Patrick Welche <prlw1@cam.ac.uk>

[netbsd-mini2440.git] / external / apache2 / mDNSResponder / dist / mDNSPosix / mDNSPosix.c

/* -*- Mode: C; tab-width: 4 -*-
 *
 * Copyright (c) 2002-2004 Apple Computer, Inc. All rights reserved.
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 * 
 *     http://www.apache.org/licenses/LICENSE-2.0
 * 
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 *
 * Formatting notes:
 * This code follows the "Whitesmiths style" C indentation rules. Plenty of discussion
 * on C indentation can be found on the web, such as <http://www.kafejo.com/komp/1tbs.htm>,
 * but for the sake of brevity here I will say just this: Curly braces are not syntactially
 * part of an "if" statement; they are the beginning and ending markers of a compound statement;
 * therefore common sense dictates that if they are part of a compound statement then they
 * should be indented to the same level as everything else in that compound statement.
 * Indenting curly braces at the same level as the "if" implies that curly braces are
 * part of the "if", which is false. (This is as misleading as people who write "char* x,y;"
 * thinking that variables x and y are both of type "char*" -- and anyone who doesn't
 * understand why variable y is not of type "char*" just proves the point that poor code
 * layout leads people to unfortunate misunderstandings about how the C language really works.)

        Change History (most recent first):

Log: mDNSPosix.c,v $
Revision 1.108  2009/01/25 03:16:46  mkrochma
Added skeleton definition of mDNSPlatformSetLocalARP

Revision 1.107  2009/01/07 08:25:03  mkrochma
Added skeleton definition of mDNSPlatformUpdateProxyList

Revision 1.106  2008/10/22 17:19:57  cheshire
Don't need to define BPF_fd any more (it's now per-interface, not global)

Revision 1.105  2008/10/03 23:34:08  cheshire
Added skeleton definition of mDNSPlatformSendRawPacket

Revision 1.104  2008/09/05 22:16:48  cheshire
<rdar://problem/3988320> Should use randomized source ports and transaction IDs to avoid DNS cache poisoning
Add "UDPSocket *src" parameter in mDNSPlatformSendUDP

Revision 1.103  2007/10/02 19:31:17  cheshire
In ParseDNSServers, should use strncasecmp for case-insensitive compare

Revision 1.102  2007/09/12 19:23:17  cheshire
Get rid of unnecessary mDNSPlatformTCPIsConnected() routine

Revision 1.101  2007/07/20 00:54:23  cheshire
<rdar://problem/4641118> Need separate SCPreferences for per-user .Mac settings

Revision 1.100  2007/07/19 21:45:30  cheshire
Fixed code spacing

Revision 1.99  2007/07/11 02:56:51  cheshire
<rdar://problem/5303807> Register IPv6-only hostname and don't create port mappings for AutoTunnel services
Remove unused mDNSPlatformDefaultRegDomainChanged

Revision 1.98  2007/06/20 01:10:13  cheshire
<rdar://problem/5280520> Sync iPhone changes into main mDNSResponder code

Revision 1.97  2007/04/26 00:35:16  cheshire
<rdar://problem/5140339> uDNS: Domain discovery not working over VPN
Fixes to make sure results update correctly when connectivity changes (e.g. a DNS server
inside the firewall may give answers where a public one gives none, and vice versa.)

Revision 1.96  2007/04/22 20:29:59  cheshire
Fix locking error

Revision 1.95  2007/04/22 20:15:46  cheshire
Add missing parameters for mDNSPosixEventCallback

Revision 1.94  2007/04/17 19:21:29  cheshire
<rdar://problem/5140339> Domain discovery not working over VPN

Revision 1.93  2007/04/16 20:49:40  cheshire
Fix compile errors for mDNSPosix build

Revision 1.92  2007/04/05 20:40:37  cheshire
Remove unused mDNSPlatformTCPGetFlags()

Revision 1.91  2007/03/22 18:31:48  cheshire
Put dst parameter first in mDNSPlatformStrCopy/mDNSPlatformMemCopy, like conventional Posix strcpy/memcpy

Revision 1.90  2007/03/21 00:31:45  cheshire
Remove unnecessary (and unimplemented) platform functions

Revision 1.89  2007/03/20 17:07:15  cheshire
Rename "struct uDNS_TCPSocket_struct" to "TCPSocket", "struct uDNS_UDPSocket_struct" to "UDPSocket"

Revision 1.88  2007/03/07 00:30:18  mkrochma
<rdar://problem/5034370> POSIX: kDNSServiceInterfaceIndexAny not correctly handled
Thanks goes to Aidan Williams of Audinate who did a lot of work in diagnosing this

Revision 1.87  2007/02/08 21:12:28  cheshire
<rdar://problem/4386497> Stop reading /etc/mDNSResponder.conf on every sleep/wake

Revision 1.86  2007/01/05 08:30:52  cheshire
Trim excessive "Log" checkin history from before 2006
(checkin history still available via "cvs log ..." of course)

Revision 1.85  2007/01/04 23:12:20  cheshire
Remove unused mDNSPlatformDefaultBrowseDomainChanged

Revision 1.84  2006/12/22 21:07:35  cheshire
<rdar://problem/4742742> Read *all* DNS keys from keychain,
 not just key for the system-wide default registration domain

Revision 1.83  2006/12/21 00:09:46  cheshire
Use mDNSPlatformMemZero instead of bzero

Revision 1.82  2006/12/19 22:43:55  cheshire
Fix compiler warnings

Revision 1.81  2006/08/14 23:24:46  cheshire
Re-licensed mDNSResponder daemon source code under Apache License, Version 2.0

Revision 1.80  2006/07/22 03:05:33  cheshire
Improve error reporting for socket creation failures

Revision 1.79  2006/07/06 00:02:16  cheshire
<rdar://problem/4472014> Add Private DNS client functionality to mDNSResponder

Revision 1.78  2006/06/28 09:12:22  cheshire
Added debugging message

Revision 1.77  2006/03/19 02:00:11  cheshire
<rdar://problem/4073825> Improve logic for delaying packets after repeated interface transitions

Revision 1.76  2006/01/09 19:29:16  cheshire
<rdar://problem/4403128> Cap number of "sendto failed" messages we allow mDNSResponder to log

Revision 1.75  2006/01/05 22:04:57  cheshire
<rdar://problem/4399479> Log error message when send fails with "operation not permitted"

Revision 1.74  2006/01/05 21:45:27  cheshire
<rdar://problem/4400118> Fix uninitialized structure member in IPv6 code

*/

#include "mDNSEmbeddedAPI.h"           // Defines the interface provided to the client layer above
#include "DNSCommon.h"
#include "mDNSPosix.h"                           // Defines the specific types needed to run mDNS on this platform
#include "dns_sd.h"

#include <assert.h>
#include <stdio.h>
#include <stdlib.h>
#include <errno.h>
#include <string.h>
#include <unistd.h>
#include <syslog.h>
#include <stdarg.h>
#include <fcntl.h>
#include <sys/types.h>
#include <sys/time.h>
#include <sys/socket.h>
#include <sys/uio.h>
#include <sys/select.h>
#include <netinet/in.h>
#include <arpa/inet.h>
#include <time.h>                   // platform support for UTC time

#if USES_NETLINK
#include <asm/types.h>
#include <linux/netlink.h>
#include <linux/rtnetlink.h>
#else // USES_NETLINK
#include <net/route.h>
#include <net/if.h>
#endif // USES_NETLINK

#include "mDNSUNP.h"
#include "GenLinkedList.h"

// ***************************************************************************
// Structures

// We keep a list of client-supplied event sources in PosixEventSource records 
struct PosixEventSource
        {
        mDNSPosixEventCallback          Callback;
        void                                            *Context;
        int                                                     fd;
        struct  PosixEventSource        *Next;
        };
typedef struct PosixEventSource PosixEventSource;

// Context record for interface change callback
struct IfChangeRec
        {
        int     NotifySD;
        mDNS *mDNS;
        };
typedef struct IfChangeRec      IfChangeRec;

// Note that static data is initialized to zero in (modern) C.
static fd_set                   gEventFDs;
static int                              gMaxFD;                                 // largest fd in gEventFDs
static GenLinkedList    gEventSources;                  // linked list of PosixEventSource's
static sigset_t                 gEventSignalSet;                // Signals which event loop listens for
static sigset_t                 gEventSignals;                  // Signals which were received while inside loop

// ***************************************************************************
// Globals (for debugging)

static int num_registered_interfaces = 0;
static int num_pkts_accepted = 0;
static int num_pkts_rejected = 0;

// ***************************************************************************
// Functions

int gMDNSPlatformPosixVerboseLevel = 0;

#define PosixErrorToStatus(errNum) ((errNum) == 0 ? mStatus_NoError : mStatus_UnknownErr)

mDNSlocal void SockAddrTomDNSAddr(const struct sockaddr *const sa, mDNSAddr *ipAddr, mDNSIPPort *ipPort)
        {
        switch (sa->sa_family)
                {
                case AF_INET:
                        {
                        struct sockaddr_in *sin          = (struct sockaddr_in*)sa;
                        ipAddr->type                     = mDNSAddrType_IPv4;
                        ipAddr->ip.v4.NotAnInteger       = sin->sin_addr.s_addr;
                        if (ipPort) ipPort->NotAnInteger = sin->sin_port;
                        break;
                        }

