core/sip/raw_sock.cpp

/*
 * $Id$
 *
 * Copyright (C) 2010 iptelorg GmbH
 *
 * Permission to use, copy, modify, and distribute this software for any
 * purpose with or without fee is hereby granted, provided that the above
 * copyright notice and this permission notice appear in all copies.
 *
 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 */
/** raw socket functions.
 *  @file raw_sock.c
 *  @ingroup core
 *  Module: @ref core
 */
/*
 * History:
 * --------
 *  2010-06-07  initial version (from older code) andrei
 *  2010-06-15  IP_HDRINCL raw socket support, including on-send
 *               fragmentation (andrei)
 */

#include "ip_util.h"
#include "log.h"

#include <errno.h>
#include <string.h>
#include <unistd.h>
#include <sys/types.h>
#include <fcntl.h>
#include <sys/socket.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <arpa/inet.h>
#ifndef __USE_BSD
#define __USE_BSD  /* on linux use bsd version of iphdr (more portable) */
#endif /* __USE_BSD */
#include <netinet/ip.h>
#define __FAVOR_BSD /* on linux use bsd version of udphdr (more portable) */
#include <netinet/udp.h>
#include <netdb.h>
#ifdef __DragonFly__
#include <err.h>
#endif

#include "raw_sock.h"

/* likely/unlikely */
#if __GNUC__ >= 3

#define likely(expr)              __builtin_expect(!!(expr), 1)
#define unlikely(expr)            __builtin_expect(!!(expr), 0)

#else /* __GNUC__ */

/* #warning "No compiler optimizations supported try gcc 4.x" */
#define likely(expr) (expr)
#define unlikely(expr) (expr)

#endif /* __GNUC__ */


// macros for converting values in the expected format
// #if OS == "freebsd" || OS == "netbsd" || OS == "darwin"
/* on freebsd and netbsd the ip offset (along with flags) and the
   ip header length must be filled in _host_ bytes order format.
   The same is true for openbsd < 2.1.
*/
#if defined(RAW_IPHDR_IP_HBO)

/** convert the ip offset in the format expected by the kernel. */
#define RAW_IPHDR_IP_OFF(off) (unsigned short)(off)
/** convert the ip total length in the format expected by the kernel. */
#define RAW_IPHDR_IP_LEN(tlen) (unsigned short)(tlen)

#else /* __OS_* */
/* linux, openbsd >= 2.1 a.s.o. */
/** convert the ip offset in the format expected by the kernel. */
#define RAW_IPHDR_IP_OFF(off)  htons((unsigned short)(off))
/** convert the ip total length in the format expected by the kernel. */
#define RAW_IPHDR_IP_LEN(tlen) htons((unsigned short)(tlen))

#endif /* __OS_* */

/** create and return a raw socket.
 * @param proto - protocol used (e.g. IPPROTO_UDP, IPPROTO_RAW)
 * @param ip - if not null the socket will be bound on this ip.
 * @param iphdr_incl - set to 1 if packets send on this socket include
 *                     a pre-built ip header (some fields, like the checksum
 *                     will still be filled by the kernel, OTOH packet
 *                     fragmentation has to be done in user space).
 * @return socket on success, -1 on error
 */
int raw_socket(int proto, sockaddr_storage* ip, int iphdr_incl)
{
	int sock;
	int t;

