int
udp_input(struct mbuf **mp, int *offp, int proto)
{
struct sockaddr_in udp_in = { sizeof udp_in, AF_INET };
int iphlen;
struct ip *ip;
struct udphdr *uh;
struct inpcb *inp;
struct mbuf *m;
struct mbuf *opts = NULL;
int len, off;
struct ip save_ip;
struct inpcbinfo *pcbinfo = &udbinfo[mycpuid];
off = *offp;
m = *mp;
*mp = NULL;
iphlen = off;
udp_stat.udps_ipackets++;
/*
* Strip IP options, if any; should skip this,
* make available to user, and use on returned packets,
* but we don't yet have a way to check the checksum
* with options still present.
*/
if (iphlen > sizeof(struct ip)) {
ip_stripoptions(m);
iphlen = sizeof(struct ip);
}
/*
* IP and UDP headers are together in first mbuf.
* Already checked and pulled up in ip_demux().
*/
KASSERT(m->m_len >= iphlen + sizeof(struct udphdr),
("UDP header not in one mbuf"));
ip = mtod(m, struct ip *);
uh = (struct udphdr *)((caddr_t)ip + iphlen);
/* destination port of 0 is illegal, based on RFC768. */
if (uh->uh_dport == 0)
goto bad;
/*
* Make mbuf data length reflect UDP length.
* If not enough data to reflect UDP length, drop.
*/
len = ntohs((u_short)uh->uh_ulen);
if (ip->ip_len != len) {
if (len > ip->ip_len || len < sizeof(struct udphdr)) {
udp_stat.udps_badlen++;
goto bad;
}
m_adj(m, len - ip->ip_len);
/* ip->ip_len = len; */
}
/*
* Save a copy of the IP header in case we want restore it
* for sending an ICMP error message in response.
*/
save_ip = *ip;
/*
* Checksum extended UDP header and data.
*/
if (uh->uh_sum) {
if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) {
if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR)
uh->uh_sum = m->m_pkthdr.csum_data;
else
uh->uh_sum = in_pseudo(ip->ip_src.s_addr,
ip->ip_dst.s_addr, htonl((u_short)len +
m->m_pkthdr.csum_data + IPPROTO_UDP));
uh->uh_sum ^= 0xffff;
} else {
char b[9];
bcopy(((struct ipovly *)ip)->ih_x1, b, 9);
bzero(((struct ipovly *)ip)->ih_x1, 9);
((struct ipovly *)ip)->ih_len = uh->uh_ulen;
uh->uh_sum = in_cksum(m, len + sizeof(struct ip));
bcopy(b, ((struct ipovly *)ip)->ih_x1, 9);
}
if (uh->uh_sum) {
udp_stat.udps_badsum++;
m_freem(m);
return(IPPROTO_DONE);
}
} else
udp_stat.udps_nosum++;
if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) {
struct inpcbhead *connhead;
struct inpcontainer *ic, *ic_marker;
struct inpcontainerhead *ichead;
struct udp_mcast_arg arg;
struct inpcb *last;
int error;
/*
* Deliver a multicast or broadcast datagram to *all* sockets
* for which the local and remote addresses and ports match
* those of the incoming datagram. This allows more than
* one process to receive multi/broadcasts on the same port.
* (This really ought to be done for unicast datagrams as
* well, but that would cause problems with existing
* applications that open both address-specific sockets and
* a wildcard socket listening to the same port -- they would
* end up receiving duplicates of every unicast datagram.
* Those applications open the multiple sockets to overcome an
* inadequacy of the UDP socket interface, but for backwards
* compatibility we avoid the problem here rather than
* fixing the interface. Maybe 4.5BSD will remedy this?)
*/
/*
* Construct sockaddr format source address.
*/
udp_in.sin_port = uh->uh_sport;
udp_in.sin_addr = ip->ip_src;
arg.udp_in = &udp_in;
/*
* Locate pcb(s) for datagram.
* (Algorithm copied from raw_intr().)
*/
last = NULL;
arg.iphlen = iphlen;
connhead = &pcbinfo->hashbase[
INP_PCBCONNHASH(ip->ip_src.s_addr, uh->uh_sport,
ip->ip_dst.s_addr, uh->uh_dport, pcbinfo->hashmask)];
LIST_FOREACH(inp, connhead, inp_hash) {
#ifdef INET6
if (!INP_ISIPV4(inp))
continue;
#endif
if (!in_hosteq(inp->inp_faddr, ip->ip_src) ||
!in_hosteq(inp->inp_laddr, ip->ip_dst) ||
inp->inp_fport != uh->uh_sport ||
inp->inp_lport != uh->uh_dport)
continue;
arg.inp = inp;
arg.last = last;
arg.ip = ip;
arg.m = m;
error = udp_mcast_input(&arg);
if (error == ERESTART)
continue;
last = arg.last;
if (error == EJUSTRETURN)
goto done;
}
ichead = &pcbinfo->wildcardhashbase[
INP_PCBWILDCARDHASH(uh->uh_dport,
pcbinfo->wildcardhashmask)];
ic_marker = in_pcbcontainer_marker(mycpuid);
GET_PCBINFO_TOKEN(pcbinfo);
LIST_INSERT_HEAD(ichead, ic_marker, ic_list);
while ((ic = LIST_NEXT(ic_marker, ic_list)) != NULL) {
LIST_REMOVE(ic_marker, ic_list);
LIST_INSERT_AFTER(ic, ic_marker, ic_list);
inp = ic->ic_inp;
if (inp->inp_flags & INP_PLACEMARKER)
continue;
#ifdef INET6
if (!INP_ISIPV4(inp))
continue;
#endif
if (inp->inp_lport != uh->uh_dport)
continue;
if (inp->inp_laddr.s_addr != INADDR_ANY &&
inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
continue;
arg.inp = inp;
arg.last = last;
arg.ip = ip;
arg.m = m;
error = udp_mcast_input(&arg);
if (error == ERESTART)
continue;
last = arg.last;
if (error == EJUSTRETURN)
break;
}
LIST_REMOVE(ic_marker, ic_list);
REL_PCBINFO_TOKEN(pcbinfo);
done:
if (last == NULL) {
/*
* No matching pcb found; discard datagram.
* (No need to send an ICMP Port Unreachable
* for a broadcast or multicast datgram.)
*/
udp_stat.udps_noportbcast++;
goto bad;
}
#ifdef IPSEC
/* check AH/ESP integrity. */
if (ipsec4_in_reject_so(m, last->inp_socket)) {
ipsecstat.in_polvio++;
goto bad;
}
#endif /*IPSEC*/
#ifdef FAST_IPSEC
/* check AH/ESP integrity. */
if (ipsec4_in_reject(m, last))
goto bad;
#endif /*FAST_IPSEC*/
udp_append(last, ip, m, iphlen + sizeof(struct udphdr),
&udp_in);
return(IPPROTO_DONE);
}
/*
* Locate pcb for datagram.
*/
inp = in_pcblookup_pkthash(pcbinfo, ip->ip_src, uh->uh_sport,
ip->ip_dst, uh->uh_dport, TRUE, m->m_pkthdr.rcvif,
udp_reuseport_ext ? m : NULL);
if (inp == NULL) {
if (log_in_vain) {
char buf[sizeof "aaa.bbb.ccc.ddd"];
strcpy(buf, inet_ntoa(ip->ip_dst));
log(LOG_INFO,
"Connection attempt to UDP %s:%d from %s:%d\n",
buf, ntohs(uh->uh_dport), inet_ntoa(ip->ip_src),
ntohs(uh->uh_sport));
}
udp_stat.udps_noport++;
if (m->m_flags & (M_BCAST | M_MCAST)) {
udp_stat.udps_noportbcast++;
goto bad;
}
if (blackhole)
goto bad;
#ifdef ICMP_BANDLIM
if (badport_bandlim(BANDLIM_ICMP_UNREACH) < 0)
goto bad;
#endif
*ip = save_ip;
ip->ip_len += iphlen;
icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0);
return(IPPROTO_DONE);
}
KASSERT(INP_ISIPV4(inp), ("not inet inpcb"));
#ifdef IPSEC
if (ipsec4_in_reject_so(m, inp->inp_socket)) {
ipsecstat.in_polvio++;
goto bad;
}
#endif /*IPSEC*/
#ifdef FAST_IPSEC
if (ipsec4_in_reject(m, inp))
goto bad;
#endif /*FAST_IPSEC*/
/*
* Check the minimum TTL for socket.
*/
if (ip->ip_ttl < inp->inp_ip_minttl)
goto bad;
/*
* Construct sockaddr format source address.
* Stuff source address and datagram in user buffer.
*/
udp_in.sin_port = uh->uh_sport;
udp_in.sin_addr = ip->ip_src;
if ((inp->inp_flags & INP_CONTROLOPTS) ||
(inp->inp_socket->so_options & SO_TIMESTAMP))
ip_savecontrol(inp, &opts, ip, m);
m_adj(m, iphlen + sizeof(struct udphdr));
lwkt_gettoken(&inp->inp_socket->so_rcv.ssb_token);
if (ssb_appendaddr(&inp->inp_socket->so_rcv,
(struct sockaddr *)&udp_in, m, opts) == 0) {
lwkt_reltoken(&inp->inp_socket->so_rcv.ssb_token);
udp_stat.udps_fullsock++;
goto bad;
}
lwkt_reltoken(&inp->inp_socket->so_rcv.ssb_token);
sorwakeup(inp->inp_socket);
return(IPPROTO_DONE);
bad:
m_freem(m);
if (opts)
m_freem(opts);
return(IPPROTO_DONE);
}
/*
* TCP input routine, follows pages 65-76 of the
* protocol specification dated September, 1981 very closely.
*/
void
tcp_input(struct mbuf *m, int iphlen, struct socket *inso)
{
struct ip save_ip, *ip;
register struct tcpiphdr *ti;
caddr_t optp = NULL;
int optlen = 0;
int len, tlen, off;
register struct tcpcb *tp = NULL;
register int tiflags;
struct socket *so = NULL;
int todrop, acked, ourfinisacked, needoutput = 0;
int iss = 0;
u_long tiwin;
int ret;
struct ex_list *ex_ptr;
Slirp *slirp;
DEBUG_CALL("tcp_input");
DEBUG_ARGS((dfd, " m = %8lx iphlen = %2d inso = %lx\n",
(long )m, iphlen, (long )inso ));
/*
* If called with m == 0, then we're continuing the connect
*/
if (m == NULL) {
so = inso;
slirp = so->slirp;
/* Re-set a few variables */
tp = sototcpcb(so);
m = so->so_m;
so->so_m = NULL;
ti = so->so_ti;
tiwin = ti->ti_win;
tiflags = ti->ti_flags;
goto cont_conn;
}
slirp = m->slirp;
/*
* Get IP and TCP header together in first mbuf.
* Note: IP leaves IP header in first mbuf.
*/
ti = mtod(m, struct tcpiphdr *);
if (iphlen > sizeof(struct ip )) {
ip_stripoptions(m, (struct mbuf *)0);
iphlen=sizeof(struct ip );
}
/* XXX Check if too short */
/*
* Save a copy of the IP header in case we want restore it
* for sending an ICMP error message in response.
*/
ip=mtod(m, struct ip *);
save_ip = *ip;
save_ip.ip_len+= iphlen;
/*
* Checksum extended TCP header and data.
*/
tlen = ((struct ip *)ti)->ip_len;
tcpiphdr2qlink(ti)->next = tcpiphdr2qlink(ti)->prev = NULL;
memset(&ti->ti_i.ih_mbuf, 0 , sizeof(struct mbuf_ptr));
ti->ti_x1 = 0;
ti->ti_len = htons((uint16_t)tlen);
len = sizeof(struct ip ) + tlen;
if(cksum(m, len)) {
goto drop;
}
/*
* Check that TCP offset makes sense,
* pull out TCP options and adjust length. XXX
*/
off = ti->ti_off << 2;
if (off < sizeof (struct tcphdr) || off > tlen) {
goto drop;
}
tlen -= off;
ti->ti_len = tlen;
if (off > sizeof (struct tcphdr)) {
optlen = off - sizeof (struct tcphdr);
optp = mtod(m, caddr_t) + sizeof (struct tcpiphdr);
}
tiflags = ti->ti_flags;
/*
* Convert TCP protocol specific fields to host format.
*/
NTOHL(ti->ti_seq);
NTOHL(ti->ti_ack);
NTOHS(ti->ti_win);
NTOHS(ti->ti_urp);
/*
* Drop TCP, IP headers and TCP options.
*/
m->m_data += sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
m->m_len -= sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
/*
* Locate pcb for segment.
*/
findso:
so = slirp->tcp_last_so;
if (so->so_fport != ti->ti_dport ||
so->so_lport != ti->ti_sport ||
so->so_laddr.s_addr != ti->ti_src.s_addr ||
so->so_faddr.s_addr != ti->ti_dst.s_addr) {
so = solookup(&slirp->tcb, ti->ti_src, ti->ti_sport,
ti->ti_dst, ti->ti_dport);
if (so)
slirp->tcp_last_so = so;
}
/*
* If the state is CLOSED (i.e., TCB does not exist) then
* all data in the incoming segment is discarded.
* If the TCB exists but is in CLOSED state, it is embryonic,
* but should either do a listen or a connect soon.
*
* state == CLOSED means we've done socreate() but haven't
* attached it to a protocol yet...
*
* XXX If a TCB does not exist, and the TH_SYN flag is
* the only flag set, then create a session, mark it
* as if it was LISTENING, and continue...
*/
if (so == NULL) {
if (slirp->restricted) {
/* Any hostfwds will have an existing socket, so we only get here
* for non-hostfwd connections. These should be dropped, unless it
* happens to be a guestfwd.
*/
for (ex_ptr = slirp->exec_list; ex_ptr; ex_ptr = ex_ptr->ex_next) {
if (ex_ptr->ex_fport == ti->ti_dport &&
ti->ti_dst.s_addr == ex_ptr->ex_addr.s_addr) {
break;
}
}
if (!ex_ptr) {
goto dropwithreset;
}
}
if ((tiflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ACK)) != TH_SYN)
goto dropwithreset;
if ((so = socreate(slirp)) == NULL)
goto dropwithreset;
if (tcp_attach(so) < 0) {
free(so); /* Not sofree (if it failed, it's not insqued) */
goto dropwithreset;
}
sbreserve(&so->so_snd, TCP_SNDSPACE);
sbreserve(&so->so_rcv, TCP_RCVSPACE);
so->so_laddr = ti->ti_src;
so->so_lport = ti->ti_sport;
so->so_faddr = ti->ti_dst;
so->so_fport = ti->ti_dport;
if ((so->so_iptos = tcp_tos(so)) == 0)
so->so_iptos = ((struct ip *)ti)->ip_tos;
tp = sototcpcb(so);
tp->t_state = TCPS_LISTEN;
}
/*
* If this is a still-connecting socket, this probably
* a retransmit of the SYN. Whether it's a retransmit SYN
* or something else, we nuke it.
*/
if (so->so_state & SS_ISFCONNECTING)
goto drop;
tp = sototcpcb(so);
/* XXX Should never fail */
if (tp == NULL)
goto dropwithreset;
if (tp->t_state == TCPS_CLOSED)
goto drop;
tiwin = ti->ti_win;
/*
* Segment received on connection.
* Reset idle time and keep-alive timer.
*/
tp->t_idle = 0;
if (SO_OPTIONS)
tp->t_timer[TCPT_KEEP] = TCPTV_KEEPINTVL;
else
tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_IDLE;
/*
* Process options if not in LISTEN state,
* else do it below (after getting remote address).
