality. At the same time, ZT does not do anything special to harm your privacy either. It's not spyware or snoop-ware.
 
-**Q:** I don't see broadcasts.  
-**A:** At the moment only Ethernet multicast is propagated, not broadcast (ff:ff:ff:ff:ff:ff). This may change in the future. IPv4 ARP uses broadcast but is handled by special code that subscribes to a multicast group using the broadcast address combined with the IPv4 address to make it address-specific. See comments in MulticastGroup.hpp for deeper technical details.
+**Q:** Is this designed to replace IP, BGP, IPv6, routers, etc.?  
+**A:** No. Its purpose is to act as a collaboration tool, a VPN alternative, a network mobility tool, a testbed for the development of software that takes advantage of fully open networking, a virtual LAN party for gamers, and so on, but it's not intended (or able) to replace the core of the Internet.
+
+**Q:** Can I bridge this to a physical port and plug in an Xbox, PlayStation, etc.?  
+**A:** Not currently, as foreign Ethernet frames are not forwarded. This may be possible in a future version.
 
 ----
 **Status**
@@ -62,9 +71,8 @@ Check out the [blog](http://blog.zerotier.com/) for announcements, in-depth arti
  * Changes in your local network configuration are generally detected and will cause peers to be re-acquired.
 
 *Known immediate issues:*
- * Multiple network support is in but there is no interface to configure it, hence it is useless. But in alpha it'll be nice to shove everyone onto "Earth" in order to stress test that little "almost unlimited size" boast. LAN party!
-   * Multiple networks would currently all have the same MAC, which some OSes may not like. There is a possible strategy for dealing with this, but it needs to be explored.
- * There is no multiple-launch protection yet. If you launch more than one instance on the same working directory everything breaks. Take care that zerotier-one is not running before launching it again. The command "sudo killall zerotier-one" is helpful.
+ * Multiple network support is in but there is no interface to configure it, hence it is useless. But in alpha it'll be nice to shove everyone onto "Earth" in order to stress test that little "almost unlimited size" boast.
+ * There is no multiple-launch protection yet and multiple instances on the same system do not work well due to route conflicts. Take care that zerotier-one is not running before launching it again. The command "sudo killall zerotier-one" is helpful.
  * Sometimes ZeroTier One doesn't like to terminate when asked nicely. This is related to issues with the tap device closing down properly. If it hangs around after a TERM signal, send it a KILL (9) signal.
  * The locally bound UDP port isn't configurable yet. It's 8993 by default.
  * Known security issues:
@@ -81,4 +89,7 @@ Check out the [blog](http://blog.zerotier.com/) for announcements, in-depth arti
 
 ----
 
+<a href="http://flattr.com/thing/1611614/ZeroTier-Networks" target="_blank"><img src="http://api.flattr.com/button/flattr-badge-large.png" alt="Flattr this" title="Flattr this" border="0" /></a>  
+[![githalytics.com alpha](https://cruel-carlota.pagodabox.com/59b2cbb9c154bf84bddb4b714402e548 "githalytics.com")](http://githalytics.com/zerotier/ZeroTierOne)
+
 (c)2012-2013 [ZeroTier Networks LLC](https://www.zerotier.com/)

RUNNING.txt

@@ -9,7 +9,7 @@ By convention, ZeroTier One will keep its state here on mac:
 ZeroTier ships with a kernel extension for its own tap device, which it
 stores in the above directory. To install this from source, type:
 
-sudo make install-mac-tap
+sudo make -f Makefile.mac install-mac-tap
 
 This will create the ZeroTier One home above if it does not exist and install
 the kext there. Note that the kext must be owned by root:wheel. The make

bump_revision.sh

@@ -1,32 +0,0 @@
-#!/bin/bash
-
-#
-# This probably won't be useful to anyone outside ZeroTier itself.
-#
-
-#
-# This bumps the revision in version.h, which triggers a build and deploy
-# to the now/ subfolder on update.zerotier.com. This allows nodes tracking
-# the bleeding edge to track the bleedingest of the bleeding edge.
-#
-
-cur_rev=`grep -F ZEROTIER_ONE_VERSION_REVISION version.h | cut -d ' ' -f 3`
-next_rev=`expr $cur_rev + 1`
-
-echo Current revision: $cur_rev
-echo Next revision: $next_rev
-
-rm -f version.h.new
-cat version.h | sed "s/ZEROTIER_ONE_VERSION_REVISION $cur_rev/ZEROTIER_ONE_VERSION_REVISION $next_rev/g" >>version.h.new
-
-new_cur_rev=`grep -F ZEROTIER_ONE_VERSION_REVISION version.h.new | cut -d ' ' -f 3`
-
-if [ "$new_cur_rev" = "$next_rev" ]; then
-	mv -f version.h.new version.h
-	echo Done.
-else
-	echo Error: version.h.new updated incorrectly, leaving in place.
-	exit 1
-fi
-
-exit 0

main.cpp

@@ -70,12 +70,18 @@ static void sighandlerQuit(int sig)
 static void sighandlerUsr(int sig)
 {
 }
+static void sighandlerHup(int sig)
+{
+	Node *n = node;
+	if (n)
+		n->updateStatusNow();
+}
 #endif
 
 int main(int argc,char **argv)
 {
 #ifndef _WIN32
-	signal(SIGHUP,SIG_IGN);
+	signal(SIGHUP,&sighandlerHup);
 	signal(SIGPIPE,SIG_IGN);
 	signal(SIGUSR1,&sighandlerUsr);
 	signal(SIGUSR2,&sighandlerUsr);