	sock = socket(PF_INET, SOCK_RAW, proto);
	if (sock==-1)
	  goto error;
	/* set socket options */
	if (iphdr_incl) {
	  t=1;
	  if (setsockopt(sock, IPPROTO_IP, IP_HDRINCL, &t, sizeof(t))<0){
	    ERROR("raw_socket: setsockopt(IP_HDRINCL) failed: %s [%d]\n",
		  strerror(errno), errno);
	    goto error;
	  }
	} else {
	  /* IP_PKTINFO makes no sense if the ip header is included */
	  /* using IP_PKTINFO */
	  t=1;
#ifdef IP_PKTINFO
	  if (setsockopt(sock, IPPROTO_IP, IP_PKTINFO, &t, sizeof(t))<0){
	    ERROR("raw_socket: setsockopt(IP_PKTINFO) failed: %s [%d]\n",
		  strerror(errno), errno);
	    goto error;
	  }
#elif defined(IP_RECVDSTADDR)
	  if (setsockopt(sock, IPPROTO_IP, IP_RECVDSTADDR, &t, sizeof(t))<0){
	    ERROR("raw_socket: setsockop(IP_RECVDSTADDR) failed: %s [%d]\n",
		  strerror(errno), errno);
	    goto error;
	  }
#else
#error "no method of getting the destination ip address supported"
#endif /* IP_RECVDSTADDR / IP_PKTINFO */
	}
#if defined (IP_MTU_DISCOVER) && defined (IP_PMTUDISC_DONT)
	t=IP_PMTUDISC_DONT;
	if(setsockopt(sock, IPPROTO_IP, IP_MTU_DISCOVER, &t, sizeof(t)) ==-1){
	  ERROR("raw_socket: setsockopt(IP_MTU_DISCOVER): %s\n", strerror(errno));
	  goto error;
	}
#endif /* IP_MTU_DISCOVER && IP_PMTUDISC_DONT */
	/* FIXME: probe_max_receive_buffer(sock) missing */
	if (ip){
	  if (bind(sock, (sockaddr*)ip, SA_len(ip))==-1){
	    char ip_str[NI_MAXHOST];
	    ERROR("raw_socket: bind(%s) failed: %s [%d]\n",
		  am_inet_ntop(ip,ip_str,NI_MAXHOST), strerror(errno), errno);
	    goto error;
	  }
	}
	return sock;
error:
	if (sock!=-1) close(sock);
	return -1;
}


/** create and return an udp over ipv4  raw socket.
 * @param iphdr_incl - set to 1 if packets send on this socket include
 *                     a pre-built ip header (some fields, like the checksum
 *                     will still be filled by the kernel, OTOH packet
 *                     fragmentation has to be done in user space).
 * @return socket on success, -1 on error
 */
int raw_udp_socket(int iphdr_incl)
{
  return raw_socket(IPPROTO_UDP, NULL, iphdr_incl);
}


/** receives an ipv4 packet using a raw socket.
 * An ipv4 packet is received in buf, using IP_PKTINFO or IP_RECVDSTADDR.
 * from and to are filled (only the ip part the ports are 0 since this
 * function doesn't try to look beyond the IP level).
 * @param sock - raw socket
 * @param buf - detination buffer.
 * @param len - buffer len (should be enough for receiving a packet +
 *               IP header).
 * @param from - result parameter, the IP address part of it will be filled
 *                with the source address and the port with 0.
 * @param to - result parameter, the IP address part of it will be filled
 *                with the destination (local) address and the port with 0.
 * @return packet len or <0 on error: -1 (check errno),
 *        -2 no IP_PKTINFO/IP_RECVDSTADDR found or AF mismatch
 */
// int recvpkt4(int sock, char* buf, int len,
// 	     sockaddr_storage* from, sockaddr_storage* to)
// {
// 	struct iovec iov[1];
// 	struct msghdr rcv_msg;
// 	struct cmsghdr* cmsg;
// #ifdef IP_PKTINFO
// 	struct in_pktinfo* rcv_pktinfo;
// #endif /* IP_PKTINFO */
// 	int n, ret;
// 	char msg_ctrl_buf[1024];