#if HAVE_IPV6
                case AF_INET6:
                        {
                        struct sockaddr_in6 *sin6        = (struct sockaddr_in6*)sa;
#ifndef NOT_HAVE_SA_LEN
                        assert(sin6->sin6_len == sizeof(*sin6));
#endif
                        ipAddr->type                     = mDNSAddrType_IPv6;
                        ipAddr->ip.v6                    = *(mDNSv6Addr*)&sin6->sin6_addr;
                        if (ipPort) ipPort->NotAnInteger = sin6->sin6_port;
                        break;
                        }
#endif

                default:
                        verbosedebugf("SockAddrTomDNSAddr: Uknown address family %d\n", sa->sa_family);
                        ipAddr->type = mDNSAddrType_None;
                        if (ipPort) ipPort->NotAnInteger = 0;
                        break;
                }
        }

#if COMPILER_LIKES_PRAGMA_MARK
#pragma mark ***** Send and Receive
#endif

// mDNS core calls this routine when it needs to send a packet.
mDNSexport mStatus mDNSPlatformSendUDP(const mDNS *const m, const void *const msg, const mDNSu8 *const end,
        mDNSInterfaceID InterfaceID, UDPSocket *src, const mDNSAddr *dst, mDNSIPPort dstPort)
        {
        int                     err = 0;
        struct sockaddr_storage to;
        PosixNetworkInterface * thisIntf = (PosixNetworkInterface *)(InterfaceID);
        int sendingsocket = -1;

        (void)src;      // Will need to use this parameter once we implement mDNSPlatformUDPSocket/mDNSPlatformUDPClose

        assert(m != NULL);
        assert(msg != NULL);
        assert(end != NULL);
        assert((((char *) end) - ((char *) msg)) > 0);
        assert(dstPort.NotAnInteger != 0);

        if (dst->type == mDNSAddrType_IPv4)
                {
                struct sockaddr_in *sin = (struct sockaddr_in*)&to;
#ifndef NOT_HAVE_SA_LEN
                sin->sin_len            = sizeof(*sin);
#endif
                sin->sin_family         = AF_INET;
                sin->sin_port           = dstPort.NotAnInteger;
                sin->sin_addr.s_addr    = dst->ip.v4.NotAnInteger;
                sendingsocket           = thisIntf ? thisIntf->multicastSocket4 : m->p->unicastSocket4;
                }

#if HAVE_IPV6
        else if (dst->type == mDNSAddrType_IPv6)
                {
                struct sockaddr_in6 *sin6 = (struct sockaddr_in6*)&to;
                mDNSPlatformMemZero(sin6, sizeof(*sin6));
#ifndef NOT_HAVE_SA_LEN
                sin6->sin6_len            = sizeof(*sin6);
#endif
                sin6->sin6_family         = AF_INET6;
                sin6->sin6_port           = dstPort.NotAnInteger;
                sin6->sin6_addr           = *(struct in6_addr*)&dst->ip.v6;
                sendingsocket             = thisIntf ? thisIntf->multicastSocket6 : m->p->unicastSocket6;
                }
#endif

        if (sendingsocket >= 0)
                err = sendto(sendingsocket, msg, (char*)end - (char*)msg, 0, (struct sockaddr *)&to, GET_SA_LEN(to));

        if      (err > 0) err = 0;
        else if (err < 0)
                {
                static int MessageCount = 0;
        // Don't report EHOSTDOWN (i.e. ARP failure), ENETDOWN, or no route to host for unicast destinations
                if (!mDNSAddressIsAllDNSLinkGroup(dst))
                        if (errno == EHOSTDOWN || errno == ENETDOWN || errno == EHOSTUNREACH || errno == ENETUNREACH) return(mStatus_TransientErr);

                if (MessageCount < 1000)
                        {
                        MessageCount++;
                        if (thisIntf)
                                LogMsg("mDNSPlatformSendUDP got error %d (%s) sending packet to %#a on interface %#a/%s/%d",
                                                          errno, strerror(errno), dst, &thisIntf->coreIntf.ip, thisIntf->intfName, thisIntf->index);
                        else
                                LogMsg("mDNSPlatformSendUDP got error %d (%s) sending packet to %#a", errno, strerror(errno), dst);
                        }
                }

        return PosixErrorToStatus(err);
        }

// This routine is called when the main loop detects that data is available on a socket.
mDNSlocal void SocketDataReady(mDNS *const m, PosixNetworkInterface *intf, int skt)
        {
        mDNSAddr   senderAddr, destAddr;
        mDNSIPPort senderPort;
        ssize_t                 packetLen;
        DNSMessage              packet;
        struct my_in_pktinfo    packetInfo;
        struct sockaddr_storage from;
        socklen_t               fromLen;
        int                     flags;
        mDNSu8                                  ttl;
        mDNSBool                reject;
        const mDNSInterfaceID InterfaceID = intf ? intf->coreIntf.InterfaceID : NULL;

        assert(m    != NULL);
        assert(skt  >= 0);

        fromLen = sizeof(from);
        flags   = 0;
        packetLen = recvfrom_flags(skt, &packet, sizeof(packet), &flags, (struct sockaddr *) &from, &fromLen, &packetInfo, &ttl);

        if (packetLen >= 0)
                {
                SockAddrTomDNSAddr((struct sockaddr*)&from, &senderAddr, &senderPort);
                SockAddrTomDNSAddr((struct sockaddr*)&packetInfo.ipi_addr, &destAddr, NULL);

                // If we have broken IP_RECVDSTADDR functionality (so far
                // I've only seen this on OpenBSD) then apply a hack to
                // convince mDNS Core that this isn't a spoof packet.
                // Basically what we do is check to see whether the
                // packet arrived as a multicast and, if so, set its
                // destAddr to the mDNS address.
                //
                // I must admit that I could just be doing something
                // wrong on OpenBSD and hence triggering this problem
                // but I'm at a loss as to how.
                //
                // If this platform doesn't have IP_PKTINFO or IP_RECVDSTADDR, then we have
                // no way to tell the destination address or interface this packet arrived on,
                // so all we can do is just assume it's a multicast

                #if HAVE_BROKEN_RECVDSTADDR || (!defined(IP_PKTINFO) && !defined(IP_RECVDSTADDR))
                        if ((destAddr.NotAnInteger == 0) && (flags & MSG_MCAST))
                                {
                                destAddr.type = senderAddr.type;
                                if      (senderAddr.type == mDNSAddrType_IPv4) destAddr.ip.v4 = AllDNSLinkGroup_v4.ip.v4;
                                else if (senderAddr.type == mDNSAddrType_IPv6) destAddr.ip.v6 = AllDNSLinkGroup_v6.ip.v6;
                                }
                #endif

                // We only accept the packet if the interface on which it came
                // in matches the interface associated with this socket.
                // We do this match by name or by index, depending on which
                // information is available.  recvfrom_flags sets the name
                // to "" if the name isn't available, or the index to -1
                // if the index is available.  This accomodates the various
                // different capabilities of our target platforms.

                reject = mDNSfalse;
                if (!intf)
                        {
                        // Ignore multicasts accidentally delivered to our unicast receiving socket
                        if (mDNSAddrIsDNSMulticast(&destAddr)) packetLen = -1;
                        }
                else
                        {
                        if      (packetInfo.ipi_ifname[0] != 0) reject = (strcmp(packetInfo.ipi_ifname, intf->intfName) != 0);
                        else if (packetInfo.ipi_ifindex != -1)  reject = (packetInfo.ipi_ifindex != intf->index);
        
                        if (reject)
                                {
                                verbosedebugf("SocketDataReady ignored a packet from %#a to %#a on interface %s/%d expecting %#a/%s/%d/%d",
                                        &senderAddr, &destAddr, packetInfo.ipi_ifname, packetInfo.ipi_ifindex,
                                        &intf->coreIntf.ip, intf->intfName, intf->index, skt);
                                packetLen = -1;
                                num_pkts_rejected++;
                                if (num_pkts_rejected > (num_pkts_accepted + 1) * (num_registered_interfaces + 1) * 2)
                                        {
                                        fprintf(stderr,
                                                "*** WARNING: Received %d packets; Accepted %d packets; Rejected %d packets because of interface mismatch\n",
                                                num_pkts_accepted + num_pkts_rejected, num_pkts_accepted, num_pkts_rejected);
                                        num_pkts_accepted = 0;
                                        num_pkts_rejected = 0;
                                        }
                                }
                        else
                                {
                                verbosedebugf("SocketDataReady got a packet from %#a to %#a on interface %#a/%s/%d/%d",
                                        &senderAddr, &destAddr, &intf->coreIntf.ip, intf->intfName, intf->index, skt);
                                num_pkts_accepted++;
                                }
                        }
                }

        if (packetLen >= 0)
                mDNSCoreReceive(m, &packet, (mDNSu8 *)&packet + packetLen,
                        &senderAddr, senderPort, &destAddr, MulticastDNSPort, InterfaceID);
        }