*/
if (optp && tp->t_state != TCPS_LISTEN)
tcp_dooptions(tp, (u_char *)optp, optlen, ti);
/*
* Header prediction: check for the two common cases
* of a uni-directional data xfer. If the packet has
* no control flags, is in-sequence, the window didn't
* change and we're not retransmitting, it's a
* candidate. If the length is zero and the ack moved
* forward, we're the sender side of the xfer. Just
* free the data acked & wake any higher level process
* that was blocked waiting for space. If the length
* is non-zero and the ack didn't move, we're the
* receiver side. If we're getting packets in-order
* (the reassembly queue is empty), add the data to
* the socket buffer and note that we need a delayed ack.
*
* XXX Some of these tests are not needed
* eg: the tiwin == tp->snd_wnd prevents many more
* predictions.. with no *real* advantage..
*/
if (tp->t_state == TCPS_ESTABLISHED &&
(tiflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ACK)) == TH_ACK &&
ti->ti_seq == tp->rcv_nxt &&
tiwin && tiwin == tp->snd_wnd &&
tp->snd_nxt == tp->snd_max) {
if (ti->ti_len == 0) {
if (SEQ_GT(ti->ti_ack, tp->snd_una) &&
SEQ_LEQ(ti->ti_ack, tp->snd_max) &&
tp->snd_cwnd >= tp->snd_wnd) {
/*
* this is a pure ack for outstanding data.
*/
if (tp->t_rtt &&
SEQ_GT(ti->ti_ack, tp->t_rtseq))
tcp_xmit_timer(tp, tp->t_rtt);
acked = ti->ti_ack - tp->snd_una;
sbdrop(&so->so_snd, acked);
tp->snd_una = ti->ti_ack;
m_free(m);
/*
* If all outstanding data are acked, stop
* retransmit timer, otherwise restart timer
* using current (possibly backed-off) value.
* If process is waiting for space,
* wakeup/selwakeup/signal. If data
* are ready to send, let tcp_output
* decide between more output or persist.
*/
if (tp->snd_una == tp->snd_max)
tp->t_timer[TCPT_REXMT] = 0;
else if (tp->t_timer[TCPT_PERSIST] == 0)
tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
/*
* This is called because sowwakeup might have
* put data into so_snd. Since we don't so sowwakeup,
* we don't need this.. XXX???
*/
if (so->so_snd.sb_cc)
(void) tcp_output(tp);
return;
}
} else if (ti->ti_ack == tp->snd_una &&
tcpfrag_list_empty(tp) &&
ti->ti_len <= sbspace(&so->so_rcv)) {
/*
* this is a pure, in-sequence data packet
* with nothing on the reassembly queue and
* we have enough buffer space to take it.
*/
tp->rcv_nxt += ti->ti_len;
/*
* Add data to socket buffer.
*/
if (so->so_emu) {
if (tcp_emu(so,m)) sbappend(so, m);
} else
sbappend(so, m);
/*
* If this is a short packet, then ACK now - with Nagel
* congestion avoidance sender won't send more until
* he gets an ACK.
*
* It is better to not delay acks at all to maximize
* TCP throughput. See RFC 2581.
*/
tp->t_flags |= TF_ACKNOW;
tcp_output(tp);
return;
}
} /* header prediction */
/*
* Calculate amount of space in receive window,
* and then do TCP input processing.
* Receive window is amount of space in rcv queue,
* but not less than advertised window.
*/
{ int win;
win = sbspace(&so->so_rcv);
if (win < 0)
win = 0;
tp->rcv_wnd = max(win, (int)(tp->rcv_adv - tp->rcv_nxt));
}
switch (tp->t_state) {
/*
* If the state is LISTEN then ignore segment if it contains an RST.
* If the segment contains an ACK then it is bad and send a RST.
* If it does not contain a SYN then it is not interesting; drop it.
* Don't bother responding if the destination was a broadcast.
* Otherwise initialize tp->rcv_nxt, and tp->irs, select an initial
* tp->iss, and send a segment:
* <SEQ=ISS><ACK=RCV_NXT><CTL=SYN,ACK>
* Also initialize tp->snd_nxt to tp->iss+1 and tp->snd_una to tp->iss.
* Fill in remote peer address fields if not previously specified.
* Enter SYN_RECEIVED state, and process any other fields of this
* segment in this state.
*/
case TCPS_LISTEN: {
if (tiflags & TH_RST)
goto drop;
if (tiflags & TH_ACK)
goto dropwithreset;
if ((tiflags & TH_SYN) == 0)
goto drop;
/*
* This has way too many gotos...
* But a bit of spaghetti code never hurt anybody :)
*/
/*
* If this is destined for the control address, then flag to
* tcp_ctl once connected, otherwise connect
*/
if ((so->so_faddr.s_addr & slirp->vnetwork_mask.s_addr) ==
slirp->vnetwork_addr.s_addr) {
if (so->so_faddr.s_addr != slirp->vhost_addr.s_addr &&
so->so_faddr.s_addr != slirp->vnameserver_addr.s_addr) {
/* May be an add exec */
for (ex_ptr = slirp->exec_list; ex_ptr;
ex_ptr = ex_ptr->ex_next) {
if(ex_ptr->ex_fport == so->so_fport &&
so->so_faddr.s_addr == ex_ptr->ex_addr.s_addr) {
so->so_state |= SS_CTL;
break;
}
}
if (so->so_state & SS_CTL) {
goto cont_input;
}
}
/* CTL_ALIAS: Do nothing, tcp_fconnect will be called on it */
}
if (so->so_emu & EMU_NOCONNECT) {
so->so_emu &= ~EMU_NOCONNECT;
goto cont_input;
}
if ((tcp_fconnect(so) == -1) &&
#if defined(_WIN32)
socket_error() != WSAEWOULDBLOCK
#else
(errno != EINPROGRESS) && (errno != EWOULDBLOCK)
#endif
) {
u_char code=ICMP_UNREACH_NET;
DEBUG_MISC((dfd, " tcp fconnect errno = %d-%s\n",
errno,strerror(errno)));
if(errno == ECONNREFUSED) {
/* ACK the SYN, send RST to refuse the connection */
tcp_respond(tp, ti, m, ti->ti_seq+1, (tcp_seq)0,
TH_RST|TH_ACK);
} else {
if(errno == EHOSTUNREACH) code=ICMP_UNREACH_HOST;
HTONL(ti->ti_seq); /* restore tcp header */
HTONL(ti->ti_ack);
HTONS(ti->ti_win);
HTONS(ti->ti_urp);
m->m_data -= sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
m->m_len += sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
*ip=save_ip;
icmp_error(m, ICMP_UNREACH,code, 0,strerror(errno));
}
tcp_close(tp);
m_free(m);
} else {
/*
* Haven't connected yet, save the current mbuf
* and ti, and return
* XXX Some OS's don't tell us whether the connect()
* succeeded or not. So we must time it out.
*/
so->so_m = m;
so->so_ti = ti;
tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_INIT;
tp->t_state = TCPS_SYN_RECEIVED;
tcp_template(tp);
}
return;
cont_conn:
/* m==NULL
* Check if the connect succeeded
*/
if (so->so_state & SS_NOFDREF) {
tp = tcp_close(tp);
goto dropwithreset;
}
cont_input:
tcp_template(tp);
if (optp)
tcp_dooptions(tp, (u_char *)optp, optlen, ti);
if (iss)
tp->iss = iss;
else
tp->iss = slirp->tcp_iss;
slirp->tcp_iss += TCP_ISSINCR/2;
tp->irs = ti->ti_seq;
tcp_sendseqinit(tp);
tcp_rcvseqinit(tp);
tp->t_flags |= TF_ACKNOW;
tp->t_state = TCPS_SYN_RECEIVED;
tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_INIT;
goto trimthenstep6;
} /* case TCPS_LISTEN */
/*
* If the state is SYN_SENT:
* if seg contains an ACK, but not for our SYN, drop the input.
* if seg contains a RST, then drop the connection.
* if seg does not contain SYN, then drop it.
* Otherwise this is an acceptable SYN segment
* initialize tp->rcv_nxt and tp->irs
* if seg contains ack then advance tp->snd_una
* if SYN has been acked change to ESTABLISHED else SYN_RCVD state
* arrange for segment to be acked (eventually)
* continue processing rest of data/controls, beginning with URG
*/
case TCPS_SYN_SENT:
if ((tiflags & TH_ACK) &&
(SEQ_LEQ(ti->ti_ack, tp->iss) ||
SEQ_GT(ti->ti_ack, tp->snd_max)))
goto dropwithreset;
if (tiflags & TH_RST) {
if (tiflags & TH_ACK) {
tcp_drop(tp, 0); /* XXX Check t_softerror! */
}
goto drop;
}
if ((tiflags & TH_SYN) == 0)
goto drop;
if (tiflags & TH_ACK) {
tp->snd_una = ti->ti_ack;
if (SEQ_LT(tp->snd_nxt, tp->snd_una))
tp->snd_nxt = tp->snd_una;
}
tp->t_timer[TCPT_REXMT] = 0;
tp->irs = ti->ti_seq;
tcp_rcvseqinit(tp);
tp->t_flags |= TF_ACKNOW;
if (tiflags & TH_ACK && SEQ_GT(tp->snd_una, tp->iss)) {
soisfconnected(so);
tp->t_state = TCPS_ESTABLISHED;
(void) tcp_reass(tp, (struct tcpiphdr *)0,
(struct mbuf *)0);
/*
* if we didn't have to retransmit the SYN,
* use its rtt as our initial srtt & rtt var.
*/
if (tp->t_rtt)
tcp_xmit_timer(tp, tp->t_rtt);
} else
tp->t_state = TCPS_SYN_RECEIVED;
trimthenstep6:
/*
* Advance ti->ti_seq to correspond to first data byte.
* If data, trim to stay within window,
* dropping FIN if necessary.
*/
ti->ti_seq++;
if (ti->ti_len > tp->rcv_wnd) {
todrop = ti->ti_len - tp->rcv_wnd;
m_adj(m, -todrop);
ti->ti_len = tp->rcv_wnd;
tiflags &= ~TH_FIN;
}
tp->snd_wl1 = ti->ti_seq - 1;
tp->rcv_up = ti->ti_seq;
goto step6;
} /* switch tp->t_state */
/*
* States other than LISTEN or SYN_SENT.
* Check that at least some bytes of segment are within
* receive window. If segment begins before rcv_nxt,
* drop leading data (and SYN); if nothing left, just ack.
*/
todrop = tp->rcv_nxt - ti->ti_seq;
if (todrop > 0) {
if (tiflags & TH_SYN) {
tiflags &= ~TH_SYN;
ti->ti_seq++;
if (ti->ti_urp > 1)
ti->ti_urp--;
else
tiflags &= ~TH_URG;
todrop--;
}
/*
* Following if statement from Stevens, vol. 2, p. 960.
*/
if (todrop > ti->ti_len
|| (todrop == ti->ti_len && (tiflags & TH_FIN) == 0)) {
/*
* Any valid FIN must be to the left of the window.
* At this point the FIN must be a duplicate or out
* of sequence; drop it.
*/
tiflags &= ~TH_FIN;
/*
* Send an ACK to resynchronize and drop any data.
* But keep on processing for RST or ACK.
*/
tp->t_flags |= TF_ACKNOW;
todrop = ti->ti_len;
}
m_adj(m, todrop);
ti->ti_seq += todrop;
ti->ti_len -= todrop;
if (ti->ti_urp > todrop)
ti->ti_urp -= todrop;
else {
tiflags &= ~TH_URG;
ti->ti_urp = 0;
}
}
/*
* If new data are received on a connection after the
* user processes are gone, then RST the other end.
*/
if ((so->so_state & SS_NOFDREF) &&
tp->t_state > TCPS_CLOSE_WAIT && ti->ti_len) {
tp = tcp_close(tp);
goto dropwithreset;
}
/*
* If segment ends after window, drop trailing data
* (and PUSH and FIN); if nothing left, just ACK.
*/
todrop = (ti->ti_seq+ti->ti_len) - (tp->rcv_nxt+tp->rcv_wnd);
if (todrop > 0) {
if (todrop >= ti->ti_len) {
/*
* If a new connection request is received
* while in TIME_WAIT, drop the old connection
* and start over if the sequence numbers
* are above the previous ones.
*/
if (tiflags & TH_SYN &&
tp->t_state == TCPS_TIME_WAIT &&
SEQ_GT(ti->ti_seq, tp->rcv_nxt)) {
iss = tp->rcv_nxt + TCP_ISSINCR;
tp = tcp_close(tp);
goto findso;
}
/*
* If window is closed can only take segments at
* window edge, and have to drop data and PUSH from
* incoming segments. Continue processing, but
* remember to ack. Otherwise, drop segment
* and ack.
*/
if (tp->rcv_wnd == 0 && ti->ti_seq == tp->rcv_nxt) {
tp->t_flags |= TF_ACKNOW;
} else {
goto dropafterack;
}
}
m_adj(m, -todrop);
ti->ti_len -= todrop;
tiflags &= ~(TH_PUSH|TH_FIN);
}
/*
* If the RST bit is set examine the state:
* SYN_RECEIVED STATE:
* If passive open, return to LISTEN state.
* If active open, inform user that connection was refused.
* ESTABLISHED, FIN_WAIT_1, FIN_WAIT2, CLOSE_WAIT STATES:
* Inform user that connection was reset, and close tcb.
* CLOSING, LAST_ACK, TIME_WAIT STATES
* Close the tcb.
*/
if (tiflags&TH_RST) switch (tp->t_state) {
case TCPS_SYN_RECEIVED:
case TCPS_ESTABLISHED:
case TCPS_FIN_WAIT_1:
case TCPS_FIN_WAIT_2:
case TCPS_CLOSE_WAIT:
tp->t_state = TCPS_CLOSED;
tcp_close(tp);
goto drop;
case TCPS_CLOSING:
case TCPS_LAST_ACK:
case TCPS_TIME_WAIT:
tcp_close(tp);
goto drop;
}
/*
* If a SYN is in the window, then this is an
* error and we send an RST and drop the connection.
*/
if (tiflags & TH_SYN) {
tp = tcp_drop(tp,0);
goto dropwithreset;
}
/*
* If the ACK bit is off we drop the segment and return.
*/
if ((tiflags & TH_ACK) == 0) goto drop;
/*
* Ack processing.
*/
switch (tp->t_state) {
/*
* In SYN_RECEIVED state if the ack ACKs our SYN then enter
* ESTABLISHED state and continue processing, otherwise
* send an RST. una<=ack<=max
*/
case TCPS_SYN_RECEIVED:
if (SEQ_GT(tp->snd_una, ti->ti_ack) ||
SEQ_GT(ti->ti_ack, tp->snd_max))
goto dropwithreset;
tp->t_state = TCPS_ESTABLISHED;
/*
* The sent SYN is ack'ed with our sequence number +1
* The first data byte already in the buffer will get
* lost if no correction is made. This is only needed for
* SS_CTL since the buffer is empty otherwise.