// 	iov[0].iov_base=buf;
// 	iov[0].iov_len=len;
// 	rcv_msg.msg_name=from;
// 	rcv_msg.msg_namelen=sizeof(sockaddr_storage);
// 	rcv_msg.msg_control=msg_ctrl_buf;
// 	rcv_msg.msg_controllen=sizeof(msg_ctrl_buf);
// 	rcv_msg.msg_iov=&iov[0];
// 	rcv_msg.msg_iovlen=1;
// 	ret=-2; /* no PKT_INFO or AF mismatch */
// retry:
// 	n=recvmsg(sock, &rcv_msg, MSG_WAITALL);
// 	if (unlikely(n==-1)){
// 	  if (errno==EINTR)
// 	    goto retry;
// 	  ret=n;
// 	  goto end;
// 	}
// 	/* find the pkt info */
// 	for (cmsg=CMSG_FIRSTHDR(&rcv_msg); cmsg;
// 	     cmsg=CMSG_NXTHDR(&rcv_msg, cmsg)){
// #ifdef IP_PKTINFO
// 	  if ((cmsg->cmsg_level==IPPROTO_IP) &&
// 	      (cmsg->cmsg_type==IP_PKTINFO)) {
// 	    rcv_pktinfo=(struct in_pktinfo*)CMSG_DATA(cmsg);
// 	    to->ss_family=AF_INET;
// 	    memcpy(&SAv4(to)->sin_addr, &rcv_pktinfo->ipi_spec_dst.s_addr
// 		   sizeof(in_addr));
// 	    am_set_port(to,0); /* not known */
// 	    /* interface no. in ipi_ifindex */
// 	    ret=n; /* success */
// 	    break;
// 	  }
// #elif defined (IP_RECVDSTADDR)
// 	  if (likely((cmsg->cmsg_level==IPPROTO_IP) &&
// 		     (cmsg->cmsg_type==IP_RECVDSTADDR))) {
// 	    to->ss_family=AF_INET;
// 	    memcpy(&SAv4(to)->sin_addr, CMSG_DATA(cmsg),
// 		   sizeof(in_addr));
// 	    am_set_port(to,0); /* not known */
// 	    ret=n; /* success */
// 	    break;
// 	  }
// #else
// #error "no method of getting the destination ip address supported"
// #endif /* IP_PKTINFO / IP_RECVDSTADDR */
// 	}
//  end:
// 	return ret;
// }


/* receive an ipv4 udp packet over a raw socket.
 * The packet is copied in *buf and *buf is advanced to point to the
 * payload.  Fills from and to.
 * @param rsock - raw socket
 * @param buf - the packet will be written to where *buf points intially and
 *              then *buf will be advanced to point to the udp payload.
 * @param len - buffer length (should be enough to hold at least the
 *               ip and udp headers + 1 byte).
 * @param from - result parameter, filled with source address and port of the
 *               packet.
 * @param from - result parameter, filled with destination (local) address and
 *               port of the packet.
 * @param rf   - filter used to decide whether or not the packet is
 *                accepted/processed. If null, all the packets are accepted.
 * @return packet len or  <0 on error (-1 and -2 on recv error @see recvpkt4,
 *         -3 if the headers are invalid and -4 if the packet doesn't
 *         match the  filter).
 */
// int raw_udp4_recv(int rsock, char** buf, int len,
// 		  sockaddr_storage* from, sockaddr_storage* to)
// {
// 	int n;
// 	unsigned short dst_port;
// 	unsigned short src_port;
// 	//struct ip_addr dst_ip;
// 	char* end;
// 	char* udph_start;
// 	char* udp_payload;
// 	struct ip iph;
// 	struct udphdr udph;
// 	unsigned short udp_len;

// 	n=recvpkt4(rsock, *buf, len, from, to);
// 	if (unlikely(n<0)) goto error;

// 	end=*buf+n;
// 	if (unlikely(n<(sizeof(struct ip)+sizeof(struct udphdr)))) {
// 		n=-3;
// 		goto error;
// 	}