mDNSexport TCPSocket *mDNSPlatformTCPSocket(mDNS * const m, TCPSocketFlags flags, mDNSIPPort * port)
        {
        (void)m;                        // Unused
        (void)flags;            // Unused
        (void)port;                     // Unused
        return NULL;
        }

mDNSexport TCPSocket *mDNSPlatformTCPAccept(TCPSocketFlags flags, int sd)
        {
        (void)flags;            // Unused
        (void)sd;                       // Unused
        return NULL;
        }

mDNSexport int mDNSPlatformTCPGetFD(TCPSocket *sock)
        {
        (void)sock;                     // Unused
        return -1;
        }

mDNSexport mStatus mDNSPlatformTCPConnect(TCPSocket *sock, const mDNSAddr *dst, mDNSOpaque16 dstport, mDNSInterfaceID InterfaceID,
                                                                                  TCPConnectionCallback callback, void *context)
        {
        (void)sock;                     // Unused
        (void)dst;                      // Unused
        (void)dstport;          // Unused
        (void)InterfaceID;      // Unused
        (void)callback;         // Unused
        (void)context;          // Unused
        return(mStatus_UnsupportedErr);
        }

mDNSexport void mDNSPlatformTCPCloseConnection(TCPSocket *sock)
        {
        (void)sock;                     // Unused
        }

mDNSexport long mDNSPlatformReadTCP(TCPSocket *sock, void *buf, unsigned long buflen, mDNSBool * closed)
        {
        (void)sock;                     // Unused
        (void)buf;                      // Unused
        (void)buflen;           // Unused
        (void)closed;           // Unused
        return 0;                       
        }

mDNSexport long mDNSPlatformWriteTCP(TCPSocket *sock, const char *msg, unsigned long len)
        {
        (void)sock;                     // Unused
        (void)msg;                      // Unused
        (void)len;                      // Unused
        return 0;
        }

mDNSexport UDPSocket *mDNSPlatformUDPSocket(mDNS * const m, mDNSIPPort port)
        {
        (void)m;                        // Unused
        (void)port;                     // Unused
        return NULL;
        }

mDNSexport void           mDNSPlatformUDPClose(UDPSocket *sock)
        {
        (void)sock;                     // Unused
        }
        
mDNSexport void mDNSPlatformUpdateProxyList(mDNS *const m, const mDNSInterfaceID InterfaceID)
        {
        (void)m;                        // Unused
        (void)InterfaceID;                      // Unused
        }

mDNSexport void mDNSPlatformSendRawPacket(const void *const msg, const mDNSu8 *const end, mDNSInterfaceID InterfaceID)
        {
        (void)msg;                      // Unused
        (void)end;                      // Unused
        (void)InterfaceID;                      // Unused
        }
        
mDNSexport void mDNSPlatformSetLocalARP(const mDNSv4Addr *const tpa, const mDNSEthAddr *const tha, mDNSInterfaceID InterfaceID)
        {
        (void)tpa;                      // Unused
        (void)tha;                      // Unused
        (void)InterfaceID;                      // Unused
        }       

mDNSexport mStatus mDNSPlatformTLSSetupCerts(void)
        {
        return(mStatus_UnsupportedErr);
        }
        
mDNSexport void mDNSPlatformTLSTearDownCerts(void)
        {
        }

#if COMPILER_LIKES_PRAGMA_MARK
#pragma mark ***** DDNS Config Platform Functions
#endif

mDNSexport void mDNSPlatformSetDNSConfig(mDNS *const m, mDNSBool setservers, mDNSBool setsearch, domainname *const fqdn, DNameListElem **RegDomains, DNameListElem **BrowseDomains)
        {
        (void) m;
        (void) setservers;
        (void) fqdn;
        (void) setsearch;
        (void) RegDomains;
        (void) BrowseDomains;
        }

mDNSexport mStatus mDNSPlatformGetPrimaryInterface(mDNS * const m, mDNSAddr * v4, mDNSAddr * v6, mDNSAddr * router)
        {
        (void) m;
        (void) v4;
        (void) v6;
        (void) router;

        return mStatus_UnsupportedErr;
        }

mDNSexport void mDNSPlatformDynDNSHostNameStatusChanged(const domainname *const dname, const mStatus status)
        {
        (void) dname;
        (void) status;
        }

#if COMPILER_LIKES_PRAGMA_MARK
#pragma mark ***** Init and Term
#endif

// This gets the current hostname, truncating it at the first dot if necessary
mDNSlocal void GetUserSpecifiedRFC1034ComputerName(domainlabel *const namelabel)
        {
        int len = 0;
        gethostname((char *)(&namelabel->c[1]), MAX_DOMAIN_LABEL);
        while (len < MAX_DOMAIN_LABEL && namelabel->c[len+1] && namelabel->c[len+1] != '.') len++;
        namelabel->c[0] = len;
        }

// On OS X this gets the text of the field labelled "Computer Name" in the Sharing Prefs Control Panel
// Other platforms can either get the information from the appropriate place,
// or they can alternatively just require all registering services to provide an explicit name
mDNSlocal void GetUserSpecifiedFriendlyComputerName(domainlabel *const namelabel)
        {
        // On Unix we have no better name than the host name, so we just use that.
        GetUserSpecifiedRFC1034ComputerName(namelabel);
        }

mDNSexport int ParseDNSServers(mDNS *m, const char *filePath)
        {
        char line[256];
        char nameserver[16];
        char keyword[11];
        int  numOfServers = 0;
        FILE *fp = fopen(filePath, "r");
        if (fp == NULL) return -1;
        while (fgets(line,sizeof(line),fp))
                {
                struct in_addr ina;
                line[255]='\0';         // just to be safe
                if (sscanf(line,"%10s %15s", keyword, nameserver) != 2) continue;       // it will skip whitespaces
                if (strncasecmp(keyword,"nameserver",10)) continue;
                if (inet_aton(nameserver, (struct in_addr *)&ina) != 0)
                        {
                        mDNSAddr DNSAddr;
                        DNSAddr.type = mDNSAddrType_IPv4;
                        DNSAddr.ip.v4.NotAnInteger = ina.s_addr;
                        mDNS_AddDNSServer(m, NULL, mDNSInterface_Any, &DNSAddr, UnicastDNSPort);
                        numOfServers++;
                        }
                }  
        fclose(fp);
        return (numOfServers > 0) ? 0 : -1;
        }

// Searches the interface list looking for the named interface.
// Returns a pointer to if it found, or NULL otherwise.
mDNSlocal PosixNetworkInterface *SearchForInterfaceByName(mDNS *const m, const char *intfName)
        {
        PosixNetworkInterface *intf;

        assert(m != NULL);
        assert(intfName != NULL);

        intf = (PosixNetworkInterface*)(m->HostInterfaces);
        while ((intf != NULL) && (strcmp(intf->intfName, intfName) != 0))
                intf = (PosixNetworkInterface *)(intf->coreIntf.next);

        return intf;
        }

mDNSexport mDNSInterfaceID mDNSPlatformInterfaceIDfromInterfaceIndex(mDNS *const m, mDNSu32 index)
        {
        PosixNetworkInterface *intf;

        assert(m != NULL);

        if (index == kDNSServiceInterfaceIndexLocalOnly) return(mDNSInterface_LocalOnly);
        if (index == kDNSServiceInterfaceIndexAny      ) return(mDNSInterface_Any);

        intf = (PosixNetworkInterface*)(m->HostInterfaces);
        while ((intf != NULL) && (mDNSu32) intf->index != index) 
                intf = (PosixNetworkInterface *)(intf->coreIntf.next);

        return (mDNSInterfaceID) intf;
        }
        
mDNSexport mDNSu32 mDNSPlatformInterfaceIndexfromInterfaceID(mDNS *const m, mDNSInterfaceID id)
        {
        PosixNetworkInterface *intf;

        assert(m != NULL);

        if (id == mDNSInterface_LocalOnly) return(kDNSServiceInterfaceIndexLocalOnly);
        if (id == mDNSInterface_Any      ) return(kDNSServiceInterfaceIndexAny);

        intf = (PosixNetworkInterface*)(m->HostInterfaces);
        while ((intf != NULL) && (mDNSInterfaceID) intf != id)
                intf = (PosixNetworkInterface *)(intf->coreIntf.next);

        return intf ? intf->index : 0;
        }

// Frees the specified PosixNetworkInterface structure. The underlying
// interface must have already been deregistered with the mDNS core.
mDNSlocal void FreePosixNetworkInterface(PosixNetworkInterface *intf)
        {
        assert(intf != NULL);
        if (intf->intfName != NULL)        free((void *)intf->intfName);
        if (intf->multicastSocket4 != -1) assert(close(intf->multicastSocket4) == 0);
#if HAVE_IPV6
        if (intf->multicastSocket6 != -1) assert(close(intf->multicastSocket6) == 0);
#endif
        free(intf);
        }