* tp->snd_una++; or:
*/
tp->snd_una=ti->ti_ack;
if (so->so_state & SS_CTL) {
/* So tcp_ctl reports the right state */
ret = tcp_ctl(so);
if (ret == 1) {
soisfconnected(so);
so->so_state &= ~SS_CTL; /* success XXX */
} else if (ret == 2) {
so->so_state &= SS_PERSISTENT_MASK;
so->so_state |= SS_NOFDREF; /* CTL_CMD */
} else {
needoutput = 1;
tp->t_state = TCPS_FIN_WAIT_1;
}
} else {
soisfconnected(so);
}
(void) tcp_reass(tp, (struct tcpiphdr *)0, (struct mbuf *)0);
tp->snd_wl1 = ti->ti_seq - 1;
/* Avoid ack processing; snd_una==ti_ack => dup ack */
goto synrx_to_est;
/* fall into ... */
/*
* In ESTABLISHED state: drop duplicate ACKs; ACK out of range
* ACKs. If the ack is in the range
* tp->snd_una < ti->ti_ack <= tp->snd_max
* then advance tp->snd_una to ti->ti_ack and drop
* data from the retransmission queue. If this ACK reflects
* more up to date window information we update our window information.
*/
case TCPS_ESTABLISHED:
case TCPS_FIN_WAIT_1:
case TCPS_FIN_WAIT_2:
case TCPS_CLOSE_WAIT:
case TCPS_CLOSING:
case TCPS_LAST_ACK:
case TCPS_TIME_WAIT:
if (SEQ_LEQ(ti->ti_ack, tp->snd_una)) {
if (ti->ti_len == 0 && tiwin == tp->snd_wnd) {
DEBUG_MISC((dfd, " dup ack m = %lx so = %lx\n",
(long )m, (long )so));
/*
* If we have outstanding data (other than
* a window probe), this is a completely
* duplicate ack (ie, window info didn't
* change), the ack is the biggest we've
* seen and we've seen exactly our rexmt
* threshold of them, assume a packet
* has been dropped and retransmit it.
* Kludge snd_nxt & the congestion
* window so we send only this one
* packet.
*
* We know we're losing at the current
* window size so do congestion avoidance
* (set ssthresh to half the current window
* and pull our congestion window back to
* the new ssthresh).
*
* Dup acks mean that packets have left the
* network (they're now cached at the receiver)
* so bump cwnd by the amount in the receiver
* to keep a constant cwnd packets in the
* network.
*/
if (tp->t_timer[TCPT_REXMT] == 0 ||
ti->ti_ack != tp->snd_una)
tp->t_dupacks = 0;
else if (++tp->t_dupacks == TCPREXMTTHRESH) {
tcp_seq onxt = tp->snd_nxt;
u_int win =
min(tp->snd_wnd, tp->snd_cwnd) / 2 /
tp->t_maxseg;
if (win < 2)
win = 2;
tp->snd_ssthresh = win * tp->t_maxseg;
tp->t_timer[TCPT_REXMT] = 0;
tp->t_rtt = 0;
tp->snd_nxt = ti->ti_ack;
tp->snd_cwnd = tp->t_maxseg;
(void) tcp_output(tp);
tp->snd_cwnd = tp->snd_ssthresh +
tp->t_maxseg * tp->t_dupacks;
if (SEQ_GT(onxt, tp->snd_nxt))
tp->snd_nxt = onxt;
goto drop;
} else if (tp->t_dupacks > TCPREXMTTHRESH) {
tp->snd_cwnd += tp->t_maxseg;
(void) tcp_output(tp);
goto drop;
}
} else
tp->t_dupacks = 0;
break;
}
synrx_to_est:
/*
* If the congestion window was inflated to account
* for the other side's cached packets, retract it.
*/
if (tp->t_dupacks > TCPREXMTTHRESH &&
tp->snd_cwnd > tp->snd_ssthresh)
tp->snd_cwnd = tp->snd_ssthresh;
tp->t_dupacks = 0;
if (SEQ_GT(ti->ti_ack, tp->snd_max)) {
goto dropafterack;
}
acked = ti->ti_ack - tp->snd_una;
/*
* If transmit timer is running and timed sequence
* number was acked, update smoothed round trip time.
* Since we now have an rtt measurement, cancel the
* timer backoff (cf., Phil Karn's retransmit alg.).
* Recompute the initial retransmit timer.
*/
if (tp->t_rtt && SEQ_GT(ti->ti_ack, tp->t_rtseq))
tcp_xmit_timer(tp,tp->t_rtt);
/*
* If all outstanding data is acked, stop retransmit
* timer and remember to restart (more output or persist).
* If there is more data to be acked, restart retransmit
* timer, using current (possibly backed-off) value.
*/
if (ti->ti_ack == tp->snd_max) {
tp->t_timer[TCPT_REXMT] = 0;
needoutput = 1;
} else if (tp->t_timer[TCPT_PERSIST] == 0)
tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
/*
* When new data is acked, open the congestion window.
* If the window gives us less than ssthresh packets
* in flight, open exponentially (maxseg per packet).
* Otherwise open linearly: maxseg per window
* (maxseg^2 / cwnd per packet).
*/
{
register u_int cw = tp->snd_cwnd;
register u_int incr = tp->t_maxseg;
if (cw > tp->snd_ssthresh)
incr = incr * incr / cw;
tp->snd_cwnd = min(cw + incr, TCP_MAXWIN<<tp->snd_scale);
}
if (acked > so->so_snd.sb_cc) {
tp->snd_wnd -= so->so_snd.sb_cc;
sbdrop(&so->so_snd, (int )so->so_snd.sb_cc);
ourfinisacked = 1;
} else {
sbdrop(&so->so_snd, acked);
tp->snd_wnd -= acked;
ourfinisacked = 0;
}
tp->snd_una = ti->ti_ack;
if (SEQ_LT(tp->snd_nxt, tp->snd_una))
tp->snd_nxt = tp->snd_una;
switch (tp->t_state) {
/*
* In FIN_WAIT_1 STATE in addition to the processing
* for the ESTABLISHED state if our FIN is now acknowledged
* then enter FIN_WAIT_2.
*/
case TCPS_FIN_WAIT_1:
if (ourfinisacked) {
/*
* If we can't receive any more
* data, then closing user can proceed.
* Starting the timer is contrary to the
* specification, but if we don't get a FIN
* we'll hang forever.
*/
if (so->so_state & SS_FCANTRCVMORE) {
tp->t_timer[TCPT_2MSL] = TCP_MAXIDLE;
}
tp->t_state = TCPS_FIN_WAIT_2;
}
break;
/*
* In CLOSING STATE in addition to the processing for
* the ESTABLISHED state if the ACK acknowledges our FIN
* then enter the TIME-WAIT state, otherwise ignore
* the segment.
*/
case TCPS_CLOSING:
if (ourfinisacked) {
tp->t_state = TCPS_TIME_WAIT;
tcp_canceltimers(tp);
tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
}
break;
/*
* In LAST_ACK, we may still be waiting for data to drain
* and/or to be acked, as well as for the ack of our FIN.
* If our FIN is now acknowledged, delete the TCB,
* enter the closed state and return.
*/
case TCPS_LAST_ACK:
if (ourfinisacked) {
tcp_close(tp);
goto drop;
}
break;
/*
* In TIME_WAIT state the only thing that should arrive
* is a retransmission of the remote FIN. Acknowledge
* it and restart the finack timer.
*/
case TCPS_TIME_WAIT:
tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
goto dropafterack;
}
} /* switch(tp->t_state) */
step6:
/*
* Update window information.
* Don't look at window if no ACK: TAC's send garbage on first SYN.
*/
if ((tiflags & TH_ACK) &&
(SEQ_LT(tp->snd_wl1, ti->ti_seq) ||
(tp->snd_wl1 == ti->ti_seq && (SEQ_LT(tp->snd_wl2, ti->ti_ack) ||
(tp->snd_wl2 == ti->ti_ack && tiwin > tp->snd_wnd))))) {
tp->snd_wnd = tiwin;
tp->snd_wl1 = ti->ti_seq;
tp->snd_wl2 = ti->ti_ack;
if (tp->snd_wnd > tp->max_sndwnd)
tp->max_sndwnd = tp->snd_wnd;
needoutput = 1;
}
/*
* Process segments with URG.
*/
if ((tiflags & TH_URG) && ti->ti_urp &&
TCPS_HAVERCVDFIN(tp->t_state) == 0) {
/*
* This is a kludge, but if we receive and accept
* random urgent pointers, we'll crash in
* soreceive. It's hard to imagine someone
* actually wanting to send this much urgent data.
*/
if (ti->ti_urp + so->so_rcv.sb_cc > so->so_rcv.sb_datalen) {
ti->ti_urp = 0;
tiflags &= ~TH_URG;
goto dodata;
}
/*
* If this segment advances the known urgent pointer,
* then mark the data stream. This should not happen
* in CLOSE_WAIT, CLOSING, LAST_ACK or TIME_WAIT STATES since
* a FIN has been received from the remote side.
* In these states we ignore the URG.
*
* According to RFC961 (Assigned Protocols),
* the urgent pointer points to the last octet
* of urgent data. We continue, however,
* to consider it to indicate the first octet
* of data past the urgent section as the original
* spec states (in one of two places).
*/
if (SEQ_GT(ti->ti_seq+ti->ti_urp, tp->rcv_up)) {
tp->rcv_up = ti->ti_seq + ti->ti_urp;
so->so_urgc = so->so_rcv.sb_cc +
(tp->rcv_up - tp->rcv_nxt); /* -1; */
tp->rcv_up = ti->ti_seq + ti->ti_urp;
}
} else
/*
* If no out of band data is expected,
* pull receive urgent pointer along
* with the receive window.
*/
if (SEQ_GT(tp->rcv_nxt, tp->rcv_up))
tp->rcv_up = tp->rcv_nxt;
dodata:
/*
* If this is a small packet, then ACK now - with Nagel
* congestion avoidance sender won't send more until
* he gets an ACK.
*/
if (ti->ti_len && (unsigned)ti->ti_len <= 5 &&
((struct tcpiphdr_2 *)ti)->first_char == (char)27) {
tp->t_flags |= TF_ACKNOW;
}
/*
* Process the segment text, merging it into the TCP sequencing queue,
* and arranging for acknowledgment of receipt if necessary.
* This process logically involves adjusting tp->rcv_wnd as data
* is presented to the user (this happens in tcp_usrreq.c,
* case PRU_RCVD). If a FIN has already been received on this
* connection then we just ignore the text.
*/
if ((ti->ti_len || (tiflags&TH_FIN)) &&
TCPS_HAVERCVDFIN(tp->t_state) == 0) {
TCP_REASS(tp, ti, m, so, tiflags);
} else {
m_free(m);
tiflags &= ~TH_FIN;
}
/*
* If FIN is received ACK the FIN and let the user know
* that the connection is closing.
*/
if (tiflags & TH_FIN) {
if (TCPS_HAVERCVDFIN(tp->t_state) == 0) {
/*
* If we receive a FIN we can't send more data,
* set it SS_FDRAIN
* Shutdown the socket if there is no rx data in the
* buffer.
* soread() is called on completion of shutdown() and
* will got to TCPS_LAST_ACK, and use tcp_output()
* to send the FIN.
*/
sofwdrain(so);
tp->t_flags |= TF_ACKNOW;
tp->rcv_nxt++;
}
switch (tp->t_state) {
/*
* In SYN_RECEIVED and ESTABLISHED STATES
* enter the CLOSE_WAIT state.
*/
case TCPS_SYN_RECEIVED:
case TCPS_ESTABLISHED:
if(so->so_emu == EMU_CTL) /* no shutdown on socket */
tp->t_state = TCPS_LAST_ACK;
else
tp->t_state = TCPS_CLOSE_WAIT;
break;
/*
* If still in FIN_WAIT_1 STATE FIN has not been acked so
* enter the CLOSING state.
*/
case TCPS_FIN_WAIT_1:
tp->t_state = TCPS_CLOSING;
break;
/*
* In FIN_WAIT_2 state enter the TIME_WAIT state,
* starting the time-wait timer, turning off the other
* standard timers.
*/
case TCPS_FIN_WAIT_2:
tp->t_state = TCPS_TIME_WAIT;
tcp_canceltimers(tp);
tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
break;
/*
* In TIME_WAIT state restart the 2 MSL time_wait timer.
*/
case TCPS_TIME_WAIT:
tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
break;
}
}
/*
* Return any desired output.
*/
if (needoutput || (tp->t_flags & TF_ACKNOW)) {
(void) tcp_output(tp);
}
return;
dropafterack:
/*
* Generate an ACK dropping incoming segment if it occupies
* sequence space, where the ACK reflects our state.
*/
if (tiflags & TH_RST)
goto drop;
m_free(m);
tp->t_flags |= TF_ACKNOW;
(void) tcp_output(tp);
return;
dropwithreset:
/* reuses m if m!=NULL, m_free() unnecessary */
if (tiflags & TH_ACK)
tcp_respond(tp, ti, m, (tcp_seq)0, ti->ti_ack, TH_RST);
else {
if (tiflags & TH_SYN) ti->ti_len++;
tcp_respond(tp, ti, m, ti->ti_seq+ti->ti_len, (tcp_seq)0,
TH_RST|TH_ACK);
}
return;
drop:
/*
* Drop space held by incoming segment and return.
*/
m_free(m);
}
void
udp_input(struct mbuf *m, int off)
{
int iphlen = off;
struct ip *ip;
struct udphdr *uh;
struct ifnet *ifp;
struct inpcb *inp;
uint16_t len, ip_len;
struct inpcbinfo *pcbinfo;
struct ip save_ip;
struct sockaddr_in udp_in;
struct m_tag *fwd_tag;
int cscov_partial;
uint8_t pr;
ifp = m->m_pkthdr.rcvif;
UDPSTAT_INC(udps_ipackets);
/*
* Strip IP options, if any; should skip this, make available to
* user, and use on returned packets, but we don't yet have a way to
* check the checksum with options still present.
*/
if (iphlen > sizeof (struct ip)) {
ip_stripoptions(m);
iphlen = sizeof(struct ip);
}
/*
* Get IP and UDP header together in first mbuf.
*/
ip = mtod(m, struct ip *);
if (m->m_len < iphlen + sizeof(struct udphdr)) {
if ((m = m_pullup(m, iphlen + sizeof(struct udphdr))) == NULL) {
UDPSTAT_INC(udps_hdrops);
return;
}
ip = mtod(m, struct ip *);
}
uh = (struct udphdr *)((caddr_t)ip + iphlen);
pr = ip->ip_p;
cscov_partial = (pr == IPPROTO_UDPLITE) ? 1 : 0;
/*
* Destination port of 0 is illegal, based on RFC768.
*/
if (uh->uh_dport == 0)
goto badunlocked;
/*
* Construct sockaddr format source address. Stuff source address
* and datagram in user buffer.
*/
bzero(&udp_in, sizeof(udp_in));
udp_in.sin_len = sizeof(udp_in);
udp_in.sin_family = AF_INET;
udp_in.sin_port = uh->uh_sport;
udp_in.sin_addr = ip->ip_src;
/*
* Make mbuf data length reflect UDP length. If not enough data to
* reflect UDP length, drop.
*/
len = ntohs((u_short)uh->uh_ulen);
ip_len = ntohs(ip->ip_len) - iphlen;
if (pr == IPPROTO_UDPLITE && len == 0) {
/* Zero means checksum over the complete packet. */
len = ip_len;
cscov_partial = 0;
}
if (ip_len != len) {
if (len > ip_len || len < sizeof(struct udphdr)) {
UDPSTAT_INC(udps_badlen);
goto badunlocked;
}
if (pr == IPPROTO_UDP)
m_adj(m, len - ip_len);
}
/*
* Save a copy of the IP header in case we want restore it for
* sending an ICMP error message in response.