// 	/* FIXME: if initial buffer is aligned, one could skip the memcpy
// 	   and directly cast ip and udphdr pointer to the memory */
// 	memcpy(&iph, *buf, sizeof(struct ip));
// 	udph_start=*buf+iph.ip_hl*4;
// 	udp_payload=udph_start+sizeof(struct udphdr);
// 	if (unlikely(udp_payload>end)){
// 		n=-3;
// 		goto error;
// 	}
// 	memcpy(&udph, udph_start, sizeof(struct udphdr));
// 	udp_len=ntohs(udph.uh_ulen);
// 	if (unlikely((udph_start+udp_len)!=end)){
// 		if ((udph_start+udp_len)>end){
// 			n=-3;
// 			goto error;
// 		}else{
// 			ERROR("udp length too small: %d/%d\n",
// 			      (int)udp_len, (int)(end-udph_start));
// 			n=-3;
// 			goto error;
// 		}
// 	}
// 	/* advance buf */
// 	*buf=udp_payload;
// 	n=(int)(end-*buf);
// 	/* fill ip from the packet (needed if no PKT_INFO is used) */
// 	dst_ip.af=AF_INET;
// 	dst_ip.len=4;
// 	dst_ip.u.addr32[0]=iph.ip_dst.s_addr;
// 	/* fill dst_port */
// 	dst_port=ntohs(udph.uh_dport);
// 	ip_addr2su(to, &dst_ip, dst_port);
// 	/* fill src_port */
// 	src_port=ntohs(udph.uh_sport);
// 	su_setport(from, src_port);
// 	// if (likely(rf)) {
// 	// 	su2ip_addr(&dst_ip, to);
// 	// 	if ( (dst_port && rf->port1 && ((dst_port<rf->port1) ||
// 	// 					(dst_port>rf->port2)) ) ||
// 	// 		(matchnet(&dst_ip, &rf->dst)!=1) ){
// 	// 		/* no match */
// 	// 		n=-4;
// 	// 		goto error;
// 	// 	}
// 	// }
// error:
// 	return n;
// }


/** udp checksum helper: compute the pseudo-header 16-bit "sum".
 * Computes the partial checksum (no complement) of the pseudo-header.
 * It is meant to be used by udpv4_chksum().
 * @param uh - filled udp header
 * @param src - source ip address in network byte order.
 * @param dst - destination ip address in network byte order.
 * @param length - payload length (not including the udp header),
 *                 in _host_ order.
 * @return the partial checksum in host order
 */
inline unsigned short udpv4_vhdr_sum(struct udphdr* uh,
				     struct in_addr* src,
				     struct in_addr* dst,
				     unsigned short length)
{
	unsigned sum;

	/* pseudo header */
	sum=(src->s_addr>>16)+(src->s_addr&0xffff)+
		(dst->s_addr>>16)+(dst->s_addr&0xffff)+
		htons(IPPROTO_UDP)+(uh->uh_ulen);
	/* udp header */
	sum+=(uh->uh_dport)+(uh->uh_sport)+(uh->uh_ulen) + 0 /*chksum*/;
	/* fold it */
	sum=(sum>>16)+(sum&0xffff);
	sum+=(sum>>16);
	/* no complement */
	return ntohs((unsigned short) sum);
}


/** compute the udp over ipv4 checksum.
 * @param u - filled udp header (except checksum).
 * @param src - source ip v4 address, in _network_ byte order.
 * @param dst - destination ip v4 address, int _network_ byte order.
 * @param data - pointer to the udp payload.
 * @param length - payload length, not including the udp header and in
 *                 _host_ order. The length mist be <= 0xffff - 8
 *                 (to allow space for the udp header).
 * @return the checksum in _host_ order */
inline static unsigned short udpv4_chksum(struct udphdr* u,
					  struct in_addr* src,
					  struct in_addr* dst,
					  unsigned char* data,
					  unsigned short length)
{
	unsigned sum;
	unsigned char* end;
	sum=udpv4_vhdr_sum(u, src, dst, length);
	end=data+(length&(~0x1)); /* make sure it's even */
	/* TODO: 16 & 32 bit aligned version */
		/* not aligned */
		for(;data<end;data+=2){
			sum+=((data[0]<<8)+data[1]);
		}
		if (length&0x1)
			sum+=((*data)<<8);

	/* fold it */
	sum=(sum>>16)+(sum&0xffff);
	sum+=(sum>>16);
	return (unsigned short)~sum;
}