node/BloomFilter.hpp

@@ -0,0 +1,137 @@
+/*
+ * ZeroTier One - Global Peer to Peer Ethernet
+ * Copyright (C) 2012-2013  ZeroTier Networks LLC
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ * --
+ *
+ * ZeroTier may be used and distributed under the terms of the GPLv3, which
+ * are available at: http://www.gnu.org/licenses/gpl-3.0.html
+ *
+ * If you would like to embed ZeroTier into a commercial application or
+ * redistribute it in a modified binary form, please contact ZeroTier Networks
+ * LLC. Start here: http://www.zerotier.com/
+ */
+
+#ifndef _ZT_BLOOMFILTER_HPP
+#define _ZT_BLOOMFILTER_HPP
+
+#include <string.h>
+#include "Utils.hpp"
+
+namespace ZeroTier {
+
+/**
+ * A simple bit field bloom filter
+ *
+ * This actually isn't a total filter, in that it does not contain a hashing
+ * algorithm. It's up to the caller to hash/sum the items being remembered
+ * so that the distribution of 'n' is random.
+ *
+ * @tparam B Size in BITS, must be a multiple of 8
+ */
+template<unsigned int B>
+class BloomFilter
+{
+public:
+	/**
+	 * Construct an empty filter
+	 */
+	BloomFilter()
+		throw()
+	{
+		memset(_field,0,sizeof(_field));
+	}
+
+	/**
+	 * Construct from a raw filter
+	 *
+	 * @param b Raw filter bits, must be exactly bytes() in length, or NULL to construct empty
+	 */
+	BloomFilter(const void *b)
+		throw()
+	{
+		if (b)
+			memcpy(_field,b,sizeof(_field));
+		else memset(_field,0,sizeof(_field));
+	}
+
+	/**
+	 * @return Size of filter in bits
+	 */
+	static inline unsigned int bits() throw() { return B; }
+
+	/**
+	 * @return Size of filter in bytes
+	 */
+	static inline unsigned int bytes() throw() { return (B / 8); }
+
+	/**
+	 * @return Pointer to portable array of bytes of bytes() length representing filter
+	 */
+	inline const unsigned char *data() const throw() { return _field; }
+
+	/**
+	 * Clear all bits in filter
+	 */
+	inline void clear()
+		throw()
+	{
+		memset(_field,0,sizeof(_field));
+	}
+
+	/**
+	 * @param n Value to set
+	 * @return True if corresponding bit was already set before this operation
+	 */
+	inline bool set(unsigned int n)
+		throw()
+	{
+		n %= B;
+		unsigned char *const x = _field + (n / 8);
+		const unsigned char m = (1 << (n % 8));
+		bool already = ((*x & m));
+		*x |= m;
+		return already;
+	}
+
+	/**
+	 * @param n Value to test
+	 * @return True if value is present is set
+	 */
+	inline bool contains(unsigned int n)
+		throw()
+	{
+		n %= B;
+		return (_field[n / 8] & (1 << (n % 8)));
+	}
+
+	/**
+	 * Clear a random bit in this bloom filter
+	 */
+	inline void decay()
+		throw()
+	{
+		const unsigned int rn = Utils::randomInt<unsigned int>();
+		_field[(rn >> 3) % (B / 8)] &= ~((unsigned char)(1 << (rn & 7)));
+	}
+
+private:
+	unsigned char _field[B / 8];
+};
+
+} // namespace ZeroTier
+
+#endif

node/Buffer.hpp

@@ -240,6 +240,22 @@ public:
 		_l += sizeof(T);
 	}
 
+	/**
+	 * Append a run of bytes
+	 *
+	 * @param c Character value to append
+	 * @param n Number of times to append
+	 * @throws std::out_of_range Attempt to append beyond capacity
+	 */
+	inline void append(unsigned char c,unsigned int n)
+		throw(std::out_of_range)
+	{
+		if ((_l + n) > C)
+			throw std::out_of_range("Buffer: append beyond capacity");
+		for(unsigned int i=0;i<n;++i)
+			_b[_l++] = (char)c;
+	}
+
 	/**
 	 * Append a C-array of bytes
 	 * 
@@ -388,7 +404,7 @@ public:
 		return !(*this < b);
 	}
 
-protected:
+private:
 	unsigned int _l;
 	char ZT_VAR_MAY_ALIAS _b[C];
 };

node/Constants.hpp

@@ -28,10 +28,55 @@
 #ifndef _ZT_CONSTANTS_HPP
 #define _ZT_CONSTANTS_HPP
 
-// Assume these are little-endian, since we don't support old PPC MACs
-// and all newer Mac or Windows systems are either x86_32, x86_64, or
-// ARM in little-endian mode.
-#if defined(__APPLE__) || defined(_WIN32)
+//
+// This include file also auto-detects and canonicalizes some environment
+// information defines:
+//
+// __LINUX__
+// __APPLE__
+// __UNIX_LIKE__ - any "unix like" OS (BSD, posix, etc.)
+// __WINDOWS__
+//
+// Also makes sure __BYTE_ORDER is defined reasonably.
+//
+
+// Canonicalize Linux... is this necessary? Do it anyway to be defensive.
+#if defined(__linux__) || defined(linux) || defined(__LINUX__) || defined(__linux)
+#ifndef __LINUX__
+#define __LINUX__
+#ifndef __UNIX_LIKE__
+#define __UNIX_LIKE__
+#endif
+#endif
+#endif
+
+// TODO: Android is what? Linux technically, but does it define it?
+
+// OSX and iOS are unix-like OSes far as we're concerned
+#ifdef __APPLE__
+#ifndef __UNIX_LIKE__
+#define __UNIX_LIKE__
+#endif
+#endif
+
+// Linux has endian.h
+#ifdef __LINUX__
+#include <endian.h>
+#endif
+
+#if defined(_WIN32) || defined(_WIN64)
+#ifndef __WINDOWS__
+#define __WINDOWS__
+#endif
+#undef __UNIX_LIKE__
+#define ZT_PATH_SEPARATOR '\\'
+#define ZT_PATH_SEPARATOR_S "\\"
+#define ZT_EOL_S "\r\n"
+#endif
+
+// Assume these are little-endian. PPC is not supported for OSX, and ARM
+// runs in little-endian mode for these OS families.
+#if defined(__APPLE__) || defined(__WINDOWS__)
 #undef __BYTE_ORDER
 #undef __LITTLE_ENDIAN
 #undef __BIG_ENDIAN
@@ -40,33 +85,23 @@
 #define __BYTE_ORDER 1234
 #endif
 
-// Linux has endian.h, which should tell us
-#if defined(__linux__) || defined(linux) || defined(__LINUX__) || defined(__linux)
-#include <endian.h>
-#endif
-
-#ifndef __BYTE_ORDER
-error_no_byte_order_defined
-#endif
-
-#ifndef ZT_OSNAME
-error_no_ZT_OSNAME
-#endif
-
-#ifndef ZT_ARCH
-error_no_ZT_ARCH
-#endif
-
-#ifdef _WIN32
-#define ZT_PATH_SEPARATOR '\\'
-#define ZT_PATH_SEPARATOR_S "\\"
-#define ZT_EOL_S "\r\n"
-#else
+#ifdef __UNIX_LIKE__
 #define ZT_PATH_SEPARATOR '/'
 #define ZT_PATH_SEPARATOR_S "/"
 #define ZT_EOL_S "\n"
 #endif
 
+// Error out if required symbols are missing
+#ifndef __BYTE_ORDER
+error_no_byte_order_defined;
+#endif
+#ifndef ZT_OSNAME
+error_no_ZT_OSNAME_defined;
+#endif
+#ifndef ZT_ARCH
+error_no_ZT_ARCH_defined;
+#endif
+
 /**
  * Length of a ZeroTier address in bytes
  */
@@ -207,6 +242,11 @@ error_no_ZT_ARCH
  */
 #define ZT_MULTICAST_DEDUP_HISTORY_EXPIRE 8000
 
+/**
+ * Number of bits to randomly "decay" in bloom filter per hop
+ */
+#define ZT_MULTICAST_BLOOM_FILTER_DECAY_RATE 2
+
 /**
  * Period between announcements of all multicast 'likes' in ms
  *
@@ -253,6 +293,11 @@ error_no_ZT_ARCH
  */
 #define ZT_AUTOCONFIGURE_CHECK_DELAY 15000
 