*/
if (!V_udp_blackhole)
save_ip = *ip;
else
memset(&save_ip, 0, sizeof(save_ip));
/*
* Checksum extended UDP header and data.
*/
if (uh->uh_sum) {
u_short uh_sum;
if ((m->m_pkthdr.csum_flags & CSUM_DATA_VALID) &&
!cscov_partial) {
if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR)
uh_sum = m->m_pkthdr.csum_data;
else
uh_sum = in_pseudo(ip->ip_src.s_addr,
ip->ip_dst.s_addr, htonl((u_short)len +
m->m_pkthdr.csum_data + pr));
uh_sum ^= 0xffff;
} else {
char b[9];
bcopy(((struct ipovly *)ip)->ih_x1, b, 9);
bzero(((struct ipovly *)ip)->ih_x1, 9);
((struct ipovly *)ip)->ih_len = (pr == IPPROTO_UDP) ?
uh->uh_ulen : htons(ip_len);
uh_sum = in_cksum(m, len + sizeof (struct ip));
bcopy(b, ((struct ipovly *)ip)->ih_x1, 9);
}
if (uh_sum) {
UDPSTAT_INC(udps_badsum);
m_freem(m);
return;
}
} else
UDPSTAT_INC(udps_nosum);
pcbinfo = get_inpcbinfo(pr);
if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
in_broadcast(ip->ip_dst, ifp)) {
struct inpcb *last;
struct inpcbhead *pcblist;
struct ip_moptions *imo;
INP_INFO_RLOCK(pcbinfo);
pcblist = get_pcblist(pr);
last = NULL;
LIST_FOREACH(inp, pcblist, inp_list) {
if (inp->inp_lport != uh->uh_dport)
continue;
#ifdef INET6
if ((inp->inp_vflag & INP_IPV4) == 0)
continue;
#endif
if (inp->inp_laddr.s_addr != INADDR_ANY &&
inp->inp_laddr.s_addr != ip->ip_dst.s_addr)
continue;
if (inp->inp_faddr.s_addr != INADDR_ANY &&
inp->inp_faddr.s_addr != ip->ip_src.s_addr)
continue;
if (inp->inp_fport != 0 &&
inp->inp_fport != uh->uh_sport)
continue;
INP_RLOCK(inp);
/*
* XXXRW: Because we weren't holding either the inpcb
* or the hash lock when we checked for a match
* before, we should probably recheck now that the
* inpcb lock is held.
*/
/*
* Handle socket delivery policy for any-source
* and source-specific multicast. [RFC3678]
*/
imo = inp->inp_moptions;
if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr))) {
struct sockaddr_in group;
int blocked;
if (imo == NULL) {
INP_RUNLOCK(inp);
continue;
}
bzero(&group, sizeof(struct sockaddr_in));
group.sin_len = sizeof(struct sockaddr_in);
group.sin_family = AF_INET;
group.sin_addr = ip->ip_dst;
blocked = imo_multi_filter(imo, ifp,
(struct sockaddr *)&group,
(struct sockaddr *)&udp_in);
if (blocked != MCAST_PASS) {
if (blocked == MCAST_NOTGMEMBER)
IPSTAT_INC(ips_notmember);
if (blocked == MCAST_NOTSMEMBER ||
blocked == MCAST_MUTED)
UDPSTAT_INC(udps_filtermcast);
INP_RUNLOCK(inp);
continue;
}
}
if (last != NULL) {
struct mbuf *n;
n = m_copy(m, 0, M_COPYALL);
udp_append(last, ip, n, iphlen, &udp_in);
INP_RUNLOCK(last);
}
last = inp;
/*
* Don't look for additional matches if this one does
* not have either the SO_REUSEPORT or SO_REUSEADDR
* socket options set. This heuristic avoids
* searching through all pcbs in the common case of a
* non-shared port. It assumes that an application
* will never clear these options after setting them.
*/
if ((last->inp_socket->so_options &
(SO_REUSEPORT|SO_REUSEADDR)) == 0)
break;
}
if (last == NULL) {
/*
* No matching pcb found; discard datagram. (No need
* to send an ICMP Port Unreachable for a broadcast
* or multicast datgram.)
*/
UDPSTAT_INC(udps_noportbcast);
if (inp)
INP_RUNLOCK(inp);
INP_INFO_RUNLOCK(pcbinfo);
goto badunlocked;
}
udp_append(last, ip, m, iphlen, &udp_in);
INP_RUNLOCK(last);
INP_INFO_RUNLOCK(pcbinfo);
return;
}
/*
* Locate pcb for datagram.
*/
/*
* Grab info from PACKET_TAG_IPFORWARD tag prepended to the chain.
*/
if ((m->m_flags & M_IP_NEXTHOP) &&
(fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL)) != NULL) {
struct sockaddr_in *next_hop;
next_hop = (struct sockaddr_in *)(fwd_tag + 1);
/*
* Transparently forwarded. Pretend to be the destination.
* Already got one like this?
*/
inp = in_pcblookup_mbuf(pcbinfo, ip->ip_src, uh->uh_sport,
ip->ip_dst, uh->uh_dport, INPLOOKUP_RLOCKPCB, ifp, m);
if (!inp) {
/*
* It's new. Try to find the ambushing socket.
* Because we've rewritten the destination address,
* any hardware-generated hash is ignored.
*/
inp = in_pcblookup(pcbinfo, ip->ip_src,
uh->uh_sport, next_hop->sin_addr,
next_hop->sin_port ? htons(next_hop->sin_port) :
uh->uh_dport, INPLOOKUP_WILDCARD |
INPLOOKUP_RLOCKPCB, ifp);
}
/* Remove the tag from the packet. We don't need it anymore. */
m_tag_delete(m, fwd_tag);
m->m_flags &= ~M_IP_NEXTHOP;
} else
inp = in_pcblookup_mbuf(pcbinfo, ip->ip_src, uh->uh_sport,
ip->ip_dst, uh->uh_dport, INPLOOKUP_WILDCARD |
INPLOOKUP_RLOCKPCB, ifp, m);
if (inp == NULL) {
if (udp_log_in_vain) {
char buf[4*sizeof "123"];
strcpy(buf, inet_ntoa(ip->ip_dst));
log(LOG_INFO,
"Connection attempt to UDP %s:%d from %s:%d\n",
buf, ntohs(uh->uh_dport), inet_ntoa(ip->ip_src),
ntohs(uh->uh_sport));
}
UDPSTAT_INC(udps_noport);
if (m->m_flags & (M_BCAST | M_MCAST)) {
UDPSTAT_INC(udps_noportbcast);
goto badunlocked;
}
if (V_udp_blackhole)
goto badunlocked;
if (badport_bandlim(BANDLIM_ICMP_UNREACH) < 0)
goto badunlocked;
*ip = save_ip;
icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PORT, 0, 0);
return;
}
/*
* Check the minimum TTL for socket.
*/
INP_RLOCK_ASSERT(inp);
if (inp->inp_ip_minttl && inp->inp_ip_minttl > ip->ip_ttl) {
INP_RUNLOCK(inp);
m_freem(m);
return;
}
if (cscov_partial) {
struct udpcb *up;
up = intoudpcb(inp);
if (up->u_rxcslen > len) {
INP_RUNLOCK(inp);
m_freem(m);
return;
}
}
UDP_PROBE(receive, NULL, inp, ip, inp, uh);
udp_append(inp, ip, m, iphlen, &udp_in);
INP_RUNLOCK(inp);
return;
badunlocked:
m_freem(m);
}
void
udp_input(register struct mbuf *m, int iphlen)
{
register struct ip *ip;
register struct udphdr *uh;
/* struct mbuf *opts = 0;*/
int len;
struct ip save_ip;
struct socket *so;
DEBUG_CALL("udp_input");
DEBUG_ARG("m = %lx", (long)m);
DEBUG_ARG("iphlen = %d", iphlen);
STAT(udpstat.udps_ipackets++);
/*
* Strip IP options, if any; should skip this,
* make available to user, and use on returned packets,
* but we don't yet have a way to check the checksum
* with options still present.
*/
if(iphlen > sizeof(struct ip)) {
ip_stripoptions(m, (struct mbuf *)0);
iphlen = sizeof(struct ip);
}
/*
* Get IP and UDP header together in first mbuf.
*/
ip = mtod(m, struct ip *);
uh = (struct udphdr *)((caddr_t)ip + iphlen);
/*
* Make mbuf data length reflect UDP length.
* If not enough data to reflect UDP length, drop.
*/
len = ntohs((u_int16_t)uh->uh_ulen);
if (ip->ip_len != len) {
if (len > ip->ip_len) {
STAT(udpstat.udps_badlen++);
goto bad;
}
m_adj(m, len - ip->ip_len);
ip->ip_len = len;
}
/*
* Save a copy of the IP header in case we want restore it
* for sending an ICMP error message in response.
*/
save_ip = *ip;
save_ip.ip_len+= iphlen; /* tcp_input subtracts this */
/*
* Checksum extended UDP header and data.
*/
if (UDPCKSUM && uh->uh_sum) {
memset(&((struct ipovly *)ip)->ih_mbuf, 0, sizeof(struct mbuf_ptr));
((struct ipovly *)ip)->ih_x1 = 0;
((struct ipovly *)ip)->ih_len = uh->uh_ulen;
/* keep uh_sum for ICMP reply
* uh->uh_sum = cksum(m, len + sizeof (struct ip));
* if (uh->uh_sum) {
*/
if(cksum(m, len + sizeof(struct ip))) {
STAT(udpstat.udps_badsum++);
goto bad;
}
}
/*
* handle DHCP/BOOTP
*/
if (ntohs(uh->uh_dport) == BOOTP_SERVER) {
bootp_input(m);
goto bad;
}
if (slirp_restrict)
goto bad;
/*
* handle TFTP
*/
if (ntohs(uh->uh_dport) == TFTP_SERVER) {
tftp_input(m);
goto bad;
}
/*
* Locate pcb for datagram.
*/
so = udp_last_so;
if (so->so_lport != uh->uh_sport ||
so->so_laddr.s_addr != ip->ip_src.s_addr) {
struct socket *tmp;
for (tmp = udb.so_next; tmp != &udb; tmp = tmp->so_next) {
if (tmp->so_lport == uh->uh_sport &&
tmp->so_laddr.s_addr == ip->ip_src.s_addr) {
tmp->so_faddr.s_addr = ip->ip_dst.s_addr;
tmp->so_fport = uh->uh_dport;
so = tmp;
break;
}
}
if (tmp == &udb) {
so = NULL;
} else {
STAT(udpstat.udpps_pcbcachemiss++);
udp_last_so = so;
}
}
if (so == NULL) {
/*
* If there's no socket for this packet,
* create one
*/
if ((so = socreate()) == NULL) goto bad;
if(udp_attach(so) == -1) {
DEBUG_MISC((dfd," udp_attach errno = %d-%s\n",
errno,strerror(errno)));
sofree(so);
goto bad;
}
/*
* Setup fields
*/
/* udp_last_so = so; */
so->so_laddr = ip->ip_src;
so->so_lport = uh->uh_sport;
if ((so->so_iptos = udp_tos(so)) == 0)
so->so_iptos = ip->ip_tos;
/*
* XXXXX Here, check if it's in udpexec_list,
* and if it is, do the fork_exec() etc.
*/
}
so->so_faddr = ip->ip_dst; /* XXX */
so->so_fport = uh->uh_dport; /* XXX */
iphlen += sizeof(struct udphdr);
m->m_len -= iphlen;
m->m_data += iphlen;
/*
* Now we sendto() the packet.
*/
if (so->so_emu)
udp_emu(so, m);
if(sosendto(so,m) == -1) {
m->m_len += iphlen;
m->m_data -= iphlen;
*ip=save_ip;
DEBUG_MISC((dfd,"udp tx errno = %d-%s\n",errno,strerror(errno)));
icmp_error(m, ICMP_UNREACH,ICMP_UNREACH_NET, 0,strerror(errno));
}
m_free(so->so_m); /* used for ICMP if error on sorecvfrom */
/* restore the orig mbuf packet */
m->m_len += iphlen;
m->m_data -= iphlen;
*ip=save_ip;
so->so_m=m; /* ICMP backup */
return;
bad:
m_freem(m);
/* if (opts) m_freem(opts); */
return;
}
/*
* [udp_validate]
*
* Validates UDP length and checksum
* Almost verbatim from udp_input
* returns: 1 -> good, 0 -> bad
*/
static int
udp_validate(register struct mbuf *m)
{
register struct ip *ip = mtod(m, struct ip *);
register struct udphdr *uh;
int len;
// int iphlen = IP_VHL_HL(ip->ip_vhl) << 2;
int iphlen = ip->ip_hl << 2;
/*
* Can't check cksum with options present
*/
if (iphlen > sizeof (struct ip)) {
printf("udp_cksum_validate: ip options present, skipping checks\n");
#ifdef NOTYET
ip_stripoptions(m, (struct mbuf *)0);
iphlen = sizeof(struct ip);
#endif
return(1);
}
/*
* Get IP and UDP header together in first mbuf.
*/
if (m->m_len < iphlen + sizeof(struct udphdr)) {
if ((m = m_pullup(m, iphlen + sizeof(struct udphdr))) == 0) {
return(0);
}
ip = mtod(m, struct ip *);
}
uh = (struct udphdr *)((caddr_t)ip + iphlen);
/*
* Make mbuf data length reflect UDP length.
* If not enough data to reflect UDP length, drop.
*/
len = ntohs((u_short)uh->uh_ulen);
if (ip->ip_len != len) {
if (len > ip->ip_len || len < sizeof(struct udphdr)) {
printf("udp_validate: bad udp length\n");
m_freem(m);
return(0);
}
}
/*
* Checksum extended UDP header and data.
*/
if (uh->uh_sum) {
if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) {
if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR)
uh->uh_sum = m->m_pkthdr.csum_data;
else
uh->uh_sum = in_pseudo(ip->ip_src.s_addr,
ip->ip_dst.s_addr, htonl((u_short)len +
m->m_pkthdr.csum_data + IPPROTO_UDP));
uh->uh_sum ^= 0xffff;
} else {
char b[9];
bcopy(((struct ipovly *)ip)->ih_x1, b, 9);
bzero(((struct ipovly *)ip)->ih_x1, 9);
((struct ipovly *)ip)->ih_len = uh->uh_ulen;
uh->uh_sum = in_cksum(m, len + sizeof (struct ip));
bcopy(b, ((struct ipovly *)ip)->ih_x1, 9);
}
if (uh->uh_sum) {
printf("udp_validate: failed, bad checksum\n");
m_freem(m);
return(0);
}
}
return(1);
}
/* m->m_data points at ip packet header
* m->m_len length ip packet
* ip->ip_len length data (IPDU)
*/
void
udp_input(PNATState pData, register struct mbuf *m, int iphlen)
{
register struct ip *ip;
register struct udphdr *uh;
int len;
struct ip save_ip;
struct socket *so;
int ret;
int ttl, tos;
LogFlowFunc(("ENTER: m = %p, iphlen = %d\n", m, iphlen));
ip = mtod(m, struct ip *);
Log2(("%RTnaipv4 iphlen = %d\n", ip->ip_dst, iphlen));
udpstat.udps_ipackets++;
/*
* Strip IP options, if any; should skip this,
* make available to user, and use on returned packets,
* but we don't yet have a way to check the checksum
* with options still present.