/** fill in an udp header.
 * @param u - udp header that will be filled.
 * @param from - source ip v4 address and port.
 * @param to -   destination ip v4 address and port.
 * @param buf - pointer to the payload.
 * @param len - payload length (not including the udp header).
 * @param do_chk - if set the udp checksum will be computed, else it will
 *                 be set to 0.
 * @return 0 on success, < 0 on error.
 */
inline static int mk_udp_hdr(struct udphdr* u,
			     const sockaddr_storage* from,
			     const sockaddr_storage* to,
			     unsigned char* buf, int len,
			     int do_chk)
{
  struct sockaddr_in *from_v4 = (sockaddr_in*)from;
  struct sockaddr_in *to_v4 = (sockaddr_in*)to;

  u->uh_ulen = htons((unsigned short)len+sizeof(struct udphdr));
  u->uh_sport = ((sockaddr_in*)from)->sin_port;
  u->uh_dport = ((sockaddr_in*)to)->sin_port;
  if (do_chk)
    u->uh_sum=htons(udpv4_chksum(u, &from_v4->sin_addr,
				 &to_v4->sin_addr, buf, len));
  else
    u->uh_sum=0; /* no checksum */
  return 0;
}


/** fill in an ip header.
 * Note: the checksum is _not_ computed.
 * WARNING: The ip header length and offset might be filled in
 * _host_ byte order or network byte order (depending on the OS, for example
 *  freebsd needs host byte order for raw sockets with IPHDR_INC, while
 *  linux needs network byte order).
 * @param iph - ip header that will be filled.
 * @param from - source ip v4 address (network byte order).
 * @param to -   destination ip v4 address (network byte order).
 * @param payload len - payload length (not including the ip header).
 * @param proto - protocol.
 * @return 0 on success, < 0 on error.
 */
inline static int mk_ip_hdr(struct ip* iph, struct in_addr* from,
			    struct in_addr* to, int payload_len,
			    unsigned char proto)
{
	iph->ip_hl = sizeof(struct ip)/4;
	iph->ip_v = 4;
	iph->ip_tos = 0;
	/* on freebsd ip_len _must_ be in _host_ byte order instead
	   of network byte order. On linux the length is ignored (it's filled
	   automatically every time). */
	iph->ip_len = RAW_IPHDR_IP_LEN(payload_len + sizeof(struct ip));
	iph->ip_id = 0; /* 0 => will be filled automatically by the kernel */
	iph->ip_off = 0; /* frag.: first 3 bits=flags=0, last 13 bits=offset */
	iph->ip_ttl = 64;//cfg_get(core, core_cfg, udp4_raw_ttl);
	iph->ip_p = proto;
	iph->ip_src = *from;
	iph->ip_dst = *to;
	iph->ip_sum = 0;

	return 0;
}


/** send an udp packet over a non-ip_hdrincl raw socket.
 * @param rsock - raw socket
 * @param buf - data
 * @param len - data len
 * @param from - source address:port (_must_ be non-null, but the ip address
 *                can be 0, in which case it will be filled by the kernel).
 * @param to - destination address:port
 * @return  <0 on error (errno set too), number of bytes sent on success
 *          (including the udp header => on success len + udpheader size).
 */
int raw_udp4_send(int rsock, char* buf, unsigned int len,
		  sockaddr_storage* from, sockaddr_storage* to)
{
	struct msghdr snd_msg;
	struct cmsghdr* cmsg;
#ifdef IP_PKTINFO
	struct in_pktinfo* snd_pktinfo;
#endif /* IP_PKTINFO */
	struct iovec iov[2];
	struct udphdr udp_hdr;
	char msg_ctrl_snd_buf[1024];
	int ret;