// Grab the first interface, deregister it, free it, and repeat until done.
mDNSlocal void ClearInterfaceList(mDNS *const m)
        {
        assert(m != NULL);

        while (m->HostInterfaces)
                {
                PosixNetworkInterface *intf = (PosixNetworkInterface*)(m->HostInterfaces);
                mDNS_DeregisterInterface(m, &intf->coreIntf, mDNSfalse);
                if (gMDNSPlatformPosixVerboseLevel > 0) fprintf(stderr, "Deregistered interface %s\n", intf->intfName);
                FreePosixNetworkInterface(intf);
                }
        num_registered_interfaces = 0;
        num_pkts_accepted = 0;
        num_pkts_rejected = 0;
        }

// Sets up a send/receive socket.
// If mDNSIPPort port is non-zero, then it's a multicast socket on the specified interface
// If mDNSIPPort port is zero, then it's a randomly assigned port number, used for sending unicast queries
mDNSlocal int SetupSocket(struct sockaddr *intfAddr, mDNSIPPort port, int interfaceIndex, int *sktPtr)
        {
        int err = 0;
        static const int kOn = 1;
        static const int kIntTwoFiveFive = 255;
        static const unsigned char kByteTwoFiveFive = 255;
        const mDNSBool JoinMulticastGroup = (port.NotAnInteger != 0);
        
        (void) interfaceIndex;  // This parameter unused on plaforms that don't have IPv6
        assert(intfAddr != NULL);
        assert(sktPtr != NULL);
        assert(*sktPtr == -1);

        // Open the socket...
        if      (intfAddr->sa_family == AF_INET) *sktPtr = socket(PF_INET,  SOCK_DGRAM, IPPROTO_UDP);
#if HAVE_IPV6
        else if (intfAddr->sa_family == AF_INET6) *sktPtr = socket(PF_INET6, SOCK_DGRAM, IPPROTO_UDP);
#endif
        else return EINVAL;

        if (*sktPtr < 0) { err = errno; perror((intfAddr->sa_family == AF_INET) ? "socket AF_INET" : "socket AF_INET6"); }

        // ... with a shared UDP port, if it's for multicast receiving
        if (err == 0 && port.NotAnInteger)
                {
                #if defined(SO_REUSEPORT)
                        err = setsockopt(*sktPtr, SOL_SOCKET, SO_REUSEPORT, &kOn, sizeof(kOn));
                #elif defined(SO_REUSEADDR)
                        err = setsockopt(*sktPtr, SOL_SOCKET, SO_REUSEADDR, &kOn, sizeof(kOn));
                #else
                        #error This platform has no way to avoid address busy errors on multicast.
                #endif
                if (err < 0) { err = errno; perror("setsockopt - SO_REUSExxxx"); }
                }

        // We want to receive destination addresses and interface identifiers.
        if (intfAddr->sa_family == AF_INET)
                {
                struct ip_mreq imr;
                struct sockaddr_in bindAddr;
                if (err == 0)
                        {
                        #if defined(IP_PKTINFO)                                                                 // Linux
                                err = setsockopt(*sktPtr, IPPROTO_IP, IP_PKTINFO, &kOn, sizeof(kOn));
                                if (err < 0) { err = errno; perror("setsockopt - IP_PKTINFO"); }
                        #elif defined(IP_RECVDSTADDR) || defined(IP_RECVIF)             // BSD and Solaris
                                #if defined(IP_RECVDSTADDR)
                                        err = setsockopt(*sktPtr, IPPROTO_IP, IP_RECVDSTADDR, &kOn, sizeof(kOn));
                                        if (err < 0) { err = errno; perror("setsockopt - IP_RECVDSTADDR"); }
                                #endif
                                #if defined(IP_RECVIF)
                                        if (err == 0)
                                                {
                                                err = setsockopt(*sktPtr, IPPROTO_IP, IP_RECVIF, &kOn, sizeof(kOn));
                                                if (err < 0) { err = errno; perror("setsockopt - IP_RECVIF"); }
                                                }
                                #endif
                        #else
                                #warning This platform has no way to get the destination interface information -- will only work for single-homed hosts
                        #endif
                        }
        #if defined(IP_RECVTTL)                                                                 // Linux
                if (err == 0)
                        {
                        setsockopt(*sktPtr, IPPROTO_IP, IP_RECVTTL, &kOn, sizeof(kOn));
                        // We no longer depend on being able to get the received TTL, so don't worry if the option fails
                        }
        #endif

                // Add multicast group membership on this interface
                if (err == 0 && JoinMulticastGroup)
                        {
                        imr.imr_multiaddr.s_addr = AllDNSLinkGroup_v4.ip.v4.NotAnInteger;
                        imr.imr_interface        = ((struct sockaddr_in*)intfAddr)->sin_addr;
                        err = setsockopt(*sktPtr, IPPROTO_IP, IP_ADD_MEMBERSHIP, &imr, sizeof(imr));
                        if (err < 0) { err = errno; perror("setsockopt - IP_ADD_MEMBERSHIP"); }
                        }

                // Specify outgoing interface too
                if (err == 0 && JoinMulticastGroup)
                        {
                        err = setsockopt(*sktPtr, IPPROTO_IP, IP_MULTICAST_IF, &((struct sockaddr_in*)intfAddr)->sin_addr, sizeof(struct in_addr));
                        if (err < 0) { err = errno; perror("setsockopt - IP_MULTICAST_IF"); }
                        }

                // Per the mDNS spec, send unicast packets with TTL 255
                if (err == 0)
                        {
                        err = setsockopt(*sktPtr, IPPROTO_IP, IP_TTL, &kIntTwoFiveFive, sizeof(kIntTwoFiveFive));
                        if (err < 0) { err = errno; perror("setsockopt - IP_TTL"); }
                        }

                // and multicast packets with TTL 255 too
                // There's some debate as to whether IP_MULTICAST_TTL is an int or a byte so we just try both.
                if (err == 0)
                        {
                        err = setsockopt(*sktPtr, IPPROTO_IP, IP_MULTICAST_TTL, &kByteTwoFiveFive, sizeof(kByteTwoFiveFive));
                        if (err < 0 && errno == EINVAL)
                                err = setsockopt(*sktPtr, IPPROTO_IP, IP_MULTICAST_TTL, &kIntTwoFiveFive, sizeof(kIntTwoFiveFive));
                        if (err < 0) { err = errno; perror("setsockopt - IP_MULTICAST_TTL"); }
                        }

                // And start listening for packets
                if (err == 0)
                        {
                        bindAddr.sin_family      = AF_INET;
                        bindAddr.sin_port        = port.NotAnInteger;
                        bindAddr.sin_addr.s_addr = INADDR_ANY; // Want to receive multicasts AND unicasts on this socket
                        err = bind(*sktPtr, (struct sockaddr *) &bindAddr, sizeof(bindAddr));
                        if (err < 0) { err = errno; perror("bind"); fflush(stderr); }
                        }
                } // endif (intfAddr->sa_family == AF_INET)

#if HAVE_IPV6
        else if (intfAddr->sa_family == AF_INET6)
                {
                struct ipv6_mreq imr6;
                struct sockaddr_in6 bindAddr6;
        #if defined(IPV6_RECVPKTINFO)
                if (err == 0)
                        {
                                err = setsockopt(*sktPtr, IPPROTO_IPV6, IPV6_RECVPKTINFO, &kOn, sizeof(kOn));
                                if (err < 0) { err = errno; perror("setsockopt - IPV6_RECVPKTINFO"); }
                        }
#elif defined(IPV6_PKTINFO)
                if (err == 0)
                        {
                                err = setsockopt(*sktPtr, IPPROTO_IPV6, IPV6_PKTINFO, &kOn, sizeof(kOn));
                                if (err < 0) { err = errno; perror("setsockopt - IPV6_PKTINFO"); }
                        }
        #else
                #warning This platform has no way to get the destination interface information for IPv6 -- will only work for single-homed hosts
        #endif
        #if defined(IPV6_RECVHOPLIMIT)
                if (err == 0)
                        {
                                err = setsockopt(*sktPtr, IPPROTO_IPV6, IPV6_RECVHOPLIMIT, &kOn, sizeof(kOn));
                                if (err < 0) { err = errno; perror("setsockopt - IPV6_RECVHOPLIMIT"); }
                        }
        #elif defined(IPV6_HOPLIMIT)
                if (err == 0)
                        {
                                err = setsockopt(*sktPtr, IPPROTO_IPV6, IPV6_HOPLIMIT, &kOn, sizeof(kOn));
                                if (err < 0) { err = errno; perror("setsockopt - IPV6_HOPLIMIT"); }
                        }
        #endif

                // Add multicast group membership on this interface
                if (err == 0 && JoinMulticastGroup)
                        {
                        imr6.ipv6mr_multiaddr       = *(const struct in6_addr*)&AllDNSLinkGroup_v6.ip.v6;
                        imr6.ipv6mr_interface       = interfaceIndex;
                        //LogMsg("Joining %.16a on %d", &imr6.ipv6mr_multiaddr, imr6.ipv6mr_interface);
                        err = setsockopt(*sktPtr, IPPROTO_IPV6, IPV6_JOIN_GROUP, &imr6, sizeof(imr6));
                        if (err < 0)
                                {
                                err = errno;
                                verbosedebugf("IPV6_JOIN_GROUP %.16a on %d failed.\n", &imr6.ipv6mr_multiaddr, imr6.ipv6mr_interface);
                                perror("setsockopt - IPV6_JOIN_GROUP");
                                }
                        }