*/
if (iphlen > sizeof(struct ip))
{
ip_stripoptions(m, (struct mbuf *)0);
iphlen = sizeof(struct ip);
}
/*
* Get IP and UDP header together in first mbuf.
*/
ip = mtod(m, struct ip *);
uh = (struct udphdr *)((caddr_t)ip + iphlen);
/*
* Make mbuf data length reflect UDP length.
* If not enough data to reflect UDP length, drop.
*/
len = RT_N2H_U16((u_int16_t)uh->uh_ulen);
Assert(ip->ip_len + iphlen == (ssize_t)m_length(m, NULL));
if (ip->ip_len != len)
{
if (len > ip->ip_len)
{
udpstat.udps_badlen++;
Log3(("NAT: IP(id: %hd) has bad size\n", ip->ip_id));
goto bad_free_mbuf;
}
m_adj(m, len - ip->ip_len);
ip->ip_len = len;
}
/*
* Save a copy of the IP header in case we want restore it
* for sending an ICMP error message in response.
*/
save_ip = *ip;
save_ip.ip_len+= iphlen; /* tcp_input subtracts this */
/*
* Checksum extended UDP header and data.
*/
if (udpcksum && uh->uh_sum)
{
memset(((struct ipovly *)ip)->ih_x1, 0, 9);
((struct ipovly *)ip)->ih_len = uh->uh_ulen;
#if 0
/* keep uh_sum for ICMP reply */
uh->uh_sum = cksum(m, len + sizeof (struct ip));
if (uh->uh_sum)
{
#endif
if (cksum(m, len + iphlen))
{
udpstat.udps_badsum++;
Log3(("NAT: IP(id: %hd) has bad (udp) cksum\n", ip->ip_id));
goto bad_free_mbuf;
}
}
#if 0
}
#endif
/*
* handle DHCP/BOOTP
*/
if (uh->uh_dport == RT_H2N_U16_C(BOOTP_SERVER))
{
bootp_input(pData, m);
goto done_free_mbuf;
}
LogFunc(("uh src: %RTnaipv4:%d, dst: %RTnaipv4:%d\n",
ip->ip_src.s_addr, RT_N2H_U16(uh->uh_sport),
ip->ip_dst.s_addr, RT_N2H_U16(uh->uh_dport)));
/*
* handle DNS host resolver without creating a socket
*/
if ( pData->fUseHostResolver
&& uh->uh_dport == RT_H2N_U16_C(53)
&& CTL_CHECK(ip->ip_dst.s_addr, CTL_DNS))
{
struct sockaddr_in dst, src;
src.sin_addr.s_addr = ip->ip_dst.s_addr;
src.sin_port = uh->uh_dport;
dst.sin_addr.s_addr = ip->ip_src.s_addr;
dst.sin_port = uh->uh_sport;
m_adj(m, sizeof(struct udpiphdr));
m = hostresolver(pData, m, ip->ip_src.s_addr, uh->uh_sport);
if (m == NULL)
goto done_free_mbuf;
slirpMbufTagService(pData, m, CTL_DNS);
udp_output2(pData, NULL, m, &src, &dst, IPTOS_LOWDELAY);
LogFlowFuncLeave();
return;
}
/*
* handle TFTP
*/
if ( uh->uh_dport == RT_H2N_U16_C(TFTP_SERVER)
&& CTL_CHECK(ip->ip_dst.s_addr, CTL_TFTP))
{
if (pData->pvTftpSessions)
slirpTftpInput(pData, m);
goto done_free_mbuf;
}
/*
* XXX: DNS proxy currently relies on the fact that each socket
* only serves one request.
*/
if ( pData->fUseDnsProxy
&& CTL_CHECK(ip->ip_dst.s_addr, CTL_DNS)
&& (uh->uh_dport == RT_H2N_U16_C(53)))
{
so = NULL;
goto new_socket;
}
/*
* Locate pcb for datagram.
*/
so = udp_last_so;
if ( so->so_lport != uh->uh_sport
|| so->so_laddr.s_addr != ip->ip_src.s_addr)
{
struct socket *tmp;
for (tmp = udb.so_next; tmp != &udb; tmp = tmp->so_next)
{
if ( tmp->so_lport == uh->uh_sport
&& tmp->so_laddr.s_addr == ip->ip_src.s_addr)
{
so = tmp;
break;
}
}
if (tmp == &udb)
so = NULL;
else
{
udpstat.udpps_pcbcachemiss++;
udp_last_so = so;
}
}
new_socket:
if (so == NULL)
{
/*
* If there's no socket for this packet,
* create one
*/
if ((so = socreate()) == NULL)
{
Log2(("NAT: IP(id: %hd) failed to create socket\n", ip->ip_id));
goto bad_free_mbuf;
}
/*
* Setup fields
*/
so->so_laddr = ip->ip_src;
so->so_lport = uh->uh_sport;
so->so_iptos = ip->ip_tos;
if (udp_attach(pData, so) <= 0)
{
Log2(("NAT: IP(id: %hd) udp_attach errno = %d (%s)\n",
ip->ip_id, errno, strerror(errno)));
sofree(pData, so);
goto bad_free_mbuf;
}
/* udp_last_so = so; */
/*
* XXXXX Here, check if it's in udpexec_list,
* and if it is, do the fork_exec() etc.
*/
}
so->so_faddr = ip->ip_dst; /* XXX */
so->so_fport = uh->uh_dport; /* XXX */
Assert(so->so_type == IPPROTO_UDP);
/*
* DNS proxy
*/
if ( pData->fUseDnsProxy
&& CTL_CHECK(ip->ip_dst.s_addr, CTL_DNS)
&& (uh->uh_dport == RT_H2N_U16_C(53)))
{
dnsproxy_query(pData, so, m, iphlen);
goto done_free_mbuf;
}
iphlen += sizeof(struct udphdr);
m->m_len -= iphlen;
m->m_data += iphlen;
ttl = ip->ip_ttl = save_ip.ip_ttl;
if (ttl != so->so_sottl) {
ret = setsockopt(so->s, IPPROTO_IP, IP_TTL,
(char *)&ttl, sizeof(ttl));
if (RT_LIKELY(ret == 0))
so->so_sottl = ttl;
}
tos = save_ip.ip_tos;
if (tos != so->so_sotos) {
ret = setsockopt(so->s, IPPROTO_IP, IP_TOS,
(char *)&tos, sizeof(tos));
if (RT_LIKELY(ret == 0))
so->so_sotos = tos;
}
{
/*
* Different OSes have different socket options for DF. We
* can't use IP_HDRINCL here as it's only valid for SOCK_RAW.
*/
# define USE_DF_OPTION(_Optname) \
const int dfopt = _Optname
#if defined(IP_MTU_DISCOVER)
USE_DF_OPTION(IP_MTU_DISCOVER);
#elif defined(IP_DONTFRAG) /* Solaris 11+, FreeBSD */
USE_DF_OPTION(IP_DONTFRAG);
#elif defined(IP_DONTFRAGMENT) /* Windows */
USE_DF_OPTION(IP_DONTFRAGMENT);
#else
USE_DF_OPTION(0);
#endif
if (dfopt) {
int df = (save_ip.ip_off & IP_DF) != 0;
#if defined(IP_MTU_DISCOVER)
df = df ? IP_PMTUDISC_DO : IP_PMTUDISC_DONT;
#endif
if (df != so->so_sodf) {
ret = setsockopt(so->s, IPPROTO_IP, dfopt,
(char *)&df, sizeof(df));
if (RT_LIKELY(ret == 0))
so->so_sodf = df;
}
}
}
if ( sosendto(pData, so, m) == -1
&& ( !soIgnorableErrorCode(errno)
&& errno != ENOTCONN))
{
m->m_len += iphlen;
m->m_data -= iphlen;
*ip = save_ip;
Log2(("NAT: UDP tx errno = %d (%s) on sent to %RTnaipv4\n",
errno, strerror(errno), ip->ip_dst));
icmp_error(pData, m, ICMP_UNREACH, ICMP_UNREACH_NET, 0, strerror(errno));
so->so_m = NULL;
LogFlowFuncLeave();
return;
}
if (so->so_m)
m_freem(pData, so->so_m); /* used for ICMP if error on sorecvfrom */
/* restore the orig mbuf packet */
m->m_len += iphlen;
m->m_data -= iphlen;
*ip = save_ip;
so->so_m = m; /* ICMP backup */
LogFlowFuncLeave();
return;
bad_free_mbuf:
Log2(("NAT: UDP(id: %hd) datagram to %RTnaipv4 with size(%d) claimed as bad\n",
ip->ip_id, &ip->ip_dst, ip->ip_len));
done_free_mbuf:
/* some services like bootp(built-in), dns(buildt-in) and dhcp don't need sockets
* and create new m'buffers to send them to guest, so we'll free their incomming
* buffers here.
*/
if (m != NULL)
m_freem(pData, m);
LogFlowFuncLeave();
return;
}
/*
* TCP input routine, follows pages 65-76 of the
* protocol specification dated September, 1981 very closely.
*/
void
tcp_input(usn_mbuf_t *m, int iphlen)
{
struct tcpiphdr *ti;
struct inpcb *inp;
u_char *optp = NULL;
int optlen;
int len, tlen, off;
struct tcpcb *tp = 0;
int tiflags;
struct usn_socket *so = 0;
int todrop, acked, ourfinisacked;
int needoutput = 0;
short ostate;
struct usn_in_addr laddr;
int dropsocket = 0;
int iss = 0;
u_long tiwin, ts_val, ts_ecr;
int ts_present = 0;
(void)needoutput;
g_tcpstat.tcps_rcvtotal++;
// Get IP and TCP header together in first mbuf.
// Note: IP leaves IP header in first mbuf.
ti = mtod(m, struct tcpiphdr *);
if (iphlen > sizeof (usn_ip_t))
ip_stripoptions(m, (usn_mbuf_t *)0);
if (m->mlen < sizeof (struct tcpiphdr)) {
if ((m = m_pullup(m, sizeof (struct tcpiphdr))) == 0) {
g_tcpstat.tcps_rcvshort++;
return;
}
ti = mtod(m, struct tcpiphdr *);
}
#ifdef DUMP_PAYLOAD
dump_chain(m,"tcp");
#endif
/*
* Checksum extended TCP header and data.
*/
tlen = ntohs(((usn_ip_t *)ti)->ip_len);
len = sizeof (usn_ip_t) + tlen;
ti->ti_next = ti->ti_prev = 0;
ti->ti_x1 = 0;
ti->ti_len = (u_short)tlen;
HTONS(ti->ti_len);
ti->ti_sum = in_cksum(m, len);
if (ti->ti_sum) {
g_tcpstat.tcps_rcvbadsum++;
goto drop;
}
/*
* Check that TCP offset makes sense,
* pull out TCP options and adjust length. XXX
*/
off = ti->ti_off << 2;
if (off < sizeof (struct tcphdr) || off > tlen) {
g_tcpstat.tcps_rcvbadoff++;
goto drop;
}
tlen -= off;
ti->ti_len = tlen;
if (off > sizeof (struct tcphdr)) {
if (m->mlen < sizeof(usn_ip_t) + off) {
if ((m = m_pullup(m, sizeof (usn_ip_t) + off)) == 0) {
g_tcpstat.tcps_rcvshort++;
return;
}
ti = mtod(m, struct tcpiphdr *);
}
optlen = off - sizeof (struct tcphdr);
optp = mtod(m, u_char *) + sizeof (struct tcpiphdr);
// Do quick retrieval of timestamp options ("options
// prediction?"). If timestamp is the only option and it's
// formatted as recommended in RFC 1323 appendix A, we
// quickly get the values now and not bother calling
// tcp_dooptions(), etc.
if ((optlen == TCPOLEN_TSTAMP_APPA ||
(optlen > TCPOLEN_TSTAMP_APPA &&
optp[TCPOLEN_TSTAMP_APPA] == TCPOPT_EOL)) &&
*(u_int *)optp == htonl(TCPOPT_TSTAMP_HDR) &&
(ti->ti_flags & TH_SYN) == 0) {
ts_present = 1;
ts_val = ntohl(*(u_long *)(optp + 4));
ts_ecr = ntohl(*(u_long *)(optp + 8));
optp = NULL; // we've parsed the options
}
}
tiflags = ti->ti_flags;
// Convert TCP protocol specific fields to host format.
NTOHL(ti->ti_seq);
NTOHL(ti->ti_ack);
NTOHS(ti->ti_win);
NTOHS(ti->ti_urp);
// Locate pcb for segment.
findpcb:
inp = g_tcp_last_inpcb;
if (inp->inp_lport != ti->ti_dport ||
inp->inp_fport != ti->ti_sport ||
inp->inp_faddr.s_addr != ti->ti_src.s_addr ||
inp->inp_laddr.s_addr != ti->ti_dst.s_addr) {
inp = in_pcblookup(&g_tcb, ti->ti_src, ti->ti_sport,
ti->ti_dst, ti->ti_dport, INPLOOKUP_WILDCARD);
if (inp)
g_tcp_last_inpcb = inp;
++g_tcpstat.tcps_pcbcachemiss;
}
// If the state is CLOSED (i.e., TCB does not exist) then
// all data in the incoming segment is discarded.
// If the TCB exists but is in CLOSED state, it is embryonic,
// but should either do a listen or a connect soon.
if (inp == 0)
goto dropwithreset;
tp = intotcpcb(inp);
DEBUG("found inp cb, laddr=%x, lport=%d, faddr=%x,"
" fport=%d, tp_state=%d, tp_flags=%d",
inp->inp_laddr.s_addr,
inp->inp_lport,
inp->inp_faddr.s_addr,
inp->inp_fport, tp->t_state, tp->t_flags);
if (tp == 0)
goto dropwithreset;
if (tp->t_state == TCPS_CLOSED)
goto drop;
// Unscale the window into a 32-bit value.
if ((tiflags & TH_SYN) == 0)
tiwin = ti->ti_win << tp->snd_scale;
else
tiwin = ti->ti_win;
so = inp->inp_socket;
DEBUG("socket info, options=%x", so->so_options);
if (so->so_options & (SO_DEBUG|SO_ACCEPTCONN)) {
if (so->so_options & SO_DEBUG) {
ostate = tp->t_state;
g_tcp_saveti = *ti;
}
if (so->so_options & SO_ACCEPTCONN) {
if ((tiflags & (TH_RST|TH_ACK|TH_SYN)) != TH_SYN) {
// Note: dropwithreset makes sure we don't
// send a reset in response to a RST.
if (tiflags & TH_ACK) {
g_tcpstat.tcps_badsyn++;
goto dropwithreset;
}
DEBUG("SYN is expected, tiflags=%d", tiflags);
goto drop;
}
so = sonewconn(so, 0);
if (so == 0) {
DEBUG("failed to create new connection, tiflags=%d", tiflags);
goto drop;
}
// Mark socket as temporary until we're
// committed to keeping it. The code at
// ``drop'' and ``dropwithreset'' check the
// flag dropsocket to see if the temporary
// socket created here should be discarded.
// We mark the socket as discardable until
// we're committed to it below in TCPS_LISTEN.
dropsocket++;
inp = (struct inpcb *)so->so_pcb;
inp->inp_laddr = ti->ti_dst;
inp->inp_lport = ti->ti_dport;
// BSD >= 4.3
inp->inp_options = ip_srcroute();
tp = intotcpcb(inp);
tp->t_state = TCPS_LISTEN;
// Compute proper scaling value from buffer space
while (tp->request_r_scale < TCP_MAX_WINSHIFT &&
TCP_MAXWIN << tp->request_r_scale < so->so_rcv->sb_hiwat)
tp->request_r_scale++;
}
}
// Segment received on connection.
