ip_output.c

VXWORKS源代码

            WV_NET_ADDROUT_EVENT_2 (NET_CORE_EVENT, WV_NET_CRITICAL, 21, 2,
                                    ip->ip_src.s_addr, ip->ip_dst.s_addr,
                                    WV_NETEVENT_IPOUT_BADFLAGS, WV_NET_SEND,
                                    dst->sin_addr.s_addr, 0)
#endif  /* INCLUDE_WVNET */
#endif

			error = EACCES;
			goto bad;
		}
		/* allow broadcast messages to be fragmented if flag set */
		if (((u_short)ip->ip_len > mtu) &&
                    !(_ipCfgFlags & IP_DO_LARGE_BCAST)) {
#ifdef WV_INSTRUMENTATION
#ifdef INCLUDE_WVNET    /* WV_NET_CRITICAL event */
            WV_NET_ADDROUT_EVENT_2 (NET_CORE_EVENT, WV_NET_CRITICAL, 22, 3,
                                    ip->ip_src.s_addr, ip->ip_dst.s_addr,
                                    WV_NETEVENT_IPOUT_CANTFRAG, WV_NET_SEND,
                                    dst->sin_addr.s_addr, ifp)
#endif  /* INCLUDE_WVNET */
#endif

			error = EMSGSIZE;
			goto bad;
		}
		m->m_flags |= M_BCAST;
	} else
		m->m_flags &= ~M_BCAST;

sendit:
        if (_func_ipsecOutput != NULL)
            {
            /* Deliver the packet to ipsec for further processing/filtering. */
            error = _func_ipsecOutput (&m0, opt, ro, flags, imo, dst, ia);
            m = m0;
            ip = mtod(m, struct ip *);
            if (error != 0)
                {
                /* If there is an error we should have already freed the packet */
                if (error == ERROR)
                    {
                    /* Unless the error is very specific we do not propagate the error; so we reset it */
                    error = OK;
                    }
                goto done;
                }
            }
	/*
	 * If small enough for interface, can just send directly.
	 */
	if ((u_short)ip->ip_len <= mtu) {
#ifdef WV_INSTRUMENTATION
#ifdef INCLUDE_WVNET    /* WV_NET_INFO event */
            WV_NET_ADDROUT_EVENT_4 (NET_CORE_EVENT, WV_NET_INFO, 54, 10,
                                    ip->ip_src.s_addr, ip->ip_dst.s_addr,
                                    WV_NETEVENT_IPOUT_SENDALL, WV_NET_SEND,
                                    ip->ip_src.s_addr, ip->ip_dst.s_addr,
                                    (u_long)ip->ip_len, mtu)
#endif  /* INCLUDE_WVNET */
#endif
		ip->ip_len = htons((u_short)ip->ip_len);
		ip->ip_off = htons((u_short)ip->ip_off);

		ip->ip_sum = 0;
                if (_ipCfgFlags & IP_DO_CHECKSUM_SND)
                    ip->ip_sum = in_cksum(m, hlen);

		error = (*ifp->if_output)(ifp, m,
				(struct sockaddr *)dst, ro->ro_rt);
		goto done;
	}
	/*
	 * Too large for interface; fragment if possible.
	 * Must be able to put at least 8 bytes per fragment.
	 */
	if (ip->ip_off & IP_DF) {
#ifdef WV_INSTRUMENTATION
#ifdef INCLUDE_WVNET    /* WV_NET_CRITICAL event */
            WV_NET_ADDROUT_EVENT_2 (NET_CORE_EVENT, WV_NET_CRITICAL, 22, 3,
                                    ip->ip_src.s_addr, ip->ip_dst.s_addr,
                                    WV_NETEVENT_IPOUT_CANTFRAG, WV_NET_SEND,
                                    dst->sin_addr.s_addr, ifp)
#endif  /* INCLUDE_WVNET */
#endif

		error = EMSGSIZE;

            /*
             * If the interface is using a new (smaller) MTU size than
             * originally assigned, update the current path MTU estimate
             * (if any) to the new (largest possible) value.
             */

            if (ro->ro_rt && (ro->ro_rt->rt_flags & (RTF_UP | RTF_HOST))
                && !(ro->ro_rt->rt_rmx.rmx_locks & RTV_MTU)
                && (ro->ro_rt->rt_rmx.rmx_mtu > mtu))
                {
                ro->ro_rt->rt_rmx.rmx_mtu = mtu;
                }