+/**
+ * Delay between updates of status file in home directory
+ */
+#define ZT_STATUS_OUTPUT_PERIOD 120000
+
 /**
  * Minimum delay in Node service loop
  * 

node/Demarc.cpp

@@ -93,9 +93,9 @@ bool Demarc::bindLocalUdp(unsigned int localPort)
 	uint64_t v4p = ((uint64_t)PORT_TYPE_UDP_SOCKET_V4 << 60) | (uint64_t)localPort;
 	uint64_t v6p = ((uint64_t)PORT_TYPE_UDP_SOCKET_V6 << 60) | (uint64_t)localPort;
 	if ((_ports.count((Port)v4p))||(_ports.count((Port)v6p)))
-		return true;
+		return true; // port already bound
 
-	UdpSocket *v4;
+	UdpSocket *v4 = (UdpSocket *)0;
 	try {
 		DemarcPortObj *v4r = &(_ports[(Port)v4p]);
 		v4r->port = (Port)v4p;
@@ -104,10 +104,10 @@ bool Demarc::bindLocalUdp(unsigned int localPort)
 		v4r->type = PORT_TYPE_UDP_SOCKET_V4;
 	} catch ( ... ) {
 		_ports.erase((Port)v4p);
-		return false;
+		v4 = (UdpSocket *)0;
 	}
 
-	UdpSocket *v6;
+	UdpSocket *v6 = (UdpSocket *)0;
 	try {
 		DemarcPortObj *v6r = &(_ports[(Port)v6p]);
 		v6r->port = (Port)v6p;
@@ -115,13 +115,11 @@ bool Demarc::bindLocalUdp(unsigned int localPort)
 		v6r->obj = v6 = new UdpSocket(localPort,true,&Demarc::_CBudpSocketPacketHandler,v6r);
 		v6r->type = PORT_TYPE_UDP_SOCKET_V6;
 	} catch ( ... ) {
-		delete v4;
-		_ports.erase((Port)v4p);
 		_ports.erase((Port)v6p);
-		return false;
+		v6 = (UdpSocket *)0;
 	}
 
-	return true;
+	return ((v4)||(v6));
 }
 
 Demarc::Port Demarc::pick(const InetAddress &to) const

node/EthernetTap.hpp

@@ -36,14 +36,13 @@
 #include <set>
 #include <string>
 #include <stdexcept>
-#include "Array.hpp"
-#include "Utils.hpp"
-#include "InetAddress.hpp"
-#include "NonCopyable.hpp"
-#include "MAC.hpp"
 #include "Constants.hpp"
+#include "InetAddress.hpp"
+#include "MAC.hpp"
 #include "Mutex.hpp"
 #include "MulticastGroup.hpp"
+#include "Thread.hpp"
+#include "Buffer.hpp"
 
 namespace ZeroTier {
 
@@ -52,21 +51,40 @@ class RuntimeEnvironment;
 /**
  * System ethernet tap device
  */
-class EthernetTap : NonCopyable
+class EthernetTap : protected Thread
 {
 public:
 	/**
 	 * Construct a new TAP device
 	 *
+	 * Handler arguments: arg,from,to,etherType,data
+	 * 
 	 * @param renv Runtime environment
 	 * @param mac MAC address of device
 	 * @param mtu MTU of device
+	 * @param handler Handler function to be called when data is received from the tap
+	 * @param arg First argument to handler function
 	 * @throws std::runtime_error Unable to allocate device
 	 */
-	EthernetTap(const RuntimeEnvironment *renv,const MAC &mac,unsigned int mtu)
+	EthernetTap(
+		const RuntimeEnvironment *renv,
+		const MAC &mac,
+		unsigned int mtu,
+		void (*handler)(void *,const MAC &,const MAC &,unsigned int,const Buffer<4096> &),
+		void *arg)
 		throw(std::runtime_error);
 
-	~EthernetTap();
+	/**
+	 * Close tap and shut down thread
+	 *
+	 * This may block for a few seconds while thread exits.
+	 */
+	virtual ~EthernetTap();
+
+	/**
+	 * Perform OS dependent actions on network configuration change detection
+	 */
+	void whack();
 
 	/**
 	 * @return MAC address of this interface
@@ -132,31 +150,10 @@ public:
 	 */
 	void put(const MAC &from,const MAC &to,unsigned int etherType,const void *data,unsigned int len);
 
-	/**
-	 * Get the next packet from the interface, blocking if none is available.
-	 *
-	 * @param from Filled with MAC address of source (normally our own)
-	 * @param to Filled with MAC address of destination
-	 * @param etherType Filled with Ethernet frame type
-	 * @param buf Buffer to fill (must have room for MTU bytes)
-	 * @return Number of bytes read or 0 if none
-	 */
-	unsigned int get(MAC &from,MAC &to,unsigned int &etherType,void *buf);
-
 	/**
 	 * @return OS-specific device or connection name
 	 */
-	std::string deviceName();
-
-	/**
-	 * @return True if tap is open
-	 */
-	bool open() const;
-
-	/**
-	 * Close this tap, invalidating the object and causing get() to abort
-	 */
-	void close();
+	std::string deviceName() const;
 
 	/**
 	 * Fill or modify a set to contain multicast groups for this device
@@ -164,11 +161,18 @@ public:
 	 * This populates a set or, if already populated, modifies it to contain
 	 * only multicast groups in which this device is interested.
 	 *
+	 * This should always include the blind wildcard MulticastGroup (MAC of
+	 * ff:ff:ff:ff:ff:ff and 0 ADI field).
+	 *
 	 * @param groups Set to modify in place
 	 * @return True if set was changed since last call
 	 */
 	bool updateMulticastGroups(std::set<MulticastGroup> &groups);
 
+protected:
+	virtual void main()
+		throw();
+
 private:
 	const MAC _mac;
 	const unsigned int _mtu;
@@ -178,20 +182,14 @@ private:
 	std::set<InetAddress> _ips;
 	Mutex _ips_m;
 
-#if defined(__APPLE__) || defined(__linux__) || defined(linux) || defined(__LINUX__) || defined(__linux)
+	void (*_handler)(void *,const MAC &,const MAC &,unsigned int,const Buffer<4096> &);
+	void *_arg;
 
+#ifdef __UNIX_LIKE__
 	char _dev[16];
-	unsigned char *_putBuf;
-	unsigned char *_getBuf;
 	int _fd;
-
-	bool _isReading;
-	pthread_t _isReadingThreadId;
-	Mutex _isReading_m;
-
-#elif defined(_WIN32) /* -------------------------------------------------- */
-
-#endif /* ----------------------------------------------------------------- */
+	int _shutdownSignalPipe[2];
+#endif
 };
 