                // Specify outgoing interface too
                if (err == 0 && JoinMulticastGroup)
                        {
                        u_int   multicast_if = interfaceIndex;
                        err = setsockopt(*sktPtr, IPPROTO_IPV6, IPV6_MULTICAST_IF, &multicast_if, sizeof(multicast_if));
                        if (err < 0) { err = errno; perror("setsockopt - IPV6_MULTICAST_IF"); }
                        }

                // We want to receive only IPv6 packets on this socket.
                // Without this option, we may get IPv4 addresses as mapped addresses.
                if (err == 0)
                        {
                        err = setsockopt(*sktPtr, IPPROTO_IPV6, IPV6_V6ONLY, &kOn, sizeof(kOn));
                        if (err < 0) { err = errno; perror("setsockopt - IPV6_V6ONLY"); }
                        }

                // Per the mDNS spec, send unicast packets with TTL 255
                if (err == 0)
                        {
                        err = setsockopt(*sktPtr, IPPROTO_IPV6, IPV6_UNICAST_HOPS, &kIntTwoFiveFive, sizeof(kIntTwoFiveFive));
                        if (err < 0) { err = errno; perror("setsockopt - IPV6_UNICAST_HOPS"); }
                        }

                // and multicast packets with TTL 255 too
                // There's some debate as to whether IPV6_MULTICAST_HOPS is an int or a byte so we just try both.
                if (err == 0)
                        {
                        err = setsockopt(*sktPtr, IPPROTO_IPV6, IPV6_MULTICAST_HOPS, &kByteTwoFiveFive, sizeof(kByteTwoFiveFive));
                        if (err < 0 && errno == EINVAL)
                                err = setsockopt(*sktPtr, IPPROTO_IPV6, IPV6_MULTICAST_HOPS, &kIntTwoFiveFive, sizeof(kIntTwoFiveFive));
                        if (err < 0) { err = errno; perror("setsockopt - IPV6_MULTICAST_HOPS"); }
                        }

                // And start listening for packets
                if (err == 0)
                        {
                        mDNSPlatformMemZero(&bindAddr6, sizeof(bindAddr6));
#ifndef NOT_HAVE_SA_LEN
                        bindAddr6.sin6_len         = sizeof(bindAddr6);
#endif
                        bindAddr6.sin6_family      = AF_INET6;
                        bindAddr6.sin6_port        = port.NotAnInteger;
                        bindAddr6.sin6_flowinfo    = 0;
                        bindAddr6.sin6_addr        = in6addr_any; // Want to receive multicasts AND unicasts on this socket
                        bindAddr6.sin6_scope_id    = 0;
                        err = bind(*sktPtr, (struct sockaddr *) &bindAddr6, sizeof(bindAddr6));
                        if (err < 0) { err = errno; perror("bind"); fflush(stderr); }
                        }
                } // endif (intfAddr->sa_family == AF_INET6)
#endif

        // Set the socket to non-blocking.
        if (err == 0)
                {
                err = fcntl(*sktPtr, F_GETFL, 0);
                if (err < 0) err = errno;
                else
                        {
                        err = fcntl(*sktPtr, F_SETFL, err | O_NONBLOCK);
                        if (err < 0) err = errno;
                        }
                }

        // Clean up
        if (err != 0 && *sktPtr != -1) { assert(close(*sktPtr) == 0); *sktPtr = -1; }
        assert((err == 0) == (*sktPtr != -1));
        return err;
        }

// Creates a PosixNetworkInterface for the interface whose IP address is
// intfAddr and whose name is intfName and registers it with mDNS core.
mDNSlocal int SetupOneInterface(mDNS *const m, struct sockaddr *intfAddr, struct sockaddr *intfMask, const char *intfName, int intfIndex)
        {
        int err = 0;
        PosixNetworkInterface *intf;
        PosixNetworkInterface *alias = NULL;

        assert(m != NULL);
        assert(intfAddr != NULL);
        assert(intfName != NULL);
        assert(intfMask != NULL);

        // Allocate the interface structure itself.
        intf = (PosixNetworkInterface*)malloc(sizeof(*intf));
        if (intf == NULL) { assert(0); err = ENOMEM; }

        // And make a copy of the intfName.
        if (err == 0)
                {
                intf->intfName = strdup(intfName);
                if (intf->intfName == NULL) { assert(0); err = ENOMEM; }
                }

        if (err == 0)
                {
                // Set up the fields required by the mDNS core.
                SockAddrTomDNSAddr(intfAddr, &intf->coreIntf.ip, NULL);
                SockAddrTomDNSAddr(intfMask, &intf->coreIntf.mask, NULL);
                //LogMsg("SetupOneInterface: %#a %#a",  &intf->coreIntf.ip,  &intf->coreIntf.mask);
                strncpy(intf->coreIntf.ifname, intfName, sizeof(intf->coreIntf.ifname));
                intf->coreIntf.ifname[sizeof(intf->coreIntf.ifname)-1] = 0;
                intf->coreIntf.Advertise = m->AdvertiseLocalAddresses;
                intf->coreIntf.McastTxRx = mDNStrue;

                // Set up the extra fields in PosixNetworkInterface.
                assert(intf->intfName != NULL);         // intf->intfName already set up above
                intf->index                = intfIndex;
                intf->multicastSocket4     = -1;
#if HAVE_IPV6
                intf->multicastSocket6     = -1;
#endif
                alias                      = SearchForInterfaceByName(m, intf->intfName);
                if (alias == NULL) alias   = intf;
                intf->coreIntf.InterfaceID = (mDNSInterfaceID)alias;

                if (alias != intf)
                        debugf("SetupOneInterface: %s %#a is an alias of %#a", intfName, &intf->coreIntf.ip, &alias->coreIntf.ip);
                }

        // Set up the multicast socket
        if (err == 0)
                {
                if (alias->multicastSocket4 == -1 && intfAddr->sa_family == AF_INET)
                        err = SetupSocket(intfAddr, MulticastDNSPort, intf->index, &alias->multicastSocket4);
#if HAVE_IPV6
                else if (alias->multicastSocket6 == -1 && intfAddr->sa_family == AF_INET6)
                        err = SetupSocket(intfAddr, MulticastDNSPort, intf->index, &alias->multicastSocket6);
#endif
                }

        // The interface is all ready to go, let's register it with the mDNS core.
        if (err == 0)
                err = mDNS_RegisterInterface(m, &intf->coreIntf, mDNSfalse);

        // Clean up.
        if (err == 0)
                {
                num_registered_interfaces++;
                debugf("SetupOneInterface: %s %#a Registered", intf->intfName, &intf->coreIntf.ip);
                if (gMDNSPlatformPosixVerboseLevel > 0)
                        fprintf(stderr, "Registered interface %s\n", intf->intfName);
                }
        else
                {
                // Use intfName instead of intf->intfName in the next line to avoid dereferencing NULL.
                debugf("SetupOneInterface: %s %#a failed to register %d", intfName, &intf->coreIntf.ip, err);
                if (intf) { FreePosixNetworkInterface(intf); intf = NULL; }
                }

        assert((err == 0) == (intf != NULL));

        return err;
        }

// Call get_ifi_info() to obtain a list of active interfaces and call SetupOneInterface() on each one.
mDNSlocal int SetupInterfaceList(mDNS *const m)
        {
        mDNSBool        foundav4       = mDNSfalse;
        int             err            = 0;
        struct ifi_info *intfList      = get_ifi_info(AF_INET, mDNStrue);
        struct ifi_info *firstLoopback = NULL;

        assert(m != NULL);
        debugf("SetupInterfaceList");

        if (intfList == NULL) err = ENOENT;

#if HAVE_IPV6
        if (err == 0)           /* Link the IPv6 list to the end of the IPv4 list */
                {
                struct ifi_info **p = &intfList;
                while (*p) p = &(*p)->ifi_next;
                *p = get_ifi_info(AF_INET6, mDNStrue);
                }
#endif

        if (err == 0)
                {
                struct ifi_info *i = intfList;
                while (i)
                        {
                        if (     ((i->ifi_addr->sa_family == AF_INET)
#if HAVE_IPV6
                                          || (i->ifi_addr->sa_family == AF_INET6)
#endif
                                ) &&  (i->ifi_flags & IFF_UP) && !(i->ifi_flags & IFF_POINTOPOINT))
                                {
                                if (i->ifi_flags & IFF_LOOPBACK)
                                        {
                                        if (firstLoopback == NULL)
                                                firstLoopback = i;
                                        }
                                else
                                        {
                                        if (SetupOneInterface(m, i->ifi_addr, i->ifi_netmask, i->ifi_name, i->ifi_index) == 0)
                                                if (i->ifi_addr->sa_family == AF_INET)
                                                        foundav4 = mDNStrue;
                                        }
                                }
                        i = i->ifi_next;
                        }