#ifdef VIRTUAL_STACK
		_ipstat.ips_cantfrag++;
#else
		ipstat.ips_cantfrag++;
#endif /* VIRTUAL_STACK */
		goto bad;
	}
	len = (mtu - hlen) &~ 7;
	if (len < 8) {
#ifdef WV_INSTRUMENTATION
#ifdef INCLUDE_WVNET    /* WV_NET_CRITICAL event */
            WV_NET_ADDROUT_EVENT_2 (NET_CORE_EVENT, WV_NET_CRITICAL, 23, 4,
                                    ip->ip_src.s_addr, ip->ip_dst.s_addr,
                                    WV_NETEVENT_IPOUT_SHORTMSG, WV_NET_SEND,
                                    dst->sin_addr.s_addr, ifp)
#endif  /* INCLUDE_WVNET */
#endif

		error = EMSGSIZE;
		goto bad;
	}

    {
	int mhlen, firstlen = len;
	struct mbuf **mnext = &m->m_nextpkt;

#ifdef WV_INSTRUMENTATION
#ifdef INCLUDE_WVNET    /* WV_NET_INFO event */
        WV_NET_ADDROUT_EVENT_4 (NET_CORE_EVENT, WV_NET_INFO, 55, 11,
                                ip->ip_src.s_addr, ip->ip_dst.s_addr,
                                WV_NETEVENT_IPOUT_SENDFRAG, WV_NET_SEND,
                                dst->sin_addr.s_addr, (u_long)ip->ip_len,
                                hlen, mtu);
#endif  /* INCLUDE_WVNET */
#endif

	/*
	 * Loop through length of segment after first fragment,
	 * make new header and copy data of each part and link onto chain.
	 */
	m0 = m;
	mhlen = sizeof (struct ip);
	for (off = hlen + len; off < (u_short)ip->ip_len; off += len) {
		m= mHdrClGet(M_DONTWAIT, MT_HEADER, CL_SIZE_128, TRUE);
		if (m == 0) {
			error = ENOBUFS;
#ifdef VIRTUAL_STACK
			_ipstat.ips_odropped++;
#else
			ipstat.ips_odropped++;
#endif /* VIRTUAL_STACK */
			goto sendorfree;
		}
		m->m_flags = m0->m_flags;
		m->m_data += max_linkhdr;
		mhip = mtod(m, struct ip *);
		*mhip = *ip;
		if (hlen > sizeof (struct ip)) {
			mhlen = ip_optcopy(ip, mhip) + sizeof (struct ip);
			mhip->ip_hl = mhlen >> 2;
		}
		m->m_len = mhlen;
		mhip->ip_off = ((off - hlen) >> 3) + (ip->ip_off & ~IP_MF);
		if (ip->ip_off & IP_MF)
			mhip->ip_off |= IP_MF;
		if (off + len >= (u_short)ip->ip_len)
			len = (u_short)ip->ip_len - off;
		else
			mhip->ip_off |= IP_MF;
		mhip->ip_len = htons((u_short)(len + mhlen));
		m->m_next = m_copy(m0, off, len);
		if (m->m_next == 0) {
			(void) m_free(m);
			error = ENOBUFS;	/* ??? */
#ifdef VIRTUAL_STACK
			_ipstat.ips_odropped++;
#else
			ipstat.ips_odropped++;
#endif /* VIRTUAL_STACK */
			goto sendorfree;
		}
		m->m_pkthdr.len = mhlen + len;
		m->m_pkthdr.rcvif = (struct ifnet *)0;
		mhip->ip_off = htons((u_short)mhip->ip_off);

                mhip->ip_sum = 0;
                if (_ipCfgFlags & IP_DO_CHECKSUM_SND)
                    mhip->ip_sum = in_cksum(m, mhlen);

		*mnext = m;
		mnext = &m->m_nextpkt;
#ifdef VIRTUAL_STACK
		_ipstat.ips_ofragments++;
#else
		ipstat.ips_ofragments++;
#endif /* VIRTUAL_STACK */
	}
	/*
	 * Update first fragment by trimming what's been copied out
	 * and updating header, then send each fragment (in order).
	 */
	m = m0;
	m_adj(m, hlen + firstlen - (u_short)ip->ip_len);
	m->m_pkthdr.len = hlen + firstlen;
	ip->ip_len = htons((u_short)m->m_pkthdr.len);
	ip->ip_off = htons((u_short)(ip->ip_off | IP_MF));

        ip->ip_sum = 0;
        if (_ipCfgFlags & IP_DO_CHECKSUM_SND)
            ip->ip_sum = in_cksum(m, hlen);
sendorfree:
	for (m = m0; m; m = m0) {
		m0 = m->m_nextpkt;
		m->m_nextpkt = 0;
		if (error == 0)
			error = (*ifp->if_output)(ifp, m,
			    (struct sockaddr *)dst, ro->ro_rt);
		else
			m_freem(m);
	}

	if (error == 0)
#ifdef VIRTUAL_STACK
		_ipstat.ips_fragmented++;
#else
		ipstat.ips_fragmented++;
#endif /* VIRTUAL_STACK */
    }
done:

#ifdef WV_INSTRUMENTATION
#ifdef INCLUDE_WVNET    /* WV_NET_ERROR event */
    WV_NET_ADDROUT_EVENT_4 (NET_CORE_EVENT, WV_NET_ERROR, 31, 9,
                            srcAddr, dstAddr,
                            WV_NETEVENT_IPOUT_FINISH, WV_NET_SEND,
                            error, ifp, srcAddr, dstAddr)
#endif  /* INCLUDE_WVNET */
#endif

	if ((ro == &iproute) && ((flags & IP_ROUTETOIF) == 0) && ro->ro_rt) {
		RTFREE(ro->ro_rt);
	}	
	return (error);
bad:
	m_freem(m);
	goto done;
}

/*
 * Insert IP options into preformed packet.
 * Adjust IP destination as required for IP source routing,
 * as indicated by a non-zero in_addr at the start of the options.
 */
static struct mbuf *
ip_insertoptions(m, opt, phlen)
	register struct mbuf *m;
	struct mbuf *opt;
	int *phlen;
{
	register struct ipoption *p = mtod(opt, struct ipoption *);
	struct mbuf *n;
	register struct ip *ip = mtod(m, struct ip *);
	unsigned optlen;

	optlen = opt->m_len - sizeof(p->ipopt_dst);
	if (optlen + (u_short)ip->ip_len > IP_MAXPACKET)
		return (m);		/* XXX should fail */
	if (p->ipopt_dst.s_addr)
		ip->ip_dst = p->ipopt_dst;
#if 0	/* XXX changed for default cluster support vinai
         * The reason why always new clusters are allocted,
         * is that, data could be overwritten since clusters are shared.
         */
	if (m->m_flags & M_EXT || m->m_data - optlen < m->m_pktdat) {
#endif
		n = mHdrClGet(M_DONTWAIT, MT_HEADER, CL_SIZE_128, TRUE);
		if (n == 0)
			return (m);
		n->m_pkthdr.len = m->m_pkthdr.len + optlen;
		m->m_len -= sizeof(struct ip);
		m->m_data += sizeof(struct ip);
		n->m_next = m;
		m = n;
		m->m_len = optlen + sizeof(struct ip);
		m->m_data += max_linkhdr;
		bcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip));
#if 0 	/* XXX changed for default cluster support vinai */               
	} else {
		m->m_data -= optlen;
		m->m_len += optlen;
		m->m_pkthdr.len += optlen;
		ovbcopy((caddr_t)ip, mtod(m, caddr_t), sizeof(struct ip));
	}
#endif        
	ip = mtod(m, struct ip *);
	bcopy((caddr_t)p->ipopt_list, (caddr_t)(ip + 1), (unsigned)optlen);
	*phlen = sizeof(struct ip) + optlen;
	ip->ip_len += optlen;
	return (m);
}

/*
 * Copy options from ip to jp,
 * omitting those not copied during fragmentation.
 */
int
ip_optcopy(ip, jp)
	struct ip *ip, *jp;
{
	register u_char *cp, *dp;
	int opt, optlen, cnt;

	cp = (u_char *)(ip + 1);
	dp = (u_char *)(jp + 1);
	cnt = (ip->ip_hl << 2) - sizeof (struct ip);
	for (; cnt > 0; cnt -= optlen, cp += optlen) {
		opt = cp[0];
		if (opt == IPOPT_EOL)
			break;
		if (opt == IPOPT_NOP) {
			/* Preserve for IP mcast tunnel's LSRR alignment. */
			*dp++ = IPOPT_NOP;
			optlen = 1;
			continue;
		} else
			optlen = cp[IPOPT_OLEN];
		/* bogus lengths should have been caught by ip_dooptions */
		if (optlen > cnt)
			optlen = cnt;
		if (IPOPT_COPIED(opt)) {
			bcopy((caddr_t)cp, (caddr_t)dp, (unsigned)optlen);
			dp += optlen;
		}
	}
	for (optlen = dp - (u_char *)(jp+1); optlen & 0x3; optlen++)
		*dp++ = IPOPT_EOL;
	return (optlen);
}

/*
 * IP socket option processing.
 */
int
ip_ctloutput(op, so, level, optname, mp)
	int op;
	struct socket *so;
	int level, optname;
	struct mbuf **mp;
{
	register struct inpcb *inp = sotoinpcb(so);
	register struct mbuf *m = *mp;
	register int optval = 0;
	int error = 0;