 } // namespace ZeroTier

node/Filter.cpp

@@ -0,0 +1,241 @@
+/*
+ * ZeroTier One - Global Peer to Peer Ethernet
+ * Copyright (C) 2012-2013  ZeroTier Networks LLC
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ * --
+ *
+ * ZeroTier may be used and distributed under the terms of the GPLv3, which
+ * are available at: http://www.gnu.org/licenses/gpl-3.0.html
+ *
+ * If you would like to embed ZeroTier into a commercial application or
+ * redistribute it in a modified binary form, please contact ZeroTier Networks
+ * LLC. Start here: http://www.zerotier.com/
+ */
+
+#include <stdint.h>
+
+#include "RuntimeEnvironment.hpp"
+#include "Logger.hpp"
+#include "Filter.hpp"
+#include "Utils.hpp"
+
+namespace ZeroTier {
+
+bool Filter::Rule::operator()(unsigned int etype,const void *data,unsigned int len) const
+{
+	if ((!_etherType)||(_etherType(etype))) { // ethertype is ANY, or matches
+		// Ethertype determines meaning of protocol and port
+		switch(etype) {
+			default:
+				if ((!_protocol)&&(!_port))
+					return true; // match other ethertypes if protocol and port are ANY, since we don't know what to do with them
+				break;
+
+			case ZT_ETHERTYPE_IPV4:
+				if (len > 20) {
+					if ((!_protocol)||(_protocol(((const uint8_t *)data)[9]))) { // IP protocol
+						if (!_port)
+							return true; // protocol matches or is ANY, port is ANY
+
+						// Don't match on fragments beyond fragment 0. If we've blocked
+						// fragment 0, further fragments will fall on deaf ears anyway.
+						if ((Utils::ntoh(((const uint16_t *)data)[3]) & 0x1fff))
+							return false;
+
+						// Internet header length determines where data begins, in multiples of 32 bits
+						unsigned int ihl = 4 * (((const uint8_t *)data)[0] & 0x0f);
+
+						switch(((const uint8_t *)data)[9]) { // port's meaning depends on IP protocol
+							case ZT_IPPROTO_ICMP:
+								return _port(((const uint8_t *)data)[ihl]); // port = ICMP type
+							case ZT_IPPROTO_TCP:
+							case ZT_IPPROTO_UDP:
+							case ZT_IPPROTO_SCTP:
+							case ZT_IPPROTO_UDPLITE:
+								return _port(((const uint16_t *)data)[(ihl / 2) + 1]); // destination port
+						}
+
+						return false; // no match on port
+					}
+				}
+				break;
+
+			case ZT_ETHERTYPE_IPV6:
+				if (len > 40) {
+					// see: http://stackoverflow.com/questions/17518951/is-the-ipv6-header-really-this-nutty
+					int nextHeader = ((const uint8_t *)data)[6];
+					unsigned int pos = 40;
+					while ((pos < len)&&(nextHeader >= 0)&&(nextHeader != 59)) { // 59 == no next header
+						fprintf(stderr,"[rule] V6: start header parse, header %.2x pos %d\n",nextHeader,pos);
+
+						switch(nextHeader) {
+							case 0: // hop-by-hop options
+							case 60: // destination options
+							case 43: // routing
+							case 135: // mobility (mobile IPv6 options)
+								if (_protocol((unsigned int)nextHeader))
+									return true; // match if our goal was to match any of these
+								nextHeader = ((const uint8_t *)data)[pos];
+								pos += 8 + (8 * ((const uint8_t *)data)[pos + 1]);
+								break;
+							case 44: // fragment
+								if (_protocol(44))
+									return true; // match if our goal was to match fragments
+								nextHeader = ((const uint8_t *)data)[pos];
+								pos += 8;
+								break;
+							case ZT_IPPROTO_AH: // AH
+								return _protocol(ZT_IPPROTO_AH); // true if AH is matched protocol, otherwise false since packet will be IPsec
+							case ZT_IPPROTO_ESP: // ESP
+								return _protocol(ZT_IPPROTO_ESP); // true if ESP is matched protocol, otherwise false since packet will be IPsec
+							case ZT_IPPROTO_ICMPV6:
+								if (_protocol(ZT_IPPROTO_ICMPV6)) { // only match ICMPv6 if specified
+									if ((!_port)||(_port(((const uint8_t *)data)[pos])))
+										return true; // protocol matches, port is ANY or matches ICMP type
+								}
+								break;
+							case ZT_IPPROTO_TCP:
+							case ZT_IPPROTO_UDP:
+							case ZT_IPPROTO_SCTP:
+							case ZT_IPPROTO_UDPLITE:
+								// If we encounter any of these, match if protocol matches or is wildcard as
+								// we'll consider these the "real payload" if present.
+								if ((!_protocol)||(_protocol(nextHeader))) {
+									if ((!_port)||(_port(((const uint16_t *)data)[(pos / 2) + 1])))
+										return true; // protocol matches or is ANY, port is ANY or matches
+								}
+								break;
+						}
+
+						fprintf(stderr,"[rule] V6: end header parse, next header %.2x, new pos %d\n",nextHeader,pos);
+					}
+				}
+				break;
+		}
+	}
+
+	return false;
+}
+
+Filter::Filter(const RuntimeEnvironment *renv) :
+	_r(renv)
+{
+}
+
+Filter::~Filter()
+{
+}
+
+void Filter::add(const Rule &r,const Action &a)
+{
+	Mutex::Lock _l(_chain_m);
+	for(std::vector<Entry>::iterator i(_chain.begin());i!=_chain.end();++i) {
+		if (i->rule == r) {
+			_chain.erase(i);
+			break;
+		}
+	}
+	_chain.push_back(Entry(r,a));
+}
+
+std::string Filter::toString(const char *sep) const
+{
+	char buf[256];
+
+	if (!sep)
+		sep = ",";
+
+	std::string s;
+
+	Mutex::Lock _l(_chain_m);
+	for(std::vector<Entry>::const_iterator i(_chain.begin());i!=_chain.end();++i) {
+		bool first = (i == _chain.begin());
+
+		s.push_back('[');
+
+		if (i->rule.etherType()) {
+			if (i->rule.etherType().magnitude() > 1)
+				sprintf(buf,"%u-%u",i->rule.etherType().start,i->rule.etherType().end);
+			else sprintf(buf,"%u",i->rule.etherType().start);
+			s.append(buf);
+		} else s.push_back('*');
+
+		s.push_back(';');
+
+		if (i->rule.protocol()) {
+			if (i->rule.protocol().magnitude() > 1)
+				sprintf(buf,"%u-%u",i->rule.protocol().start,i->rule.protocol().end);
+			else sprintf(buf,"%u",i->rule.protocol().start);
+			s.append(buf);
+		} else s.push_back('*');
+
+		s.push_back(';');
+
+		if (i->rule.port()) {
+			if (i->rule.port().magnitude() > 1)
+				sprintf(buf,"%u-%u",i->rule.port().start,i->rule.port().end);
+			else sprintf(buf,"%u",i->rule.port().start);
+			s.append(buf);
+		} else s.push_back('*');
+
+		s.append("]:");
+
+		switch(i->action) {
+			case ACTION_DENY:
+				s.append("DENY");
+				break;
+			case ACTION_ALLOW:
+				s.append("ALLOW");
+				break;
+			case ACTION_LOG:
+				s.append("LOG");
+				break;
+		}
+
+		if (!first)
+			s.append(sep);
+	}
+
+	return s;
+}
+
+const char *Filter::etherTypeName(const unsigned int etherType)
+	throw()
+{
+	static char tmp[6];
+	switch(etherType) {
+		case ZT_ETHERTYPE_IPV4:
+			return "IPV4";
+		case ZT_ETHERTYPE_ARP:
+			return "ARP";
+		case ZT_ETHERTYPE_RARP:
+			return "RARP";
+		case ZT_ETHERTYPE_ATALK:
+			return "ATALK";
+		case ZT_ETHERTYPE_AARP:
+			return "AARP";
+		case ZT_ETHERTYPE_IPX_A:
+			return "IPX_A";
+		case ZT_ETHERTYPE_IPX_B:
+			return "IPX_B";
+		case ZT_ETHERTYPE_IPV6:
+			return "IPV6";
+	}
+	sprintf(tmp,"%.4x",etherType);
+	return tmp; // technically not thread safe, but we're only going to see this in debugging if ever
+}
+
+} // namespace ZeroTier