                // If we found no normal interfaces but we did find a loopback interface, register the
                // loopback interface.  This allows self-discovery if no interfaces are configured.
                // Temporary workaround: Multicast loopback on IPv6 interfaces appears not to work.
                // In the interim, we skip loopback interface only if we found at least one v4 interface to use
                // if ((m->HostInterfaces == NULL) && (firstLoopback != NULL))
                if (!foundav4 && firstLoopback)
                        (void) SetupOneInterface(m, firstLoopback->ifi_addr, firstLoopback->ifi_netmask, firstLoopback->ifi_name, firstLoopback->ifi_index);
                }

        // Clean up.
        if (intfList != NULL) free_ifi_info(intfList);
        return err;
        }

#if USES_NETLINK

// See <http://www.faqs.org/rfcs/rfc3549.html> for a description of NetLink

// Open a socket that will receive interface change notifications
mDNSlocal mStatus OpenIfNotifySocket(int *pFD)
        {
        mStatus                                 err = mStatus_NoError;
        struct sockaddr_nl              snl;
        int sock;
        int ret;

        sock = socket(AF_NETLINK, SOCK_RAW, NETLINK_ROUTE);
        if (sock < 0)
                return errno;

        // Configure read to be non-blocking because inbound msg size is not known in advance
        (void) fcntl(sock, F_SETFL, O_NONBLOCK);

        /* Subscribe the socket to Link & IP addr notifications. */
        mDNSPlatformMemZero(&snl, sizeof snl);
        snl.nl_family = AF_NETLINK;
        snl.nl_groups = RTMGRP_LINK | RTMGRP_IPV4_IFADDR;
        ret = bind(sock, (struct sockaddr *) &snl, sizeof snl);
        if (0 == ret)
                *pFD = sock;
        else
                err = errno;

        return err;
        }

#if MDNS_DEBUGMSGS
mDNSlocal void          PrintNetLinkMsg(const struct nlmsghdr *pNLMsg)
        {
        const char *kNLMsgTypes[] = { "", "NLMSG_NOOP", "NLMSG_ERROR", "NLMSG_DONE", "NLMSG_OVERRUN" };
        const char *kNLRtMsgTypes[] = { "RTM_NEWLINK", "RTM_DELLINK", "RTM_GETLINK", "RTM_NEWADDR", "RTM_DELADDR", "RTM_GETADDR" };

        printf("nlmsghdr len=%d, type=%s, flags=0x%x\n", pNLMsg->nlmsg_len, 
                        pNLMsg->nlmsg_type < RTM_BASE ? kNLMsgTypes[pNLMsg->nlmsg_type] : kNLRtMsgTypes[pNLMsg->nlmsg_type - RTM_BASE], 
                        pNLMsg->nlmsg_flags);

        if (RTM_NEWLINK <= pNLMsg->nlmsg_type && pNLMsg->nlmsg_type <= RTM_GETLINK)
                {
                struct ifinfomsg        *pIfInfo = (struct ifinfomsg*) NLMSG_DATA(pNLMsg);
                printf("ifinfomsg family=%d, type=%d, index=%d, flags=0x%x, change=0x%x\n", pIfInfo->ifi_family, 
                                pIfInfo->ifi_type, pIfInfo->ifi_index, pIfInfo->ifi_flags, pIfInfo->ifi_change);
                
                }
        else if (RTM_NEWADDR <= pNLMsg->nlmsg_type && pNLMsg->nlmsg_type <= RTM_GETADDR)
                {
                struct ifaddrmsg        *pIfAddr = (struct ifaddrmsg*) NLMSG_DATA(pNLMsg);
                printf("ifaddrmsg family=%d, index=%d, flags=0x%x\n", pIfAddr->ifa_family, 
                                pIfAddr->ifa_index, pIfAddr->ifa_flags);
                }
        printf("\n");
        }
#endif

mDNSlocal mDNSu32               ProcessRoutingNotification(int sd)
// Read through the messages on sd and if any indicate that any interface records should
// be torn down and rebuilt, return affected indices as a bitmask. Otherwise return 0.
        {
        ssize_t                                 readCount;
        char                                    buff[4096];     
        struct nlmsghdr                 *pNLMsg = (struct nlmsghdr*) buff;
        mDNSu32                         result = 0;
        
        // The structure here is more complex than it really ought to be because,
        // unfortunately, there's no good way to size a buffer in advance large
        // enough to hold all pending data and so avoid message fragmentation.
        // (Note that FIONREAD is not supported on AF_NETLINK.)

        readCount = read(sd, buff, sizeof buff);
        while (1)
                {
                // Make sure we've got an entire nlmsghdr in the buffer, and payload, too.
                // If not, discard already-processed messages in buffer and read more data.
                if (((char*) &pNLMsg[1] > (buff + readCount)) ||        // i.e. *pNLMsg extends off end of buffer 
                         ((char*) pNLMsg + pNLMsg->nlmsg_len > (buff + readCount)))
                        {
                        if (buff < (char*) pNLMsg)              // we have space to shuffle
                                {
                                // discard processed data
                                readCount -= ((char*) pNLMsg - buff);
                                memmove(buff, pNLMsg, readCount);
                                pNLMsg = (struct nlmsghdr*) buff;

                                // read more data
                                readCount += read(sd, buff + readCount, sizeof buff - readCount);
                                continue;                                       // spin around and revalidate with new readCount
                                }
                        else
                                break;  // Otherwise message does not fit in buffer
                        }

#if MDNS_DEBUGMSGS
                PrintNetLinkMsg(pNLMsg);
#endif

                // Process the NetLink message
                if (pNLMsg->nlmsg_type == RTM_GETLINK || pNLMsg->nlmsg_type == RTM_NEWLINK)
                        result |= 1 << ((struct ifinfomsg*) NLMSG_DATA(pNLMsg))->ifi_index;
                else if (pNLMsg->nlmsg_type == RTM_DELADDR || pNLMsg->nlmsg_type == RTM_NEWADDR)
                        result |= 1 << ((struct ifaddrmsg*) NLMSG_DATA(pNLMsg))->ifa_index;

                // Advance pNLMsg to the next message in the buffer
                if ((pNLMsg->nlmsg_flags & NLM_F_MULTI) != 0 && pNLMsg->nlmsg_type != NLMSG_DONE)
                        {
                        ssize_t len = readCount - ((char*)pNLMsg - buff);
                        pNLMsg = NLMSG_NEXT(pNLMsg, len);
                        }
                else
                        break;  // all done!
                }

        return result;
        }

#else // USES_NETLINK

// Open a socket that will receive interface change notifications
mDNSlocal mStatus OpenIfNotifySocket(int *pFD)
        {
        *pFD = socket(AF_ROUTE, SOCK_RAW, 0);

        if (*pFD < 0)
                return mStatus_UnknownErr;

        // Configure read to be non-blocking because inbound msg size is not known in advance
        (void) fcntl(*pFD, F_SETFL, O_NONBLOCK);

        return mStatus_NoError;
        }

#if MDNS_DEBUGMSGS
mDNSlocal void          PrintRoutingSocketMsg(const struct ifa_msghdr *pRSMsg)
        {
        const char *kRSMsgTypes[] = { "", "RTM_ADD", "RTM_DELETE", "RTM_CHANGE", "RTM_GET", "RTM_LOSING",
                                        "RTM_REDIRECT", "RTM_MISS", "RTM_LOCK", "RTM_OLDADD", "RTM_OLDDEL", "RTM_RESOLVE", 
                                        "RTM_NEWADDR", "RTM_DELADDR", "RTM_IFINFO", "RTM_NEWMADDR", "RTM_DELMADDR" };

        int             index = pRSMsg->ifam_type == RTM_IFINFO ? ((struct if_msghdr*) pRSMsg)->ifm_index : pRSMsg->ifam_index;

        printf("ifa_msghdr len=%d, type=%s, index=%d\n", pRSMsg->ifam_msglen, kRSMsgTypes[pRSMsg->ifam_type], index);
        }
#endif

mDNSlocal mDNSu32               ProcessRoutingNotification(int sd)
// Read through the messages on sd and if any indicate that any interface records should
// be torn down and rebuilt, return affected indices as a bitmask. Otherwise return 0.
        {
        ssize_t                                 readCount;
        char                                    buff[4096];     
        struct ifa_msghdr               *pRSMsg = (struct ifa_msghdr*) buff;
        mDNSu32                         result = 0;

        readCount = read(sd, buff, sizeof buff);
        if (readCount < (ssize_t) sizeof(struct ifa_msghdr))
                return mStatus_UnsupportedErr;          // cannot decipher message

#if MDNS_DEBUGMSGS
        PrintRoutingSocketMsg(pRSMsg);
#endif

        // Process the message
        if (pRSMsg->ifam_type == RTM_NEWADDR || pRSMsg->ifam_type == RTM_DELADDR ||
                 pRSMsg->ifam_type == RTM_IFINFO)
                {
                if (pRSMsg->ifam_type == RTM_IFINFO)
                        result |= 1 << ((struct if_msghdr*) pRSMsg)->ifm_index;
                else
                        result |= 1 << pRSMsg->ifam_index;
                }

        return result;
        }

#endif // USES_NETLINK

// Called when data appears on interface change notification socket
mDNSlocal void InterfaceChangeCallback(int fd, short filter, void *context)
        {
        IfChangeRec             *pChgRec = (IfChangeRec*) context;
        fd_set                  readFDs;
        mDNSu32         changedInterfaces = 0;
        struct timeval  zeroTimeout = { 0, 0 };