#ifdef WV_INSTRUMENTATION
#ifdef INCLUDE_WVNET    /* WV_NET_INFO event */
    WV_NET_MARKER_4 (NET_AUX_EVENT, WV_NET_INFO, 11, 6,
                     WV_NETEVENT_IPOUTCTRL_START,
                     so->so_fd, op, level, optname)
#endif  /* INCLUDE_WVNET */
#endif

	if (level != IPPROTO_IP) {
		error = EINVAL;
		if (op == PRCO_SETOPT && *mp)
			(void) m_free(*mp);
	} else switch (op) {

	case PRCO_SETOPT:
		switch (optname) {
		case IP_OPTIONS:
#ifdef notyet
		case IP_RETOPTS:
			return (ip_pcbopts(optname, &inp->inp_options, m));
#else
			return (ip_pcbopts(&inp->inp_options, m));
#endif

		case IP_TOS:
		case IP_TTL:
		case IP_RECVOPTS:
		case IP_RECVRETOPTS:
		case IP_RECVDSTADDR:
			if (m->m_len != sizeof(int))
				error = EINVAL;
			else {
				optval = *mtod(m, int *);
				switch (optname) {

				case IP_TOS:
					inp->inp_ip.ip_tos = optval;
					break;

				case IP_TTL:
					inp->inp_ip.ip_ttl = optval;
					break;
#define	OPTSET(bit) \
	if (optval) \
		inp->inp_flags |= bit; \
	else \
		inp->inp_flags &= ~bit;

				case IP_RECVOPTS:
					OPTSET(INP_RECVOPTS);
					break;

				case IP_RECVRETOPTS:
					OPTSET(INP_RECVRETOPTS);
					break;

				case IP_RECVDSTADDR:
					OPTSET(INP_RECVDSTADDR);
					break;
				}
			}
			break;
#undef OPTSET

		case IP_MULTICAST_IF:
		case IP_MULTICAST_TTL:
		case IP_MULTICAST_LOOP:
		case IP_ADD_MEMBERSHIP:
		case IP_DROP_MEMBERSHIP:
			error = ip_setmoptions(optname, inp, m);
			break;

		default:
#ifdef WV_INSTRUMENTATION
#ifdef INCLUDE_WVNET    /* WV_NET_ERROR event */
                    WV_NET_MARKER_4 (NET_AUX_EVENT, WV_NET_ERROR, 30, 5,
                                     WV_NETEVENT_IPOUTCTRL_BADCMD,
                                     so->so_fd, op, level, optname)
#endif  /* INCLUDE_WVNET */
#endif

			error = ENOPROTOOPT;
			break;
		}
		if (m)
			(void)m_free(m);
		break;

	case PRCO_GETOPT:
		switch (optname) {
		case IP_OPTIONS:
		case IP_RETOPTS:
			*mp = m = mBufClGet(M_WAIT, MT_SOOPTS, CL_SIZE_128,
					    TRUE);
			if (m == NULL)
			    {
			    error = ENOBUFS;
			    break;
			    }

			if (inp->inp_options) {
				m->m_len = inp->inp_options->m_len;
				bcopy(mtod(inp->inp_options, caddr_t),
				    mtod(m, caddr_t), (unsigned)m->m_len);
			} else
				m->m_len = 0;
			break;

		case IP_TOS:
		case IP_TTL:
		case IP_RECVOPTS:
		case IP_RECVRETOPTS:
		case IP_RECVDSTADDR:
			*mp = m = mBufClGet(M_WAIT, MT_SOOPTS, CL_SIZE_128,
					    TRUE);
			if (m == NULL)
			    {
			    error = ENOBUFS;
			    break;
			    }

			m->m_len = sizeof(int);
			switch (optname) {

			case IP_TOS:
				optval = inp->inp_ip.ip_tos;
				break;

			case IP_TTL:
				optval = inp->inp_ip.ip_ttl;
				break;

#define	OPTBIT(bit)	(inp->inp_flags & bit ? 1 : 0)

			case IP_RECVOPTS:
				optval = OPTBIT(INP_RECVOPTS);
				break;

			case IP_RECVRETOPTS:
				optval = OPTBIT(INP_RECVRETOPTS);
				break;

			case IP_RECVDSTADDR:
				optval = OPTBIT(INP_RECVDSTADDR);
				break;
			}
			*mtod(m, int *) = optval;
			break;

		case IP_MULTICAST_IF:
		case IP_MULTICAST_TTL:
		case IP_MULTICAST_LOOP:
		case IP_ADD_MEMBERSHIP:
		case IP_DROP_MEMBERSHIP:
			error = ip_getmoptions(optname, inp->inp_moptions, mp);
			break;

		default:
#ifdef WV_INSTRUMENTATION
#ifdef INCLUDE_WVNET    /* WV_NET_ERROR event */
                    WV_NET_MARKER_4 (NET_AUX_EVENT, WV_NET_ERROR, 30, 5,
                                     WV_NETEVENT_IPOUTCTRL_BADCMD,
                                     so->so_fd, op, level, optname)
#endif  /* INCLUDE_WVNET */
#endif

			error = ENOPROTOOPT;