node/Filter.hpp

@@ -0,0 +1,300 @@
+/*
+ * ZeroTier One - Global Peer to Peer Ethernet
+ * Copyright (C) 2012-2013  ZeroTier Networks LLC
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ * --
+ *
+ * ZeroTier may be used and distributed under the terms of the GPLv3, which
+ * are available at: http://www.gnu.org/licenses/gpl-3.0.html
+ *
+ * If you would like to embed ZeroTier into a commercial application or
+ * redistribute it in a modified binary form, please contact ZeroTier Networks
+ * LLC. Start here: http://www.zerotier.com/
+ */
+
+#ifndef _ZT_FILTER_HPP
+#define _ZT_FILTER_HPP
+
+#include <stdio.h>
+#include <string.h>
+#include <string>
+#include <vector>
+#include <utility>
+
+#include "Mutex.hpp"
+#include "Range.hpp"
+
+/* Ethernet frame types that might be relevant to us */
+#define ZT_ETHERTYPE_IPV4 0x0800
+#define ZT_ETHERTYPE_ARP 0x0806
+#define ZT_ETHERTYPE_RARP 0x8035
+#define ZT_ETHERTYPE_ATALK 0x809b
+#define ZT_ETHERTYPE_AARP 0x80f3
+#define ZT_ETHERTYPE_IPX_A 0x8137
+#define ZT_ETHERTYPE_IPX_B 0x8138
+#define ZT_ETHERTYPE_IPV6 0x86dd
+
+/* IP protocols we might care about */
+#define ZT_IPPROTO_ICMP 0x01
+#define ZT_IPPROTO_IGMP 0x02
+#define ZT_IPPROTO_TCP 0x06
+#define ZT_IPPROTO_UDP 0x11
+#define ZT_IPPROTO_GRE 0x2f
+#define ZT_IPPROTO_ESP 0x32
+#define ZT_IPPROTO_AH 0x33
+#define ZT_IPPROTO_ICMPV6 0x3a
+#define ZT_IPPROTO_OSPF 0x59
+#define ZT_IPPROTO_IPIP 0x5e
+#define ZT_IPPROTO_IPCOMP 0x6c
+#define ZT_IPPROTO_L2TP 0x73
+#define ZT_IPPROTO_SCTP 0x84
+#define ZT_IPPROTO_FC 0x85
+#define ZT_IPPROTO_UDPLITE 0x88
+#define ZT_IPPROTO_HIP 0x8b
+
+/* IPv4 ICMP types */
+#define ZT_ICMP_ECHO_REPLY 0
+#define ZT_ICMP_DESTINATION_UNREACHABLE 3
+#define ZT_ICMP_SOURCE_QUENCH 4
+#define ZT_ICMP_REDIRECT 5
+#define ZT_ICMP_ALTERNATE_HOST_ADDRESS 6
+#define ZT_ICMP_ECHO_REQUEST 8
+#define ZT_ICMP_ROUTER_ADVERTISEMENT 9
+#define ZT_ICMP_ROUTER_SOLICITATION 10
+#define ZT_ICMP_TIME_EXCEEDED 11
+#define ZT_ICMP_BAD_IP_HEADER 12
+#define ZT_ICMP_TIMESTAMP 13
+#define ZT_ICMP_TIMESTAMP_REPLY 14
+#define ZT_ICMP_INFORMATION_REQUEST 15
+#define ZT_ICMP_INFORMATION_REPLY 16
+#define ZT_ICMP_ADDRESS_MASK_REQUEST 17
+#define ZT_ICMP_ADDRESS_MASK_REPLY 18
+#define ZT_ICMP_TRACEROUTE 30
+#define ZT_ICMP_MOBILE_HOST_REDIRECT 32
+#define ZT_ICMP_MOBILE_REGISTRATION_REQUEST 35
+#define ZT_ICMP_MOBILE_REGISTRATION_REPLY 36
+
+/* IPv6 ICMP types */
+#define ZT_ICMP6_DESTINATION_UNREACHABLE 1
+#define ZT_ICMP6_PACKET_TOO_BIG 2
+#define ZT_ICMP6_TIME_EXCEEDED 3
+#define ZT_ICMP6_PARAMETER_PROBLEM 4
+#define ZT_ICMP6_ECHO_REQUEST 128
+#define ZT_ICMP6_ECHO_REPLY 129
+#define ZT_ICMP6_MULTICAST_LISTENER_QUERY 130
+#define ZT_ICMP6_MULTICAST_LISTENER_REPORT 131
+#define ZT_ICMP6_MULTICAST_LISTENER_DONE 132
+#define ZT_ICMP6_ROUTER_SOLICITATION 133
+#define ZT_ICMP6_ROUTER_ADVERTISEMENT 134
+#define ZT_ICMP6_NEIGHBOR_SOLICITATION 135
+#define ZT_ICMP6_NEIGHBOR_ADVERTISEMENT 136
+#define ZT_ICMP6_REDIRECT_MESSAGE 137
+#define ZT_ICMP6_ROUTER_RENUMBERING 138
+#define ZT_ICMP6_NODE_INFORMATION_QUERY 139
+#define ZT_ICMP6_NODE_INFORMATION_RESPONSE 140
+#define ZT_ICMP6_INV_NEIGHBOR_SOLICITATION 141
+#define ZT_ICMP6_INV_NEIGHBOR_ADVERTISEMENT 142
+#define ZT_ICMP6_MLDV2 143
+#define ZT_ICMP6_HOME_AGENT_ADDRESS_DISCOVERY_REQUEST 144
+#define ZT_ICMP6_HOME_AGENT_ADDRESS_DISCOVERY_REPLY 145
+#define ZT_ICMP6_MOBILE_PREFIX_SOLICITATION 146
+#define ZT_ICMP6_MOBILE_PREFIX_ADVERTISEMENT 147
+#define ZT_ICMP6_CERTIFICATION_PATH_SOLICITATION 148
+#define ZT_ICMP6_CERTIFICATION_PATH_ADVERTISEMENT 149
+#define ZT_ICMP6_MULTICAST_ROUTER_ADVERTISEMENT 151
+#define ZT_ICMP6_MULTICAST_ROUTER_SOLICITATION 152
+#define ZT_ICMP6_MULTICAST_ROUTER_TERMINATION 153
+#define ZT_ICMP6_RPL_CONTROL_MESSAGE 155
+
+namespace ZeroTier {
+
+class RuntimeEnvironment;
+
+/**
+ * A simple Ethernet frame level filter supporting basic IP port DENY
+ */
+class Filter
+{
+public:
+	/**
+	 * A filter rule
+	 *
+	 * This behaves as an immutable value object.
+	 */
+	class Rule
+	{
+	public:
+		Rule()
+			throw() :
+			_etherType(),
+			_protocol(),
+			_port()
+		{
+		}
+
+		/**
+		 * Construct a new rule