        (void)fd; // Unused
        (void)filter; // Unused

        FD_ZERO(&readFDs);
        FD_SET(pChgRec->NotifySD, &readFDs);
        
        do
        {
                changedInterfaces |= ProcessRoutingNotification(pChgRec->NotifySD);
        }
        while (0 < select(pChgRec->NotifySD + 1, &readFDs, (fd_set*) NULL, (fd_set*) NULL, &zeroTimeout));

        // Currently we rebuild the entire interface list whenever any interface change is
        // detected. If this ever proves to be a performance issue in a multi-homed 
        // configuration, more care should be paid to changedInterfaces.
        if (changedInterfaces)
                mDNSPlatformPosixRefreshInterfaceList(pChgRec->mDNS);
        }

// Register with either a Routing Socket or RtNetLink to listen for interface changes.
mDNSlocal mStatus WatchForInterfaceChange(mDNS *const m)
        {
        mStatus         err;
        IfChangeRec     *pChgRec;

        pChgRec = (IfChangeRec*) mDNSPlatformMemAllocate(sizeof *pChgRec);
        if (pChgRec == NULL)
                return mStatus_NoMemoryErr;

        pChgRec->mDNS = m;
        err = OpenIfNotifySocket(&pChgRec->NotifySD);
        if (err == 0)
                err = mDNSPosixAddFDToEventLoop(pChgRec->NotifySD, InterfaceChangeCallback, pChgRec);

        return err;
        }

// Test to see if we're the first client running on UDP port 5353, by trying to bind to 5353 without using SO_REUSEPORT.
// If we fail, someone else got here first. That's not a big problem; we can share the port for multicast responses --
// we just need to be aware that we shouldn't expect to successfully receive unicast UDP responses.
mDNSlocal mDNSBool mDNSPlatformInit_CanReceiveUnicast(void)
        {
        int err;
        int s = socket(AF_INET, SOCK_DGRAM, IPPROTO_UDP);
        struct sockaddr_in s5353;
        s5353.sin_family      = AF_INET;
        s5353.sin_port        = MulticastDNSPort.NotAnInteger;
        s5353.sin_addr.s_addr = 0;
        err = bind(s, (struct sockaddr *)&s5353, sizeof(s5353));
        close(s);
        if (err) debugf("No unicast UDP responses");
        else     debugf("Unicast UDP responses okay");
        return(err == 0);
        }

// mDNS core calls this routine to initialise the platform-specific data.
mDNSexport mStatus mDNSPlatformInit(mDNS *const m)
        {
        int err = 0;
        struct sockaddr sa;
        assert(m != NULL);

        if (mDNSPlatformInit_CanReceiveUnicast()) m->CanReceiveUnicastOn5353 = mDNStrue;

        // Tell mDNS core the names of this machine.

        // Set up the nice label
        m->nicelabel.c[0] = 0;
        GetUserSpecifiedFriendlyComputerName(&m->nicelabel);
        if (m->nicelabel.c[0] == 0) MakeDomainLabelFromLiteralString(&m->nicelabel, "Computer");

        // Set up the RFC 1034-compliant label
        m->hostlabel.c[0] = 0;
        GetUserSpecifiedRFC1034ComputerName(&m->hostlabel);
        if (m->hostlabel.c[0] == 0) MakeDomainLabelFromLiteralString(&m->hostlabel, "Computer");

        mDNS_SetFQDN(m);

        sa.sa_family = AF_INET;
        m->p->unicastSocket4 = -1;
        if (err == mStatus_NoError) err = SetupSocket(&sa, zeroIPPort, 0, &m->p->unicastSocket4);
#if HAVE_IPV6
        sa.sa_family = AF_INET6;
        m->p->unicastSocket6 = -1;
        if (err == mStatus_NoError) err = SetupSocket(&sa, zeroIPPort, 0, &m->p->unicastSocket6);
#endif

        // Tell mDNS core about the network interfaces on this machine.
        if (err == mStatus_NoError) err = SetupInterfaceList(m);

        // Tell mDNS core about DNS Servers
        mDNS_Lock(m);
        if (err == mStatus_NoError) ParseDNSServers(m, uDNS_SERVERS_FILE);
        mDNS_Unlock(m);

        if (err == mStatus_NoError)
                {
                err = WatchForInterfaceChange(m);
                // Failure to observe interface changes is non-fatal.
                if (err != mStatus_NoError)
                        {
                        fprintf(stderr, "mDNS(%d) WARNING: Unable to detect interface changes (%d).\n", getpid(), err);
                        err = mStatus_NoError;
                        }
                }

        // We don't do asynchronous initialization on the Posix platform, so by the time
        // we get here the setup will already have succeeded or failed.  If it succeeded,
        // we should just call mDNSCoreInitComplete() immediately.
        if (err == mStatus_NoError)
                mDNSCoreInitComplete(m, mStatus_NoError);

        return PosixErrorToStatus(err);
        }

// mDNS core calls this routine to clean up the platform-specific data.
// In our case all we need to do is to tear down every network interface.
mDNSexport void mDNSPlatformClose(mDNS *const m)
        {
        assert(m != NULL);
        ClearInterfaceList(m);
        if (m->p->unicastSocket4 != -1) assert(close(m->p->unicastSocket4) == 0);
#if HAVE_IPV6
        if (m->p->unicastSocket6 != -1) assert(close(m->p->unicastSocket6) == 0);
#endif
        }

mDNSexport mStatus mDNSPlatformPosixRefreshInterfaceList(mDNS *const m)
        {
        int err;
        ClearInterfaceList(m);
        err = SetupInterfaceList(m);
        return PosixErrorToStatus(err);
        }

#if COMPILER_LIKES_PRAGMA_MARK
#pragma mark ***** Locking
#endif

// On the Posix platform, locking is a no-op because we only ever enter
// mDNS core on the main thread.

// mDNS core calls this routine when it wants to prevent
// the platform from reentering mDNS core code.
mDNSexport void    mDNSPlatformLock   (const mDNS *const m)
        {
        (void) m;       // Unused
        }

// mDNS core calls this routine when it release the lock taken by
// mDNSPlatformLock and allow the platform to reenter mDNS core code.
mDNSexport void    mDNSPlatformUnlock (const mDNS *const m)
        {
        (void) m;       // Unused
        }

#if COMPILER_LIKES_PRAGMA_MARK
#pragma mark ***** Strings
#endif

// mDNS core calls this routine to copy C strings.
// On the Posix platform this maps directly to the ANSI C strcpy.
mDNSexport void    mDNSPlatformStrCopy(void *dst, const void *src)
        {
        strcpy((char *)dst, (char *)src);
        }

// mDNS core calls this routine to get the length of a C string.
// On the Posix platform this maps directly to the ANSI C strlen.
mDNSexport mDNSu32  mDNSPlatformStrLen (const void *src)
        {
        return strlen((char*)src);
        }

// mDNS core calls this routine to copy memory.
// On the Posix platform this maps directly to the ANSI C memcpy.
mDNSexport void    mDNSPlatformMemCopy(void *dst, const void *src, mDNSu32 len)
        {
        memcpy(dst, src, len);
        }

// mDNS core calls this routine to test whether blocks of memory are byte-for-byte
// identical. On the Posix platform this is a simple wrapper around ANSI C memcmp.
mDNSexport mDNSBool mDNSPlatformMemSame(const void *dst, const void *src, mDNSu32 len)
        {
        return memcmp(dst, src, len) == 0;
        }

// mDNS core calls this routine to clear blocks of memory.
// On the Posix platform this is a simple wrapper around ANSI C memset.
mDNSexport void    mDNSPlatformMemZero(void *dst, mDNSu32 len)
        {
        memset(dst, 0, len);
        }

mDNSexport void *  mDNSPlatformMemAllocate(mDNSu32 len) { return(malloc(len)); }
mDNSexport void    mDNSPlatformMemFree    (void *mem)   { free(mem); }

mDNSexport mDNSu32 mDNSPlatformRandomSeed(void)
        {
        struct timeval tv;
        gettimeofday(&tv, NULL);
        return(tv.tv_usec);
        }

mDNSexport mDNSs32  mDNSPlatformOneSecond = 1024;

mDNSexport mStatus mDNSPlatformTimeInit(void)
        {
        // No special setup is required on Posix -- we just use gettimeofday();
        // This is not really safe, because gettimeofday can go backwards if the user manually changes the date or time
        // We should find a better way to do this
        return(mStatus_NoError);
        }

mDNSexport mDNSs32  mDNSPlatformRawTime()
        {
        struct timeval tv;
        gettimeofday(&tv, NULL);
        // tv.tv_sec is seconds since 1st January 1970 (GMT, with no adjustment for daylight savings time)
        // tv.tv_usec is microseconds since the start of this second (i.e. values 0 to 999999)
        // We use the lower 22 bits of tv.tv_sec for the top 22 bits of our result
        // and we multiply tv.tv_usec by 16 / 15625 to get a value in the range 0-1023 to go in the bottom 10 bits.
        // This gives us a proper modular (cyclic) counter that has a resolution of roughly 1ms (actually 1/1024 second)
        // and correctly cycles every 2^22 seconds (4194304 seconds = approx 48 days).
        return((tv.tv_sec << 10) | (tv.tv_usec * 16 / 15625));
        }

mDNSexport mDNSs32 mDNSPlatformUTC(void)
        {
        return time(NULL);
        }

mDNSlocal void mDNSPosixAddToFDSet(int *nfds, fd_set *readfds, int s)
        {
        if (*nfds < s + 1) *nfds = s + 1;
        FD_SET(s, readfds);
        }

mDNSexport void mDNSPosixGetFDSet(mDNS *m, int *nfds, fd_set *readfds, struct timeval *timeout)
        {
        mDNSs32 ticks;
        struct timeval interval;