// Reset idle time and keep-alive timer.
tp->t_idle = 0;
tp->t_timer[TCPT_KEEP] = g_tcp_keepidle;
// Process options if not in LISTEN state,
// else do it below (after getting remote address).
if (optp && tp->t_state != TCPS_LISTEN)
tcp_dooptions(tp, optp, optlen, ti,
&ts_present, &ts_val, &ts_ecr);
// Header prediction: check for the two common cases
// of a uni-directional data xfer. If the packet has
// no control flags, is in-sequence, the window didn't
// change and we're not retransmitting, it's a
// candidate. If the length is zero and the ack moved
// forward, we're the sender side of the xfer. Just
// free the data acked & wake any higher level process
// that was blocked waiting for space. If the length
// is non-zero and the ack didn't move, we're the
// receiver side. If we're getting packets in-order
// (the reassembly queue is empty), add the data to
// the socket buffer and note that we need a delayed ack.
if (tp->t_state == TCPS_ESTABLISHED &&
(tiflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ACK)) == TH_ACK &&
(!ts_present || TSTMP_GEQ(ts_val, tp->ts_recent)) &&
ti->ti_seq == tp->rcv_nxt &&
tiwin && tiwin == tp->snd_wnd &&
tp->snd_nxt == tp->snd_max) {
// If last ACK falls within this segment's sequence numbers,
// record the timestamp.
if ( ts_present && TSTMP_GEQ(ts_val, tp->ts_recent) &&
SEQ_LEQ(ti->ti_seq, tp->last_ack_sent) ){
tp->ts_recent_age = g_tcp_now;
tp->ts_recent = ts_val;
}
if (ti->ti_len == 0) {
if (SEQ_GT(ti->ti_ack, tp->snd_una) &&
SEQ_LEQ(ti->ti_ack, tp->snd_max) &&
tp->snd_cwnd >= tp->snd_wnd) {
// this is a pure ack for outstanding data.
++g_tcpstat.tcps_predack;
if (ts_present)
tcp_xmit_timer(tp, g_tcp_now-ts_ecr+1);
else if (tp->t_rtt &&
SEQ_GT(ti->ti_ack, tp->t_rtseq))
tcp_xmit_timer(tp, tp->t_rtt);
acked = ti->ti_ack - tp->snd_una;
g_tcpstat.tcps_rcvackpack++;
g_tcpstat.tcps_rcvackbyte += acked;
TRACE("drop so_snd buffer, drop_bytes=%d, len=%d",
acked, so->so_snd.sb_cc);
sbdrop(so->so_snd, acked);
tp->snd_una = ti->ti_ack;
usn_free_cmbuf(m);
// If all outstanding data are acked, stop
// retransmit timer, otherwise restart timer
// using current (possibly backed-off) value.
// If process is waiting for space,
// wakeup/selwakeup/signal. If data
// are ready to send, let tcp_output
// decide between more output or persist.
if (tp->snd_una == tp->snd_max)
tp->t_timer[TCPT_REXMT] = 0;
else if (tp->t_timer[TCPT_PERSIST] == 0)
tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
if (so->so_options & SO_DEBUG)
tcp_trace(TA_INPUT, ostate, tp, &g_tcp_saveti, 0);
//if (so->so_snd->sb_flags & SB_NOTIFY) {
// usnet_tcpin_wwakeup(so, USN_TCP_IN, usn_tcpev_sbnotify, 0);
// sowwakeup(so);
//}
// send buffer is available for app thread.
usnet_tcpin_wwakeup(so, USN_TCP_IN, USN_TCPEV_WRITE, 0);
if (so->so_snd->sb_cc)
tcp_output(tp);
return;
}
} else if (ti->ti_ack == tp->snd_una &&
tp->seg_next == (struct tcpiphdr *)tp &&
ti->ti_len <= sbspace(so->so_rcv)) {
// this is a pure, in-sequence data packet
// with nothing on the reassembly queue and
// we have enough buffer space to take it.
++g_tcpstat.tcps_preddat;
tp->rcv_nxt += ti->ti_len;
g_tcpstat.tcps_rcvpack++;
g_tcpstat.tcps_rcvbyte += ti->ti_len;
// Drop TCP, IP headers and TCP options then add data
// to socket buffer.
m->head += sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
m->mlen -= sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
TRACE("add data to rcv buf");
sbappend(so->so_rcv, m);
sorwakeup(so);
// new data is available for app threads.
usnet_tcpin_rwakeup(so, USN_TCP_IN, USN_TCPEV_READ, m);
if (so->so_options & SO_DEBUG) {
TRACE("tcp trace, so_options=%d", so->so_options);
tcp_trace(TA_INPUT, ostate, tp, &g_tcp_saveti, 0);
}
tp->t_flags |= TF_DELACK;
return;
}
}
// Drop TCP, IP headers and TCP options.
m->head += sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
m->mlen -= sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
// Calculate amount of space in receive window,
// and then do TCP input processing.
// Receive window is amount of space in rcv queue,
// but not less than advertised window.
{
int win;
win = sbspace(so->so_rcv);
if (win < 0)
win = 0;
tp->rcv_wnd = max(win, (int)(tp->rcv_adv - tp->rcv_nxt));
}
switch (tp->t_state) {
// If the state is LISTEN then ignore segment if it contains an RST.
// If the segment contains an ACK then it is bad and send a RST.
// If it does not contain a SYN then it is not interesting; drop it.
// Don't bother responding if the destination was a broadcast.
// Otherwise initialize tp->rcv_nxt, and tp->irs, select an initial
// tp->iss, and send a segment:
// <SEQ=ISS><ACK=RCV_NXT><CTL=SYN,ACK>
// Also initialize tp->snd_nxt to tp->iss+1 and tp->snd_una to tp->iss.
// Fill in remote peer address fields if not previously specified.
// Enter SYN_RECEIVED state, and process any other fields of this
// segment in this state.
case TCPS_LISTEN: {
usn_mbuf_t *am;
struct usn_sockaddr_in *sin;
if (tiflags & TH_RST)
goto drop;
if (tiflags & TH_ACK)
goto dropwithreset;
if ((tiflags & TH_SYN) == 0)
goto drop;
// RFC1122 4.2.3.10, p. 104: discard bcast/mcast SYN
// in_broadcast() should never return true on a received
// packet with M_BCAST not set.
//if (m->m_flags & (M_BCAST|M_MCAST) ||
// IN_MULTICAST(ntohl(ti->ti_dst.s_addr)))
// goto drop;
am = usn_get_mbuf(0, BUF_MSIZE, 0); // XXX: the size!
if (am == NULL)
goto drop;
am->mlen = sizeof (struct usn_sockaddr_in);
sin = mtod(am, struct usn_sockaddr_in *);
sin->sin_family = AF_INET;
sin->sin_len = sizeof(*sin);
sin->sin_addr = ti->ti_src;
sin->sin_port = ti->ti_sport;
bzero((caddr_t)sin->sin_zero, sizeof(sin->sin_zero));
laddr = inp->inp_laddr;
if (inp->inp_laddr.s_addr == USN_INADDR_ANY)
inp->inp_laddr = ti->ti_dst;
if (in_pcbconnect(inp, am)) {
inp->inp_laddr = laddr;
usn_free_mbuf(am);
goto drop;
}
usn_free_mbuf(am);
tp->t_template = tcp_template(tp);
if (tp->t_template == 0) {
tp = tcp_drop(tp, ENOBUFS);
dropsocket = 0; // socket is already gone
goto drop;
}
if (optp)
tcp_dooptions(tp, optp, optlen, ti,
&ts_present, &ts_val, &ts_ecr);
if (iss)
tp->iss = iss;
else
tp->iss = g_tcp_iss;
g_tcp_iss += TCP_ISSINCR/4;
tp->irs = ti->ti_seq;
tcp_sendseqinit(tp);
tcp_rcvseqinit(tp);
tp->t_flags |= TF_ACKNOW;
TRACE("change tcp state to TCPS_SYN_RECEIVED, state=%d, tp_flags=%d",
tp->t_state, tp->t_flags);
tp->t_state = TCPS_SYN_RECEIVED;
// tcp event
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_SYN_RECEIVED, 0);
tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_INIT;
dropsocket = 0; // committed to socket
g_tcpstat.tcps_accepts++;
goto trimthenstep6;
}
// If the state is SYN_SENT:
// if seg contains an ACK, but not for our SYN, drop the input.
// if seg contains a RST, then drop the connection.
// if seg does not contain SYN, then drop it.
// Otherwise this is an acceptable SYN segment
// initialize tp->rcv_nxt and tp->irs
// if seg contains ack then advance tp->snd_una
// if SYN has been acked change to ESTABLISHED else SYN_RCVD state
// arrange for segment to be acked (eventually)
// continue processing rest of data/controls, beginning with URG
case TCPS_SYN_SENT:
if ((tiflags & TH_ACK) &&
(SEQ_LEQ(ti->ti_ack, tp->iss) ||
SEQ_GT(ti->ti_ack, tp->snd_max)))
goto dropwithreset;
if (tiflags & TH_RST) {
if (tiflags & TH_ACK)
tp = tcp_drop(tp, ECONNREFUSED);
goto drop;
}
if ((tiflags & TH_SYN) == 0)
goto drop;
if (tiflags & TH_ACK) {
tp->snd_una = ti->ti_ack;
if (SEQ_LT(tp->snd_nxt, tp->snd_una))
tp->snd_nxt = tp->snd_una;
tp->t_timer[TCPT_REXMT] = 0;
}
tp->irs = ti->ti_seq;
tcp_rcvseqinit(tp);
tp->t_flags |= TF_ACKNOW;
TRACE("ack now, tp flags=%d", tp->t_flags);
// XXX: remove second test.
if (tiflags & TH_ACK /*&& SEQ_GT(tp->snd_una, tp->iss)*/) {
g_tcpstat.tcps_connects++;
soisconnected(so);
TRACE("change tcp state to TCPS_ESTABLISHED,"
" state=%d, tp_flags=%d", tp->t_state, tp->t_flags);
tp->t_state = TCPS_ESTABLISHED;
// Do window scaling on this connection?
if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) ==
(TF_RCVD_SCALE|TF_REQ_SCALE)) {
tp->snd_scale = tp->requested_s_scale;
tp->rcv_scale = tp->request_r_scale;
}
tcp_reass(tp, (struct tcpiphdr *)0, (usn_mbuf_t *)0);
// if we didn't have to retransmit the SYN,
// use its rtt as our initial srtt & rtt var.
if (tp->t_rtt)
tcp_xmit_timer(tp, tp->t_rtt);
} else {
TRACE("change tcp state to TCPS_SYN_RECEIVED, state=%d, tp_flags=%d",
tp->t_state, tp->t_flags);
tp->t_state = TCPS_SYN_RECEIVED;
// tcp event
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_SYN_RECEIVED, 0);
}
trimthenstep6:
// Advance ti->ti_seq to correspond to first data byte.
// If data, trim to stay within window,
// dropping FIN if necessary.
ti->ti_seq++;
if (ti->ti_len > tp->rcv_wnd) {
todrop = ti->ti_len - tp->rcv_wnd;
m_adj(m, -todrop);
ti->ti_len = tp->rcv_wnd;
tiflags &= ~TH_FIN;
g_tcpstat.tcps_rcvpackafterwin++;
g_tcpstat.tcps_rcvbyteafterwin += todrop;
}
tp->snd_wl1 = ti->ti_seq - 1;
tp->rcv_up = ti->ti_seq;
goto step6;
}
// States other than LISTEN or SYN_SENT.
// First check timestamp, if present.
// Then check that at least some bytes of segment are within
// receive window. If segment begins before rcv_nxt,
// drop leading data (and SYN); if nothing left, just ack.
//
// RFC 1323 PAWS: If we have a timestamp reply on this segment
// and it's less than ts_recent, drop it.
if (ts_present && (tiflags & TH_RST) == 0 && tp->ts_recent &&
TSTMP_LT(ts_val, tp->ts_recent)) {
// Check to see if ts_recent is over 24 days old.
if ((int)(g_tcp_now - tp->ts_recent_age) > TCP_PAWS_IDLE) {
// Invalidate ts_recent. If this segment updates
// ts_recent, the age will be reset later and ts_recent
// will get a valid value. If it does not, setting
// ts_recent to zero will at least satisfy the
// requirement that zero be placed in the timestamp
// echo reply when ts_recent isn't valid. The
// age isn't reset until we get a valid ts_recent
// because we don't want out-of-order segments to be
// dropped when ts_recent is old.
tp->ts_recent = 0;
} else {
g_tcpstat.tcps_rcvduppack++;
g_tcpstat.tcps_rcvdupbyte += ti->ti_len;
g_tcpstat.tcps_pawsdrop++;
goto dropafterack;
}
}
todrop = tp->rcv_nxt - ti->ti_seq;
if (todrop > 0) {
if (tiflags & TH_SYN) {
tiflags &= ~TH_SYN;
ti->ti_seq++;
if (ti->ti_urp > 1)
ti->ti_urp--;
else
tiflags &= ~TH_URG;
todrop--;
}
if ( todrop >= ti->ti_len ||
( todrop == ti->ti_len && (tiflags & TH_FIN ) == 0 ) ) {
// Any valid FIN must be to the left of the window.
// At this point the FIN must be a duplicate or
// out of sequence; drop it.
tiflags &= ~TH_FIN;
// Send an ACK to resynchronize and drop any data
// But keep on processing for RST or ACK.
tp->t_flags |= TF_ACKNOW;
TRACE("send ack now to resync, tp_flags=%d", tp->t_flags);
todrop = ti->ti_len;
g_tcpstat.tcps_rcvdupbyte += ti->ti_len;
g_tcpstat.tcps_rcvduppack++;
} else {
g_tcpstat.tcps_rcvpartduppack++;
g_tcpstat.tcps_rcvpartdupbyte += ti->ti_len;
}
m_adj(m, todrop);
ti->ti_seq += todrop;
ti->ti_len -= todrop;
if (ti->ti_urp > todrop)
ti->ti_urp -= todrop;
else {
tiflags &= ~TH_URG;
ti->ti_urp = 0;
}
}
// If new data are received on a connection after the
// user processes are gone, then RST the other end.
if ((so->so_state & USN_NOFDREF) &&
tp->t_state > TCPS_CLOSE_WAIT && ti->ti_len) {
tp = tcp_close(tp);
g_tcpstat.tcps_rcvafterclose++;
goto dropwithreset;
}
// If segment ends after window, drop trailing data
// (and PUSH and FIN); if nothing left, just ACK.
todrop = (ti->ti_seq+ti->ti_len) - (tp->rcv_nxt+tp->rcv_wnd);
if (todrop > 0) {
g_tcpstat.tcps_rcvpackafterwin++;
if (todrop >= ti->ti_len) {
g_tcpstat.tcps_rcvbyteafterwin += ti->ti_len;
// If a new connection request is received
// while in TIME_WAIT, drop the old connection
// and start over if the sequence numbers
// are above the previous ones.
if (tiflags & TH_SYN &&
tp->t_state == TCPS_TIME_WAIT &&
SEQ_GT(ti->ti_seq, tp->rcv_nxt)) {
iss = tp->snd_nxt + TCP_ISSINCR;
tp = tcp_close(tp);
goto findpcb;
}
// If window is closed can only take segments at
// window edge, and have to drop data and PUSH from
// incoming segments. Continue processing, but
// remember to ack. Otherwise, drop segment
// and ack.
if (tp->rcv_wnd == 0 && ti->ti_seq == tp->rcv_nxt) {
tp->t_flags |= TF_ACKNOW;
g_tcpstat.tcps_rcvwinprobe++;
} else
goto dropafterack;
} else
g_tcpstat.tcps_rcvbyteafterwin += todrop;
m_adj(m, -todrop);
ti->ti_len -= todrop;
tiflags &= ~(TH_PUSH|TH_FIN);
}
// check valid timestamp. Replace code above.
if (ts_present && TSTMP_GEQ(ts_val, tp->ts_recent) &&
SEQ_LEQ(ti->ti_seq, tp->last_ack_sent) ) {
tp->ts_recent_age = g_tcp_now;
tp->ts_recent = ts_val;
}
// If the RST bit is set examine the state:
// SYN_RECEIVED STATE:
// If passive open, return to LISTEN state.