+		 *
+		 * @param etype Ethernet type or empty range for ANY
+		 * @param prot Protocol or empty range for ANY (meaning depends on ethertype, e.g. IP protocol numbers)
+		 * @param prt Port or empty range for ANY (only applies to some protocols)
+		 */
+		Rule(const Range<unsigned int> &etype,const Range<unsigned int> &prot,const Range<unsigned int> &prt)
+			throw() :
+			_etherType(etype),
+			_protocol(prot),
+			_port(prt)
+		{
+		}
+
+		inline const Range<unsigned int> &etherType() const throw() { return _etherType; }
+		inline const Range<unsigned int> &protocol() const throw() { return _protocol; }
+		inline const Range<unsigned int> &port() const throw() { return _port; }
+
+		/**
+		 * Test this rule against a frame
+		 *
+		 * @param etype Type of ethernet frame
+		 * @param data Ethernet frame data
+		 * @param len Length of ethernet frame
+		 * @return True if rule matches
+		 */
+		bool operator()(unsigned int etype,const void *data,unsigned int len) const;
+
+		inline bool operator==(const Rule &r) const throw() { return ((_etherType == r._etherType)&&(_protocol == r._protocol)&&(_port == r._port)); }
+		inline bool operator!=(const Rule &r) const throw() { return !(*this == r); }
+		inline bool operator<(const Rule &r) const
+			throw()
+		{
+			if (_etherType < r._etherType)
+				return true;
+			else if (_etherType == r._etherType) {
+				if (_protocol < r._protocol)
+					return true;
+				else if (_protocol == r._protocol) {
+					if (_port < r._port)
+						return true;
+				}
+			}
+			return false;
+		}
+		inline bool operator>(const Rule &r) const throw() { return (r < *this); }
+		inline bool operator<=(const Rule &r) const throw() { return !(r < *this); }
+		inline bool operator>=(const Rule &r) const throw() { return !(*this < r); }
+
+	private:
+		Range<unsigned int> _etherType;
+		Range<unsigned int> _protocol;
+		Range<unsigned int> _port;
+	};
+
+	/**
+	 * Action if a rule matches
+	 */
+	enum Action
+	{
+		ACTION_DENY = 0,
+		ACTION_ALLOW = 1,
+		ACTION_LOG = 2
+	};
+
+	/**
+	 * Entry in filter chain
+	 */
+	struct Entry
+	{
+		Entry() {}
+		Entry(const Rule &r,const Action &a) :
+			rule(r),
+			action(a)
+		{
+		}
+
+		Rule rule;
+		Action action;
+	};
+
+	Filter(const RuntimeEnvironment *renv);
+	~Filter();
+
+	/**
+	 * Remove all filter entries
+	 */
+	inline void clear()
+	{
+		Mutex::Lock _l(_chain_m);
+		_chain.clear();
+	}
+
+	/**
+	 * Append a rule/action pair to this chain
+	 *
+	 * If an identical rule already exists it is removed and a new entry is
+	 * added to the end with the new action. (Two identical rules with the
+	 * same action wouldn't make sense.)
+	 *
+	 * @param r Rule to add
+	 * @param a Action if rule matches
+	 */
+	void add(const Rule &r,const Action &a);
+
+	/**
+	 * @return Number of rules in filter chain
+	 */
+	inline unsigned int length() const
+		throw()
+	{
+		Mutex::Lock _l(_chain_m);
+		return _chain.size();
+	}
+
+	/**
+	 * @return Entry in filter chain or null entry if out of bounds
+	 */
+	inline Entry operator[](const unsigned int i) const
+		throw()
+	{
+		Mutex::Lock _l(_chain_m);
+		if (i < _chain.size())
+			return _chain[i];
+		return Entry();
+	}
+
+	/**
+	 * Get a string representation of this filter
+	 *
+	 * @param sep Separator between filter rules, or NULL for comma (default)
+	 * @return Human-readable string
+	 */
+	std::string toString(const char *sep = (const char *)0) const;
+
+	/**
+	 * @param etherType Ethernet type ID
+	 * @return Name of Ethernet protocol (e.g. ARP, IPV4)
+	 */
+	static const char *etherTypeName(const unsigned int etherType)
+		throw();
+
+private:
+	const RuntimeEnvironment *_r;
+
+	std::vector<Entry> _chain;
+	Mutex _chain_m;
+};
+
+} // namespace ZeroTier
+
+#endif

node/MulticastGroup.hpp

@@ -36,7 +36,7 @@
 namespace ZeroTier {
 
 /**
- * A multicast group composed of a multicast MAC and a 64-bit ADI field
+ * A multicast group composed of a multicast MAC and a 32-bit ADI field
  *
  * ADI stands for additional distinguishing information. ADI is primarily for
  * adding additional information to broadcast (ff:ff:ff:ff:ff:ff) memberships,