        // 1. Call mDNS_Execute() to let mDNSCore do what it needs to do
        mDNSs32 nextevent = mDNS_Execute(m);

        // 2. Build our list of active file descriptors
        PosixNetworkInterface *info = (PosixNetworkInterface *)(m->HostInterfaces);
        if (m->p->unicastSocket4 != -1) mDNSPosixAddToFDSet(nfds, readfds, m->p->unicastSocket4);
#if HAVE_IPV6
        if (m->p->unicastSocket6 != -1) mDNSPosixAddToFDSet(nfds, readfds, m->p->unicastSocket6);
#endif
        while (info)
                {
                if (info->multicastSocket4 != -1) mDNSPosixAddToFDSet(nfds, readfds, info->multicastSocket4);
#if HAVE_IPV6
                if (info->multicastSocket6 != -1) mDNSPosixAddToFDSet(nfds, readfds, info->multicastSocket6);
#endif
                info = (PosixNetworkInterface *)(info->coreIntf.next);
                }

        // 3. Calculate the time remaining to the next scheduled event (in struct timeval format)
        ticks = nextevent - mDNS_TimeNow(m);
        if (ticks < 1) ticks = 1;
        interval.tv_sec  = ticks >> 10;                                         // The high 22 bits are seconds
        interval.tv_usec = ((ticks & 0x3FF) * 15625) / 16;      // The low 10 bits are 1024ths

        // 4. If client's proposed timeout is more than what we want, then reduce it
        if (timeout->tv_sec > interval.tv_sec ||
                (timeout->tv_sec == interval.tv_sec && timeout->tv_usec > interval.tv_usec))
                *timeout = interval;
        }

mDNSexport void mDNSPosixProcessFDSet(mDNS *const m, fd_set *readfds)
        {
        PosixNetworkInterface *info;
        assert(m       != NULL);
        assert(readfds != NULL);
        info = (PosixNetworkInterface *)(m->HostInterfaces);

        if (m->p->unicastSocket4 != -1 && FD_ISSET(m->p->unicastSocket4, readfds))
                {
                FD_CLR(m->p->unicastSocket4, readfds);
                SocketDataReady(m, NULL, m->p->unicastSocket4);
                }
#if HAVE_IPV6
        if (m->p->unicastSocket6 != -1 && FD_ISSET(m->p->unicastSocket6, readfds))
                {
                FD_CLR(m->p->unicastSocket6, readfds);
                SocketDataReady(m, NULL, m->p->unicastSocket6);
                }
#endif

        while (info)
                {
                if (info->multicastSocket4 != -1 && FD_ISSET(info->multicastSocket4, readfds))
                        {
                        FD_CLR(info->multicastSocket4, readfds);
                        SocketDataReady(m, info, info->multicastSocket4);
                        }
#if HAVE_IPV6
                if (info->multicastSocket6 != -1 && FD_ISSET(info->multicastSocket6, readfds))
                        {
                        FD_CLR(info->multicastSocket6, readfds);
                        SocketDataReady(m, info, info->multicastSocket6);
                        }
#endif
                info = (PosixNetworkInterface *)(info->coreIntf.next);
                }
        }

// update gMaxFD
mDNSlocal void  DetermineMaxEventFD(void)
        {
        PosixEventSource        *iSource;
        
        gMaxFD = 0;
        for (iSource=(PosixEventSource*)gEventSources.Head; iSource; iSource = iSource->Next)
                if (gMaxFD < iSource->fd)
                        gMaxFD = iSource->fd;
        }

// Add a file descriptor to the set that mDNSPosixRunEventLoopOnce() listens to.
mStatus mDNSPosixAddFDToEventLoop(int fd, mDNSPosixEventCallback callback, void *context)
        {
        PosixEventSource        *newSource;
        
        if (gEventSources.LinkOffset == 0)
                InitLinkedList(&gEventSources, offsetof(PosixEventSource, Next));

        if (fd >= (int) FD_SETSIZE || fd < 0)
                return mStatus_UnsupportedErr;
        if (callback == NULL)
                return mStatus_BadParamErr;

        newSource = (PosixEventSource*) malloc(sizeof *newSource);
        if (NULL == newSource)
                return mStatus_NoMemoryErr;

        newSource->Callback = callback;
        newSource->Context = context;
        newSource->fd = fd;

        AddToTail(&gEventSources, newSource);
        FD_SET(fd, &gEventFDs);

        DetermineMaxEventFD();

        return mStatus_NoError;
        }

// Remove a file descriptor from the set that mDNSPosixRunEventLoopOnce() listens to.
mStatus mDNSPosixRemoveFDFromEventLoop(int fd)
        {
        PosixEventSource        *iSource;
        
        for (iSource=(PosixEventSource*)gEventSources.Head; iSource; iSource = iSource->Next)
                {
                if (fd == iSource->fd)
                        {
                        FD_CLR(fd, &gEventFDs);
                        RemoveFromList(&gEventSources, iSource);
                        free(iSource);
                        DetermineMaxEventFD();
                        return mStatus_NoError;
                        }
                }
        return mStatus_NoSuchNameErr;
        }

// Simply note the received signal in gEventSignals.
mDNSlocal void  NoteSignal(int signum)
        {
        sigaddset(&gEventSignals, signum);
        }

// Tell the event package to listen for signal and report it in mDNSPosixRunEventLoopOnce().
mStatus mDNSPosixListenForSignalInEventLoop(int signum)
        {
        struct sigaction        action;
        mStatus                         err;

        mDNSPlatformMemZero(&action, sizeof action);            // more portable than member-wise assignment
        action.sa_handler = NoteSignal;
        err = sigaction(signum, &action, (struct sigaction*) NULL);
        
        sigaddset(&gEventSignalSet, signum);

        return err;
        }

// Tell the event package to stop listening for signal in mDNSPosixRunEventLoopOnce().
mStatus mDNSPosixIgnoreSignalInEventLoop(int signum)
        {
        struct sigaction        action;
        mStatus                         err;

        mDNSPlatformMemZero(&action, sizeof action);            // more portable than member-wise assignment
        action.sa_handler = SIG_DFL;
        err = sigaction(signum, &action, (struct sigaction*) NULL);
        
        sigdelset(&gEventSignalSet, signum);

        return err;
        }

// Do a single pass through the attendent event sources and dispatch any found to their callbacks.
// Return as soon as internal timeout expires, or a signal we're listening for is received.
mStatus mDNSPosixRunEventLoopOnce(mDNS *m, const struct timeval *pTimeout, 
                                                                        sigset_t *pSignalsReceived, mDNSBool *pDataDispatched)
        {
        fd_set                  listenFDs = gEventFDs;
        int                             fdMax = 0, numReady;
        struct timeval  timeout = *pTimeout;
        
        // Include the sockets that are listening to the wire in our select() set
        mDNSPosixGetFDSet(m, &fdMax, &listenFDs, &timeout);     // timeout may get modified
        if (fdMax < gMaxFD)
                fdMax = gMaxFD;

        numReady = select(fdMax + 1, &listenFDs, (fd_set*) NULL, (fd_set*) NULL, &timeout);

        // If any data appeared, invoke its callback
        if (numReady > 0)
                {
                PosixEventSource        *iSource;

                (void) mDNSPosixProcessFDSet(m, &listenFDs);    // call this first to process wire data for clients

                for (iSource=(PosixEventSource*)gEventSources.Head; iSource; iSource = iSource->Next)
                        {
                        if (FD_ISSET(iSource->fd, &listenFDs))
                                {
                                iSource->Callback(iSource->fd, 0, iSource->Context);
                                break;  // in case callback removed elements from gEventSources
                                }
                        }
                *pDataDispatched = mDNStrue;
                }
        else
                *pDataDispatched = mDNSfalse;

        (void) sigprocmask(SIG_BLOCK, &gEventSignalSet, (sigset_t*) NULL);
        *pSignalsReceived = gEventSignals;
        sigemptyset(&gEventSignals);
        (void) sigprocmask(SIG_UNBLOCK, &gEventSignalSet, (sigset_t*) NULL);

        return mStatus_NoError;
        }