// If active open, inform user that connection was refused.
// ESTABLISHED, FIN_WAIT_1, FIN_WAIT2, CLOSE_WAIT STATES:
// Inform user that connection was reset, and close tcb.
// CLOSING, LAST_ACK, TIME_WAIT STATES
// Close the tcb.
if (tiflags&TH_RST) switch (tp->t_state) {
case TCPS_SYN_RECEIVED:
so->so_error = ECONNREFUSED;
goto close;
case TCPS_ESTABLISHED:
case TCPS_FIN_WAIT_1:
case TCPS_FIN_WAIT_2:
case TCPS_CLOSE_WAIT:
so->so_error = ECONNRESET;
close:
DEBUG("change tcp state to TCPS_CLOSED, state=%d", tp->t_state);
tp->t_state = TCPS_CLOSED;
// tcp event
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_CLOSED, 0);
g_tcpstat.tcps_drops++;
tp = tcp_close(tp);
goto drop;
case TCPS_CLOSING:
case TCPS_LAST_ACK:
case TCPS_TIME_WAIT:
tp = tcp_close(tp);
goto drop;
}
// If a SYN is in the window, then this is an
// error and we send an RST and drop the connection.
if (tiflags & TH_SYN) {
tp = tcp_drop(tp, ECONNRESET);
goto dropwithreset;
}
// If the ACK bit is off we drop the segment and return.
if ((tiflags & TH_ACK) == 0)
goto drop;
// Ack processing.
switch (tp->t_state) {
// In SYN_RECEIVED state if the ack ACKs our SYN then enter
// ESTABLISHED state and continue processing, otherwise
// send an RST.
case TCPS_SYN_RECEIVED:
if (SEQ_GT(tp->snd_una, ti->ti_ack) ||
SEQ_GT(ti->ti_ack, tp->snd_max))
goto dropwithreset;
g_tcpstat.tcps_connects++;
DEBUG("change tcp state to TCPS_ESTABLISHED, state=%d", tp->t_state);
tp->t_state = TCPS_ESTABLISHED;
soisconnected(so);
// Do window scaling?
if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) ==
(TF_RCVD_SCALE|TF_REQ_SCALE)) {
tp->snd_scale = tp->requested_s_scale;
tp->rcv_scale = tp->request_r_scale;
}
tcp_reass(tp, (struct tcpiphdr *)0, (usn_mbuf_t *)0);
tp->snd_wl1 = ti->ti_seq - 1;
// fall into ...
// In ESTABLISHED state: drop duplicate ACKs; ACK out of range
// ACKs. If the ack is in the range
// tp->snd_una < ti->ti_ack <= tp->snd_max
// then advance tp->snd_una to ti->ti_ack and drop
// data from the retransmission queue. If this ACK reflects
// more up to date window information we update our window information.
case TCPS_ESTABLISHED:
case TCPS_FIN_WAIT_1:
case TCPS_FIN_WAIT_2:
case TCPS_CLOSE_WAIT:
case TCPS_CLOSING:
case TCPS_LAST_ACK:
case TCPS_TIME_WAIT:
if (SEQ_LEQ(ti->ti_ack, tp->snd_una)) {
if (ti->ti_len == 0 && tiwin == tp->snd_wnd) {
g_tcpstat.tcps_rcvdupack++;
// If we have outstanding data (other than
// a window probe), this is a completely
// duplicate ack (ie, window info didn't
// change), the ack is the biggest we've
// seen and we've seen exactly our rexmt
// threshhold of them, assume a packet
// has been dropped and retransmit it.
// Kludge snd_nxt & the congestion
// window so we send only this one
// packet.
//
// We know we're losing at the current
// window size so do congestion avoidance
// (set ssthresh to half the current window
// and pull our congestion window back to
// the new ssthresh).
//
// Dup acks mean that packets have left the
// network (they're now cached at the receiver)
// so bump cwnd by the amount in the receiver
// to keep a constant cwnd packets in the
// network.
if (tp->t_timer[TCPT_REXMT] == 0 ||
ti->ti_ack != tp->snd_una)
tp->t_dupacks = 0;
else if (++tp->t_dupacks == g_tcprexmtthresh) {
// congestion avoidance
tcp_seq onxt = tp->snd_nxt;
u_int win =
min(tp->snd_wnd, tp->snd_cwnd) / 2 / tp->t_maxseg;
if (win < 2)
win = 2;
tp->snd_ssthresh = win * tp->t_maxseg;
tp->t_timer[TCPT_REXMT] = 0;
tp->t_rtt = 0;
tp->snd_nxt = ti->ti_ack;
tp->snd_cwnd = tp->t_maxseg;
tcp_output(tp);
tp->snd_cwnd = tp->snd_ssthresh +
tp->t_maxseg * tp->t_dupacks;
if (SEQ_GT(onxt, tp->snd_nxt))
tp->snd_nxt = onxt;
goto drop;
} else if (tp->t_dupacks > g_tcprexmtthresh) {
tp->snd_cwnd += tp->t_maxseg;
tcp_output(tp);
goto drop;
}
} else
tp->t_dupacks = 0;
break;
}
// If the congestion window was inflated to account
// for the other side's cached packets, retract it.
if (tp->t_dupacks > g_tcprexmtthresh &&
tp->snd_cwnd > tp->snd_ssthresh)
tp->snd_cwnd = tp->snd_ssthresh;
tp->t_dupacks = 0;
if (SEQ_GT(ti->ti_ack, tp->snd_max)) {
g_tcpstat.tcps_rcvacktoomuch++;
goto dropafterack;
}
acked = ti->ti_ack - tp->snd_una;
g_tcpstat.tcps_rcvackpack++;
g_tcpstat.tcps_rcvackbyte += acked;
// If we have a timestamp reply, update smoothed
// round trip time. If no timestamp is present but
// transmit timer is running and timed sequence
// number was acked, update smoothed round trip time.
// Since we now have an rtt measurement, cancel the
// timer backoff (cf., Phil Karn's retransmit alg.).
// Recompute the initial retransmit timer.
if (ts_present)
tcp_xmit_timer(tp, g_tcp_now-ts_ecr+1);
else if (tp->t_rtt && SEQ_GT(ti->ti_ack, tp->t_rtseq))
tcp_xmit_timer(tp,tp->t_rtt);
// If all outstanding data is acked, stop retransmit
// timer and remember to restart (more output or persist).
// If there is more data to be acked, restart retransmit
// timer, using current (possibly backed-off) value.
if (ti->ti_ack == tp->snd_max) {
tp->t_timer[TCPT_REXMT] = 0;
DEBUG("change needoutput to 1");
needoutput = 1;
tp->t_flags |= TF_NEEDOUTPUT;
} else if (tp->t_timer[TCPT_PERSIST] == 0)
tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
// When new data is acked, open the congestion window.
// If the window gives us less than ssthresh packets
// in flight, open exponentially (maxseg per packet).
// Otherwise open linearly: maxseg per window
// (maxseg * (maxseg / cwnd) per packet).
{
u_int cw = tp->snd_cwnd;
u_int incr = tp->t_maxseg;
if (cw > tp->snd_ssthresh)
incr = incr * incr / cw;
tp->snd_cwnd = min(cw + incr, TCP_MAXWIN<<tp->snd_scale);
}
if (acked > so->so_snd->sb_cc) {
tp->snd_wnd -= so->so_snd->sb_cc;
DEBUG("drop all so_snd buffer, drop_bytes=%d, acked=%d",
so->so_snd->sb_cc, acked);
sbdrop(so->so_snd, (int)so->so_snd->sb_cc);
ourfinisacked = 1;
} else {
DEBUG("drop so_snd buffer, drop_bytes=%d, len=%d", acked, so->so_snd->sb_cc);
sbdrop(so->so_snd, acked);
tp->snd_wnd -= acked;
ourfinisacked = 0;
}
//if (so->so_snd->sb_flags & SB_NOTIFY) {
sowwakeup(so);
usnet_tcpin_wwakeup(so, USN_TCP_IN, USN_TCPEV_WRITE, 0);
//}
tp->snd_una = ti->ti_ack;
if (SEQ_LT(tp->snd_nxt, tp->snd_una))
tp->snd_nxt = tp->snd_una;
switch (tp->t_state) {
// In FIN_WAIT_1 STATE in addition to the processing
// for the ESTABLISHED state if our FIN is now acknowledged
// then enter FIN_WAIT_2.
case TCPS_FIN_WAIT_1:
if (ourfinisacked) {
// If we can't receive any more
// data, then closing user can proceed.
// Starting the timer is contrary to the
// specification, but if we don't get a FIN
// we'll hang forever.
if (so->so_state & USN_CANTRCVMORE) {
soisdisconnected(so);
tp->t_timer[TCPT_2MSL] = g_tcp_maxidle;
}
DEBUG("change tcp state to TCPS_FIN_WAIT_2, state=%d", tp->t_state);
tp->t_state = TCPS_FIN_WAIT_2;
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_FIN_WAIT2, 0);
}
break;
// In CLOSING STATE in addition to the processing for
// the ESTABLISHED state if the ACK acknowledges our FIN
// then enter the TIME-WAIT state, otherwise ignore
// the segment.
case TCPS_CLOSING:
if (ourfinisacked) {
DEBUG("change tcp state to TCPS_TIME_WAIT, state=%d", tp->t_state);
tp->t_state = TCPS_TIME_WAIT;
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_TIME_WAIT, 0);
tcp_canceltimers(tp);
tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
soisdisconnected(so);
}
break;
// In LAST_ACK, we may still be waiting for data to drain
// and/or to be acked, as well as for the ack of our FIN.
// If our FIN is now acknowledged, delete the TCB,
// enter the closed state and return.
case TCPS_LAST_ACK:
if (ourfinisacked) {
tp = tcp_close(tp);
goto drop;
}
break;
// In TIME_WAIT state the only thing that should arrive
// is a retransmission of the remote FIN. Acknowledge
// it and restart the finack timer.
case TCPS_TIME_WAIT:
tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
goto dropafterack;
}
}
step6:
// Update window information.
// Don't look at window if no ACK: TAC's send garbage on first SYN.
if ((tiflags & TH_ACK) &&
(SEQ_LT(tp->snd_wl1, ti->ti_seq) ||
(tp->snd_wl1 == ti->ti_seq && (SEQ_LT(tp->snd_wl2, ti->ti_ack) ||
(tp->snd_wl2 == ti->ti_ack && tiwin > tp->snd_wnd) )) )) {
// keep track of pure window updates
if (ti->ti_len == 0 &&
tp->snd_wl2 == ti->ti_ack && tiwin > tp->snd_wnd)
g_tcpstat.tcps_rcvwinupd++;
tp->snd_wnd = tiwin;
tp->snd_wl1 = ti->ti_seq;
tp->snd_wl2 = ti->ti_ack;
if (tp->snd_wnd > tp->max_sndwnd)
tp->max_sndwnd = tp->snd_wnd;
DEBUG("change needoutput to 1");
tp->t_flags |= TF_NEEDOUTPUT;
needoutput = 1;
}
// Process segments with URG.
if ((tiflags & TH_URG) && ti->ti_urp &&
TCPS_HAVERCVDFIN(tp->t_state) == 0) {
// This is a kludge, but if we receive and accept
// random urgent pointers, we'll crash in
// soreceive. It's hard to imagine someone
// actually wanting to send this much urgent data.
if (ti->ti_urp + so->so_rcv->sb_cc > g_sb_max) {
ti->ti_urp = 0; // XXX
tiflags &= ~TH_URG; // XXX
goto dodata; // XXX
}
// If this segment advances the known urgent pointer,
// then mark the data stream. This should not happen
// in CLOSE_WAIT, CLOSING, LAST_ACK or TIME_WAIT STATES since
// a FIN has been received from the remote side.
// In these states we ignore the URG.
//
// According to RFC961 (Assigned Protocols),
// the urgent pointer points to the last octet
// of urgent data. We continue, however,
// to consider it to indicate the first octet
// of data past the urgent section as the original
// spec states (in one of two places).
if (SEQ_GT(ti->ti_seq+ti->ti_urp, tp->rcv_up)) {
tp->rcv_up = ti->ti_seq + ti->ti_urp;
so->so_oobmark = so->so_rcv->sb_cc +
(tp->rcv_up - tp->rcv_nxt) - 1;
if (so->so_oobmark == 0)
so->so_state |= USN_RCVATMARK;
sohasoutofband(so);
// send async event to app threads.
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPEV_OUTOFBOUND, 0);
tp->t_oobflags &= ~(TCPOOB_HAVEDATA | TCPOOB_HADDATA);
}
// Remove out of band data so doesn't get presented to user.
// This can happen independent of advancing the URG pointer,
// but if two URG's are pending at once, some out-of-band
// data may creep in... ick.
if (ti->ti_urp <= ti->ti_len
#ifdef SO_OOBINLINE
&& (so->so_options & SO_OOBINLINE) == 0
#endif
)
tcp_pulloutofband(so, ti, m);
} else
// If no out of band data is expected,
// pull receive urgent pointer along
// with the receive window.
if (SEQ_GT(tp->rcv_nxt, tp->rcv_up))
tp->rcv_up = tp->rcv_nxt;
dodata: // XXX
#ifdef DUMP_PAYLOAD
DEBUG("Handle data");
dump_chain(m,"tcp");
#endif
// Process the segment text, merging it into the TCP sequencing queue,
// and arranging for acknowledgment of receipt if necessary.