node/Multicaster.hpp

@@ -0,0 +1,274 @@
+/*
+ * ZeroTier One - Global Peer to Peer Ethernet
+ * Copyright (C) 2012-2013  ZeroTier Networks LLC
+ *
+ * This program is free software: you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ *
+ * --
+ *
+ * ZeroTier may be used and distributed under the terms of the GPLv3, which
+ * are available at: http://www.gnu.org/licenses/gpl-3.0.html
+ *
+ * If you would like to embed ZeroTier into a commercial application or
+ * redistribute it in a modified binary form, please contact ZeroTier Networks
+ * LLC. Start here: http://www.zerotier.com/
+ */
+
+#ifndef _ZT_MULTICASTER_HPP
+#define _ZT_MULTICASTER_HPP
+
+#include <stdint.h>
+#include <string.h>
+
+#include <utility>
+#include <algorithm>
+#include <map>
+#include <set>
+#include <vector>
+
+#include "Constants.hpp"
+#include "Buffer.hpp"
+#include "Packet.hpp"
+#include "MulticastGroup.hpp"
+#include "Utils.hpp"
+#include "MAC.hpp"
+#include "Address.hpp"
+#include "SharedPtr.hpp"
+#include "BloomFilter.hpp"
+
+// Maximum sample size to pick during choice of multicast propagation peers
+#define ZT_MULTICAST_PICK_MAX_SAMPLE_SIZE 64
+
+namespace ZeroTier {
+
+/**
+ * Multicast propagation engine
+ *
+ * This is written as a generic class so that it can be mocked and tested
+ * in simulation. It also always takes 'now' as an argument, permitting
+ * running in simulated time.
+ */
+class Multicaster
+{
+public:
+	/**
+	 * 256-bit simple bloom filter included with multicast frame packets
+	 */
+	typedef BloomFilter<ZT_PROTO_VERB_MULTICAST_FRAME_BLOOM_FILTER_SIZE_BITS> MulticastBloomFilter;
+
+	Multicaster()
+	{
+		memset(_multicastHistory,0,sizeof(_multicastHistory));
+	}
+
+	/**
+	 * Update the most recent LIKE time for an address in a given multicast group on a given network
+	 *
+	 * @param nwid Network ID
+	 * @param mg Multicast group
+	 * @param addr Address that likes group on given network
+	 * @param now Current timestamp
+	 */
+	inline void likesMulticastGroup(const uint64_t nwid,const MulticastGroup &mg,const Address &addr,const uint64_t now)
+	{
+		_multicastMemberships[MulticastChannel(nwid,mg)][addr] = now;
+	}
+
+	/**
+	 * Check multicast history to see if this is a duplicate, and add/update entry
+	 *
+	 * @param from Ultimate sending MAC address
+	 * @param to Destination multicast group
+	 * @param payload Multicast packet payload
+	 * @param len Length of packet
+	 * @param nwid Network ID
+	 * @param now Current time
+	 * @return True if this appears to be a duplicate to within history expiration time
+	 */
+	inline bool checkAndUpdateMulticastHistory(
+		const MAC &from,
+		const MulticastGroup &to,
+		const void *payload,
+		unsigned int len,
+		const uint64_t nwid,
+		const uint64_t now)
+		throw()
+	{
+		// Note: CRCs aren't transmitted over the network, so portability and
+		// byte order don't matter. This calculation can be changed. We just
+		// want a unique code.
+		uint64_t crc = Utils::crc64(0,from.data,6);
+		crc = Utils::crc64(crc,to.mac().data,6);
+		crc ^= (uint64_t)to.adi();
+		crc = Utils::crc64(crc,payload,len);
+		crc ^= nwid; // also include network ID in CRC
+
+		// Replace existing entry or pick one to replace with new entry
+		uint64_t earliest = 0xffffffffffffffffULL;
+		unsigned long earliestIdx = 0;
+		for(unsigned int i=0;i<ZT_MULTICAST_DEDUP_HISTORY_LENGTH;++i) {
+			if (_multicastHistory[i][0] == crc) {
+				uint64_t then = _multicastHistory[i][1];
+				_multicastHistory[i][1] = now;
+				return ((now - then) < ZT_MULTICAST_DEDUP_HISTORY_EXPIRE);
+			} else if (_multicastHistory[i][1] < earliest) {
+				earliest = _multicastHistory[i][1];
+				earliestIdx = i;
+			}
+		}
+
+		_multicastHistory[earliestIdx][0] = crc; // replace oldest entry
+		_multicastHistory[earliestIdx][1] = now;
+
+		return false;
+	}
+
+	/**
+	 * Choose peers to send a propagating multicast to
+	 *
+	 * @param topology Topology object or mock thereof
+	 * @param nwid Network ID
+	 * @param mg Multicast group
+	 * @param upstream Address from which message originated, or null (0) address if none
+	 * @param bf Bloom filter, updated in place with sums of addresses in chosen peers and/or decay
+	 * @param max Maximum number of peers to pick
+	 * @param peers Array of objects of type P to fill with up to [max] peers
+	 * @param now Current timestamp
+	 * @return Number of peers actually stored in peers array
+	 * @tparam T Type of topology, which is Topology in running code or a mock in simulation
+	 * @tparam P Type of peers, which is SharedPtr<Peer> in running code or a mock in simulation (mock must behave like a pointer type)
+	 */
+	template<typename T,typename P>
+	inline unsigned int pickNextPropagationPeers(
+		T &topology,
+		uint64_t nwid,
+		const MulticastGroup &mg,
+		const Address &upstream,
+		MulticastBloomFilter &bf,
+		unsigned int max,
+		P *peers,
+		uint64_t now)
+	{
+		P toConsider[ZT_MULTICAST_PICK_MAX_SAMPLE_SIZE];
+		unsigned int sampleSize = 0;
+
+		{
+			Mutex::Lock _l(_multicastMemberships_m);
+
+			// Sample a random subset of peers that we know have LIKEd this multicast
+			// group on this network.
+			std::map< MulticastChannel,std::map<Address,uint64_t> >::iterator channelMembers(_multicastMemberships.find(MulticastChannel(nwid,mg)));
+			if ((channelMembers != _multicastMemberships.end())&&(!channelMembers->second.empty())) {
+				unsigned long numEntriesPermittedToSkip = (channelMembers->second.size() > ZT_MULTICAST_PICK_MAX_SAMPLE_SIZE) ? (unsigned long)(channelMembers->second.size() - ZT_MULTICAST_PICK_MAX_SAMPLE_SIZE) : (unsigned long)0;
+				double skipWhatFraction = (double)numEntriesPermittedToSkip / (double)channelMembers->second.size();
+
+				std::map<Address,uint64_t>::iterator channelMemberEntry(channelMembers->second.begin());
+
+				while (channelMemberEntry != channelMembers->second.end()) {
+					// Auto-clean the channel members map if their LIKEs are expired. This will
+					// technically skew the random distribution of chosen members just a little, but
+					// it's unlikely that enough will expire in any single pick to make much of a
+					// difference overall.
+					if ((now - channelMemberEntry->second) > ZT_MULTICAST_LIKE_EXPIRE) {
+						channelMembers->second.erase(channelMemberEntry++);
+						continue;
+					}
+
+					// Skip some fraction of entries so that our sampling will be randomly distributed,
+					// since there is no other good way to sample randomly from a map.
+					if (numEntriesPermittedToSkip) {
+						double skipThis = (double)(Utils::randomInt<uint32_t>()) / 4294967296.0;
+						if (skipThis <= skipWhatFraction) {
+							--numEntriesPermittedToSkip;
+							++channelMemberEntry;
+							continue;
+						}
+					}
+
+					// If it's not expired and it's from our random sample, add it to the set of peers
+					// to consider.
+					P peer = topology.getPeer(channelMemberEntry->first);
+					if (peer) {
+						toConsider[sampleSize++] = peer;
+						if (sampleSize >= ZT_MULTICAST_PICK_MAX_SAMPLE_SIZE)
+							break; // abort if we have enough candidates
+					}
+					++channelMemberEntry;
+				}
+
+				// Auto-clean: erase whole map if there are no more LIKEs for this channel
+				if (channelMembers->second.empty())
+					_multicastMemberships.erase(channelMembers);
+			}
+		}
+
+		// Sort in descending order of most recent direct unicast frame, picking
+		// peers with whom we have recently communicated. This is "implicit social
+		// switching."
+		std::sort(&(toConsider[0]),&(toConsider[sampleSize]),PeerPropagationPrioritySortOrder<P>());
+
+		// Decay a few random bits in bloom filter to probabilistically eliminate
+		// false positives as we go. The odds of decaying an already-set bit
+		// increases as the bloom filter saturates, so in the early hops of
+		// propagation this likely won't have any effect.
+		for(unsigned int i=0;i<ZT_MULTICAST_BLOOM_FILTER_DECAY_RATE;++i)
+			bf.decay();
+
+		// Pick peers not in the bloom filter, setting bloom filter bits accordingly to
+		// remember and pass on these picks.
+		unsigned int picked = 0;
+		for(unsigned int i=0;((i<sampleSize)&&(picked < max));++i) {
+			if (!bf.set(toConsider[i]->address().sum()))
+				peers[picked++] = toConsider[i];
+		}
+
+		// Add a supernode if there's nowhere else to go. Supernodes know of all multicast
+		// LIKEs and so can act to bridge sparse multicast groups. We do not remember them
+		// in the bloom filter, since such bridging may very well need to happen more than
+		// once.
+		if (!picked) {
+			P peer = topology.getBestSupernode();
+			if (peer)
+				peers[picked++] = peer;
+		}
+
+		return picked;
+	}
+
+private:
+	// Sort order for chosen propagation peers
+	template<typename P>
+	struct PeerPropagationPrioritySortOrder
+	{
+		inline bool operator()(const P &p1,const P &p2) const
+		{
+			return (p1->lastUnicastFrame() >= p2->lastUnicastFrame());
+		}
+	};
+
+	// [0] - CRC, [1] - timestamp
+	uint64_t _multicastHistory[ZT_MULTICAST_DEDUP_HISTORY_LENGTH][2];
+
+	// A multicast channel, essentially a pub/sub channel. It consists of a
+	// network ID and a multicast group within that network.
+	typedef std::pair<uint64_t,MulticastGroup> MulticastChannel;
+
+	// Address and time of last LIKE, by network ID and multicast group
+	std::map< MulticastChannel,std::map<Address,uint64_t> > _multicastMemberships;
+	Mutex _multicastMemberships_m;
+};
+
+} // namespace ZeroTier
+
+#endif