	memset(&snd_msg, 0, sizeof(snd_msg));
	snd_msg.msg_name=SAv4(to);
	snd_msg.msg_namelen=SA_len(to);
	snd_msg.msg_iov=&iov[0];
	/* prepare udp header */
	mk_udp_hdr(&udp_hdr, from, to, (unsigned char*)buf, len, 1);
	iov[0].iov_base=(char*)&udp_hdr;
	iov[0].iov_len=sizeof(udp_hdr);
	iov[1].iov_base=buf;
	iov[1].iov_len=len;
	snd_msg.msg_iovlen=2;
	snd_msg.msg_control=msg_ctrl_snd_buf;
	snd_msg.msg_controllen=sizeof(msg_ctrl_snd_buf);
	/* init pktinfo cmsg */
	cmsg=CMSG_FIRSTHDR(&snd_msg);
	cmsg->cmsg_level=IPPROTO_IP;
#ifdef IP_PKTINFO
	cmsg->cmsg_type=IP_PKTINFO;
	cmsg->cmsg_len=CMSG_LEN(sizeof(struct in_pktinfo));
	snd_pktinfo=(struct in_pktinfo*)CMSG_DATA(cmsg);
	snd_pktinfo->ipi_ifindex=0;
	snd_pktinfo->ipi_spec_dst.s_addr=SAv4(&from)->sin_addr.s_addr;
#elif defined (IP_SENDSRCADDR)
	cmsg->cmsg_type=IP_SENDSRCADDR;
	cmsg->cmsg_len=CMSG_LEN(sizeof(struct in_addr));
	memcpy(CMSG_DATA(cmsg), &SAv4(&from)->sin_addr.s_addr,
	       sizeof(struct in_addr));
#else
#ifdef __DragonFly__
	errx(1, "missing support for IP_SENDSRCADDR"); /* XXX FIXME */
#else
#error "no method of setting the source ip supported"
#endif
#endif /* IP_PKTINFO / IP_SENDSRCADDR */
	snd_msg.msg_controllen=cmsg->cmsg_len;
	snd_msg.msg_flags=0;
	ret=sendmsg(rsock, &snd_msg, 0);
	return ret;
}


/** send an udp packet over an IP_HDRINCL raw socket.
 * If needed, send several fragments.
 * @param rsock - raw socket
 * @param buf - data
 * @param len - data len
 * @param from - source address:port (_must_ be non-null, but the ip address
 *                can be 0, in which case it will be filled by the kernel).
 * @param to - destination address:port
 * @param mtu - maximum datagram size (including the ip header, excluding
 *              link layer headers). Minimum allowed size is 28
 *               (sizeof(ip_header + udp_header)). If mtu is lower, it will
 *               be ignored (the packet will be sent un-fragmented).
 *              0 can be used to disable fragmentation.
 * @return  <0 on error (-2: datagram too big, -1: check errno),
 *          number of bytes sent on success
 *          (including the ip & udp headers =>
 *               on success len + udpheader + ipheader size).
 */
int raw_iphdr_udp4_send(int rsock, const char* buf, unsigned int len,
			const sockaddr_storage* from,
			const sockaddr_storage* to,
			unsigned short mtu)
{
	struct msghdr snd_msg;
	struct iovec iov[2];
	struct ip_udp_hdr {
		struct ip ip;
		struct udphdr udp;
	} hdr;
	unsigned int totlen;
#ifndef RAW_IPHDR_INC_AUTO_FRAG
	unsigned int ip_frag_size; /* fragment size */
	unsigned int last_frag_extra; /* extra bytes possible in the last frag */
	unsigned int ip_payload;
	unsigned int last_frag_offs;
	void* last_frag_start;
	int frg_no;
#endif /* RAW_IPHDR_INC_AUTO_FRAG */
	int ret;