node/Network.cpp

@@ -32,49 +32,29 @@ namespace ZeroTier {
 
 Network::Network(const RuntimeEnvironment *renv,uint64_t id)
 	throw(std::runtime_error) :
-	Thread(),
 	_r(renv),
 	_id(id),
-	_tap(renv,renv->identity.address().toMAC(),ZT_IF_MTU),
+	_tap(renv,renv->identity.address().toMAC(),ZT_IF_MTU,&_CBhandleTapData,this),
 	_members(),
 	_open(false),
 	_lock()
 {
-	TRACE("new network %llu created, TAP device: %s",id,_tap.deviceName().c_str());
-	start();
 }
 
 Network::~Network()
 {
-	_tap.close();
-	join();
-	TRACE("network %llu (%s) closed",_id,_tap.deviceName().c_str());
 }
 
-void Network::main()
-	throw()
+void Network::_CBhandleTapData(void *arg,const MAC &from,const MAC &to,unsigned int etherType,const Buffer<4096> &data)
 {
-	Buffer<4096> buf;
-	MAC from,to;
-	unsigned int etherType = 0;
-
-	while (_tap.open()) {
-		unsigned int len = _tap.get(from,to,etherType,buf.data());
-		if (len) {
-			buf.setSize(len);
-			try {
-				if (!*__refCount)
-					break; // sanity check
-				_r->sw->onLocalEthernet(SharedPtr<Network>(this),from,to,etherType,buf);
-			} catch (std::exception &exc) {
-				TRACE("unexpected exception handling local packet: %s",exc.what());
-			} catch ( ... ) {
-				TRACE("unexpected exception handling local packet");
-			}
-		} else break;
+	const RuntimeEnvironment *_r = ((Network *)arg)->_r;
+	try {
+		_r->sw->onLocalEthernet(SharedPtr<Network>((Network *)arg),from,to,etherType,data);
+	} catch (std::exception &exc) {
+		TRACE("unexpected exception handling local packet: %s",exc.what());
+	} catch ( ... ) {
+		TRACE("unexpected exception handling local packet");
 	}
-
-	TRACE("network %llu thread terminating",_id);
 }
 
 } // namespace ZeroTier

node/Network.hpp

@@ -40,8 +40,9 @@
 #include "SharedPtr.hpp"
 #include "AtomicCounter.hpp"
 #include "RuntimeEnvironment.hpp"
-#include "Thread.hpp"
 #include "MulticastGroup.hpp"
+#include "NonCopyable.hpp"
+#include "MAC.hpp"
 
 namespace ZeroTier {
 
@@ -50,17 +51,17 @@ class NodeConfig;
 /**
  * Local network endpoint
  */
-class Network : protected Thread
+class Network : NonCopyable
 {
 	friend class SharedPtr<Network>;
 	friend class NodeConfig;
 
 private:
-	virtual ~Network();
-
 	Network(const RuntimeEnvironment *renv,uint64_t id)
 		throw(std::runtime_error);
 
+	~Network();
+
 public:
 	/**
 	 * @return Network ID
@@ -111,6 +112,19 @@ public:
 		return ((_open)||(_members.count(addr) > 0));
 	}
 
+	/**
+	 * Shortcut to check open(), whether MAC is ZeroTier, then isMember()
+	 *
+	 * @param mac MAC address to check
+	 * @return True if MAC is allowed
+	 */
+	inline bool isAllowed(const MAC &mac) const
+		throw()
+	{
+		Mutex::Lock _l(_lock);
+		return ((_open)||((mac.isZeroTier())&&(_members.count(Address(mac)) > 0)));
+	}
+
 	/**
 	 * @return True if network is open (no membership required)
 	 */
@@ -141,11 +155,9 @@ public:
 		return _multicastGroups;
 	}
 
-protected:
-	virtual void main()
-		throw();
-
 private:
+	static void _CBhandleTapData(void *arg,const MAC &from,const MAC &to,unsigned int etherType,const Buffer<4096> &data);
+
 	const RuntimeEnvironment *_r;
 	uint64_t _id;
 	EthernetTap _tap;