	totlen = len + sizeof(hdr);
	if (unlikely(totlen) > 65535)
		return -2;
	memset(&snd_msg, 0, sizeof(snd_msg));
	snd_msg.msg_name=SAv4(to);
	snd_msg.msg_namelen=SA_len(to);
	snd_msg.msg_iov=&iov[0];
	/* prepare the udp & ip headers */
	mk_udp_hdr(&hdr.udp, from, to, (unsigned char*)buf, len, 1);
	mk_ip_hdr(&hdr.ip, &SAv4(from)->sin_addr, &SAv4(to)->sin_addr,
		  len + sizeof(hdr.udp), IPPROTO_UDP);
	iov[0].iov_base=(char*)&hdr;
	iov[0].iov_len=sizeof(hdr);
	snd_msg.msg_iovlen=2;
	snd_msg.msg_control=0;
	snd_msg.msg_controllen=0;
	snd_msg.msg_flags=0;
	/* this part changes for different fragments */
	/* packets are fragmented if mtu has a valid value (at least an
	   IP header + UDP header fit in it) and if the total length is greater
	   then the mtu */
#ifndef RAW_IPHDR_INC_AUTO_FRAG
	if (likely(totlen <= mtu || mtu <= sizeof(hdr))) {
#endif /* RAW_IPHDR_INC_AUTO_FRAG */
	  iov[1].iov_base=(void*)buf;
	  iov[1].iov_len=len;
	  ret=sendmsg(rsock, &snd_msg, 0);
#ifndef RAW_IPHDR_INC_AUTO_FRAG
	} else {
	  int bytes_sent;
	  ip_payload = len + sizeof(hdr.udp);
	  /* a fragment offset must be a multiple of 8 => its size must
	     also be a multiple of 8, except for the last fragment */
	  ip_frag_size = (mtu -sizeof(hdr.ip)) & (~7);
	  last_frag_extra = (mtu - sizeof(hdr.ip)) & 7; /* rest */
	  frg_no = ip_payload / ip_frag_size +
	    ((ip_payload % ip_frag_size) > last_frag_extra);
	  /*ip_last_frag_size = ip_payload % frag_size +
	    ((ip_payload % frag_size) <= last_frag_extra) *
	    ip_frag_size; */
	  last_frag_offs = (frg_no - 1) * ip_frag_size;
	  /* if we are here mtu => sizeof(ip_h+udp_h) && payload > mtu
	     => last_frag_offs >= sizeof(hdr.udp) */
	  last_frag_start = (void*)(buf + last_frag_offs - sizeof(hdr.udp));
	  /* random id, should be != 0
	     (if 0 the kernel will fill it) */
	  hdr.ip.ip_id = 0; //fastrand_max(65534) + 1;
	  /* send the first fragment */
	  iov[1].iov_base=(void*)buf;
	  /* ip_frag_size >= sizeof(hdr.udp) because we are here only
	     if mtu >= sizeof(hdr.ip) + sizeof(hdr.udp) */
	  iov[1].iov_len=ip_frag_size - sizeof(hdr.udp);
	  hdr.ip.ip_len = RAW_IPHDR_IP_LEN(ip_frag_size + sizeof(hdr.ip));
	  hdr.ip.ip_off = RAW_IPHDR_IP_OFF(0x2000); /* set MF */
	  ret=sendmsg(rsock, &snd_msg, 0);
	  if (unlikely(ret < 0))
	    goto end;
	  bytes_sent = ret;
	  /* all the other fragments, include only the ip header */
	  iov[0].iov_len = sizeof(hdr.ip);
	  iov[1].iov_base =  (char*)iov[1].iov_base + iov[1].iov_len;
	  /* fragments between the first and the last */
	  while(unlikely(iov[1].iov_base < last_frag_start)) {
	    iov[1].iov_len = ip_frag_size;
	    hdr.ip.ip_len = RAW_IPHDR_IP_LEN(iov[1].iov_len + sizeof(hdr.ip));
	    /* set MF  */
	    hdr.ip.ip_off =
	      RAW_IPHDR_IP_OFF((unsigned short)
			       (((char*)iov[1].iov_base
				 - (char*)buf + sizeof(hdr.udp))
				/ 8) | 0x2000 );
	    ret=sendmsg(rsock, &snd_msg, 0);
	    if (unlikely(ret < 0))
	      goto end;
	    bytes_sent+=ret;
	    iov[1].iov_base =  (char*)iov[1].iov_base + iov[1].iov_len;
	  }
	  /* last fragment */
	  iov[1].iov_len = buf + len - (char*)iov[1].iov_base;
	  hdr.ip.ip_len = RAW_IPHDR_IP_LEN(iov[1].iov_len + sizeof(hdr.ip));
	  /* don't set MF (last fragment) */
	  hdr.ip.ip_off = RAW_IPHDR_IP_OFF((unsigned short)
					   (((char*)iov[1].iov_base
					     - (char*)buf + sizeof(hdr.udp))
					    / 8) );
	  ret=sendmsg(rsock, &snd_msg, 0);
	  if (unlikely(ret < 0))
	    goto end;
	  ret+=bytes_sent;
	}
end:
#endif /* RAW_IPHDR_INC_AUTO_FRAG */
	return ret;
}