// This process logically involves adjusting tp->rcv_wnd as data
// is presented to the user (this happens in tcp_usrreq.c,
// case PRU_RCVD). If a FIN has already been received on this
// connection then we just ignore the text.
if ((ti->ti_len || (tiflags&TH_FIN)) &&
TCPS_HAVERCVDFIN(tp->t_state) == 0) {
TCP_REASS(tp, ti, m, so, tiflags);
// Note the amount of data that peer has sent into
// our window, in order to estimate the sender's
// buffer size.
len = so->so_rcv->sb_hiwat - (tp->rcv_adv - tp->rcv_nxt);
} else {
usn_free_cmbuf(m);
tiflags &= ~TH_FIN;
}
// If FIN is received ACK the FIN and let the user know
// that the connection is closing.
if (tiflags & TH_FIN) {
if (TCPS_HAVERCVDFIN(tp->t_state) == 0) {
socantrcvmore(so);
tp->t_flags |= TF_ACKNOW;
TRACE("ack FIN now, tp flags=%d", tp->t_flags);
tp->rcv_nxt++;
}
switch (tp->t_state) {
// In SYN_RECEIVED and ESTABLISHED STATES
// enter the CLOSE_WAIT state.
case TCPS_SYN_RECEIVED:
case TCPS_ESTABLISHED:
TRACE("change tcp state to TCPS_CLOSE_WAIT, state=%d", tp->t_state);
tp->t_state = TCPS_CLOSE_WAIT;
soewakeup(so, 0);
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_CLOSE_WAIT, 0);
break;
// If still in FIN_WAIT_1 STATE FIN has not been acked so
// enter the CLOSING state.
case TCPS_FIN_WAIT_1:
TRACE("change tcp state to TCPS_CLOSING, state=%d", tp->t_state);
tp->t_state = TCPS_CLOSING;
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_CLOSING, 0);
break;
// In FIN_WAIT_2 state enter the TIME_WAIT state,
// starting the time-wait timer, turning off the other
// standard timers.
case TCPS_FIN_WAIT_2:
TRACE("change tcp state to TCPS_TIME_WAIT, state=%d", tp->t_state);
tp->t_state = TCPS_TIME_WAIT;
tcp_canceltimers(tp);
tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
soisdisconnected(so);
usnet_tcpin_ewakeup(so, USN_TCP_IN, USN_TCPST_TIME_WAIT, 0);
break;
// In TIME_WAIT state restart the 2 MSL time_wait timer.
case TCPS_TIME_WAIT:
tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
break;
}
}
if (so->so_options & SO_DEBUG) {
TRACE("tcp trace, so_options=%d", so->so_options);
tcp_trace(TA_INPUT, ostate, tp, &g_tcp_saveti, 0);
}
// Return any desired output.
//if (needoutput || (tp->t_flags & TF_ACKNOW)){
if (tp->t_flags & TF_NEEDOUTPUT || (tp->t_flags & TF_ACKNOW)){
TRACE("ack now or need to ouput, tp->t_flags=%d", tp->t_flags);
tcp_output(tp);
}
return;
dropafterack:
TRACE("dropafterack");
// Generate an ACK dropping incoming segment if it occupies
// sequence space, where the ACK reflects our state.
if (tiflags & TH_RST)
goto drop;
usn_free_cmbuf(m);
tp->t_flags |= TF_ACKNOW;
TRACE("ack now, tp flags=%d", tp->t_flags);
tcp_output(tp);
return;
dropwithreset:
TRACE("dropwithreset");
// Generate a RST, dropping incoming segment.
// Make ACK acceptable to originator of segment.
// Don't bother to respond if destination was broadcast/multicast.
#define USN_MULTICAST(i) (((u_int)(i) & 0xf0000000) == 0xe0000000)
if ((tiflags & TH_RST) || m->flags & (BUF_BCAST|BUF_MCAST) ||
USN_MULTICAST(ntohl(ti->ti_dst.s_addr)))
goto drop;
if (tiflags & TH_ACK)
tcp_respond(tp, ti, m, (tcp_seq)0, ti->ti_ack, TH_RST);
else {
if (tiflags & TH_SYN)
ti->ti_len++;
tcp_respond(tp, ti, m, ti->ti_seq+ti->ti_len, (tcp_seq)0,
TH_RST|TH_ACK);
}
// destroy temporarily created socket
if (dropsocket)
soabort(so);
return;
drop:
TRACE("drop");
// Drop space held by incoming segment and return.
if (tp && (tp->t_inpcb->inp_socket->so_options & SO_DEBUG)) {
TRACE("tcp trace: drop a socket");
tcp_trace(TA_DROP, ostate, tp, &g_tcp_saveti, 0);
}
usn_free_cmbuf(m);
// destroy temporarily created socket
if (dropsocket)
soabort(so);
return;
}
void tcp_input(IO_Rec *recs, int nr_recs, iphost_st *host_st,
intf_st *ifs, uint32_t m_flags, uint8_t iphlen)
{
/* iphlen: size of ip header incl. possible options */
register struct tcpiphdr *ti;
uint16_t len, tlen; /* tlen: ip length without IP header start of code */
struct tcphdr *tcphdr;
uint16_t sum;
uint8_t tcphlen; /* tcp header length, incl. options */
TRC(printf("tcp input with ip header length %d\n", iphlen));
host_st->tcpstat.tcps_rcvtotal++;
ti = ( struct tcpiphdr *)(recs[0].base + sizeof(struct ether_header));
tcphdr = (struct tcphdr *)(recs[0].base +
sizeof(struct ether_header) +
iphlen);
tcphlen = tcphdr->th_off << 2;
TRC(printf("tcp header length %d\n", tcphlen));
if (iphlen > sizeof (struct ip))
ip_stripoptions(recs);
/* XXX AND: iphlen now wrong? */
if ((iphlen + tcphlen) < sizeof (struct tcpiphdr)) {
host_st->tcpstat.tcps_rcvshort++;
goto drop;
}
/*
* Checksum extended TCP header and data.
*/
tlen = ((struct ip *)ti)->ip_len;
TRC(printf("tcp option length %d\n", tcphlen- sizeof(struct tcphdr)));
len = sizeof (struct ip) + tlen;
TRC(printf("tlen %d, len %d\n", tlen, len));
ti->ti_next = ti->ti_prev = 0;
ti->ti_x1 = 0;
ti->ti_len = htons((uint16_t)tlen);
if (recs[1].len == 0)
sum = in_cksum((uint16_t *)ti, len);
else {
sum = cksum_morerecs((uint16_t *)ti,
recs[0].len - sizeof(struct ether_header),
recs,
nr_recs);
TRC(printf("tcp_input: checksum more recs %d\n", sum));
}
if (sum) {
printf("flowman: tcp_input: checksum error (%x)\n", sum);
host_st->tcpstat.tcps_rcvbadsum++;
goto drop;
}
TRC(else
printf("tcp_input: checksum OK\n"));
TRC(show_all(ti));
/*
* Convert TCP protocol specific fields to host format.
*/
ti->ti_seq = ntoh32(ti->ti_seq);
ti->ti_ack = ntoh32(ti->ti_ack);
ti->ti_win = ntoh16(ti->ti_win);
ti->ti_urp = ntoh16(ti->ti_urp);
if (tcphdr->th_flags & TH_SYN) {
TRC(printf("tcp syn message\n"));
goto dropwithreset;
}
/* AND: shouldn't really get here */
#define F(f, c) (tcphdr->th_flags & f)? c : '-'
printf("flowman: tcp_input: no SYN set (%c%c%c%c%c%c), ignoring\n",
F(TH_FIN, 'f'),
F(TH_SYN, 's'),
F(TH_RST, 'r'),
F(TH_PUSH, 'p'),
F(TH_ACK, 'a'),
F(TH_URG, 'u'));
#undef F
return;
dropwithreset:
/*
* Generate a RST, dropping incoming segment.
* Make ACK acceptable to originator of segment.
* Don't bother to respond if destination was broadcast/multicast.
*/
if ((tcphdr->th_flags & TH_RST) || m_flags & (M_BCAST|M_MCAST) ||
IN_MULTICAST(ntoh32(ti->ti_dst.s_addr)))
goto drop;
if (tcphdr->th_flags & TH_ACK) {
tcp_respond(ti, recs, nr_recs, host_st, ifs,
(tcp_seq)0, ti->ti_ack, TH_RST);
} else {
TRC(printf("tcp_input: else ti_len %d, iphlen %d \n",
ntoh16(ti->ti_len), iphlen));
/* Thiemo setting ti_len to host order and reducing length field*/
ti->ti_len = (ntoh16(ti->ti_len) -
iphlen -
(tcphlen - sizeof(struct tcphdr)));
if (tcphdr->th_flags & TH_SYN)
ti->ti_len++;
TRC(printf("ti-ti_len network: %d\n", ti->ti_len));
tcp_respond(ti, recs, nr_recs, host_st, ifs,
ti->ti_seq + ti->ti_len, (tcp_seq)0,
TH_RST|TH_ACK);
}
return;
drop:
printf("flowman: tcp_input: ditched packet\n");
}
static void
rxk_input(struct mbuf *am, int iphlen)
{
void (*tproc) ();
register unsigned short *tsp;
int hdr;
struct udphdr *tu;
register struct ip *ti;
struct udpiphdr *tvu;
register int i;
char *phandle;
afs_int32 code;
struct sockaddr_in taddr;
int tlen;
short port;
int data_len, comp_sum;
SPLVAR;
NETPRI;
/* make sure we have base ip and udp headers in first mbuf */
if (iphlen > sizeof(struct ip)) {
ip_stripoptions(am, NULL);
iphlen = sizeof(struct ip);
}
if (am->m_len < sizeof(struct udpiphdr)) {
am = m_pullup(am, sizeof(struct udpiphdr));
if (!am) {
USERPRI;
return;
}
}
ti = mtod(am, struct ip *);
/* skip basic ip hdr */
tu = (struct udphdr *)(((char *)ti) + sizeof(struct ip));
/* now read the port out */
port = tu->uh_dport;
if (port) {
for (tsp = rxk_ports, i = 0; i < MAXRXPORTS; i++) {
if (*tsp++ == port) {
/* checksum the packet */
/*
* Make mbuf data length reflect UDP length.
* If not enough data to reflect UDP length, drop.
*/
tvu = (struct udpiphdr *)ti;
tlen = ntohs((u_short) tvu->ui_ulen);
if ((int)ti->ip_len != tlen) {
if (tlen > (int)ti->ip_len) {
m_free(am);
USERPRI;
return;
}
m_adj(am, tlen - (int)ti->ip_len);
}
/* deliver packet to rx */
taddr.sin_family = AF_INET; /* compute source address */
taddr.sin_port = tu->uh_sport;
taddr.sin_addr.s_addr = ti->ip_src.s_addr;
taddr.sin_len = sizeof(taddr);
tvu = (struct udpiphdr *)ti; /* virtual udp structure, for cksum */
/* handle the checksum. Note that this code damages the actual ip
* header (replacing it with the virtual one, which is the same size),
* so we must ensure we get everything out we need, first */
if (tu->uh_sum != 0) {
/* if the checksum is there, always check it. It's crazy not
* to, unless you can really be sure that your
* underlying network (and interfaces and drivers and
* DMA hardware, etc!) is error-free. First, fill
* in entire virtual ip header. */
memset(tvu->ui_i.ih_x1, 0, 9);
tvu->ui_len = tvu->ui_ulen;
tlen = ntohs((unsigned short)(tvu->ui_ulen));
if (in_cksum(am, sizeof(struct ip) + tlen)) {
/* checksum, including cksum field, doesn't come out 0, so
* this packet is bad */
m_freem(am);
USERPRI;
return;
}
}
/*
* 28 is IP (20) + UDP (8) header. ulen includes
* udp header, and we *don't* tell RX about udp
* header either. So, we remove those 8 as well.
*/
data_len = ntohs(tu->uh_ulen);
data_len -= 8;
if (!(*rxk_GetPacketProc) (&phandle, data_len)) {
if (rx_mb_to_packet(am, m_freem, 28, data_len, phandle)) {
/* XXX should just increment counter here.. */
printf("rx: truncated UDP packet\n");
rxi_FreePacket(phandle);
} else
(*rxk_PacketArrivalProc) (phandle, &taddr,
rxk_portRocks[i], data_len);
} else
m_freem(am);
USERPRI;
return;
}
}
}
/* if we get here, try to deliver packet to udp */
if (tproc = parent_proto.pr_input)
(*tproc) (am, iphlen);
USERPRI;
return;
}
/* m->m_data points at ip packet header
* m->m_len length ip packet
* ip->ip_len length data (IPDU)
*/
void
udp_input(PNATState pData, register struct mbuf *m, int iphlen)
{
register struct ip *ip;
register struct udphdr *uh;
int len;
struct ip save_ip;
struct socket *so;
int ret;
int ttl;
LogFlowFunc(("ENTER: m = %p, iphlen = %d\n", m, iphlen));
ip = mtod(m, struct ip *);
Log2(("%RTnaipv4 iphlen = %d\n", ip->ip_dst, iphlen));
udpstat.udps_ipackets++;
/*
* Strip IP options, if any; should skip this,
* make available to user, and use on returned packets,
* but we don't yet have a way to check the checksum
* with options still present.
*/
if (iphlen > sizeof(struct ip))
{
ip_stripoptions(m, (struct mbuf *)0);
iphlen = sizeof(struct ip);
}
/*
* Get IP and UDP header together in first mbuf.
*/
ip = mtod(m, struct ip *);
uh = (struct udphdr *)((caddr_t)ip + iphlen);
/*
* Make mbuf data length reflect UDP length.
* If not enough data to reflect UDP length, drop.
*/
len = RT_N2H_U16((u_int16_t)uh->uh_ulen);
Assert((ip->ip_len == len));
Assert((ip->ip_len + iphlen == m_length(m, NULL)));
if (ip->ip_len != len)
{
if (len > ip->ip_len)
{
udpstat.udps_badlen++;
Log3(("NAT: IP(id: %hd) has bad size\n", ip->ip_id));
}
m_adj(m, len - ip->ip_len);
ip->ip_len = len;
}
/*
* Save a copy of the IP header in case we want restore it
* for sending an ICMP error message in response.
*/
save_ip = *ip;
save_ip.ip_len+= iphlen; /* tcp_input subtracts this */
/*
* Checksum extended UDP header and data.
*/
if (udpcksum && uh->uh_sum)
{
memset(((struct ipovly *)ip)->ih_x1, 0, 9);
((struct ipovly *)ip)->ih_len = uh->uh_ulen;
#if 0
/* keep uh_sum for ICMP reply */
uh->uh_sum = cksum(m, len + sizeof (struct ip));
if (uh->uh_sum)
{
#endif
if (cksum(m, len + iphlen))
{
udpstat.udps_badsum++;
Log3(("NAT: IP(id: %hd) has bad (udp) cksum\n", ip->ip_id));
goto bad_free_mbuf;
}
}
#if 0
}
#endif
/*
* handle DHCP/BOOTP
*/
if (uh->uh_dport == RT_H2N_U16_C(BOOTP_SERVER))
{
bootp_input(pData, m);
goto done_free_mbuf;
}
LogFunc(("uh src: %RTnaipv4:%d, dst: %RTnaipv4:%d\n", ip->ip_src, RT_H2N_U16_C(uh->uh_sport), ip->ip_dst, RT_H2N_U16_C(uh->uh_dport)));
if ( pData->fUseHostResolver
&& uh->uh_dport == RT_H2N_U16_C(53)
&& CTL_CHECK(ip->ip_dst.s_addr, CTL_DNS))
{
struct sockaddr_in dst, src;
src.sin_addr.s_addr = ip->ip_dst.s_addr;
src.sin_port = uh->uh_dport;
dst.sin_addr.s_addr = ip->ip_src.s_addr;
dst.sin_port = uh->uh_sport;
slirpMbufTagService(pData, m, CTL_DNS);
/* udp_output2() expects a pointer to the body of UDP packet. */
m->m_data += sizeof(struct udpiphdr);
m->m_len -= sizeof(struct udpiphdr);
udp_output2(pData, NULL, m, &src, &dst, IPTOS_LOWDELAY);
LogFlowFuncLeave();
return;
}
/*
* handle TFTP
*/
if ( uh->uh_dport == RT_H2N_U16_C(TFTP_SERVER)
&& CTL_CHECK(ip->ip_dst.s_addr, CTL_TFTP))
{
if (pData->pvTftpSessions)
slirpTftpInput(pData, m);
goto done_free_mbuf;
}
/*
* Locate pcb for datagram.
*/
so = udp_last_so;
if ( so->so_lport != uh->uh_sport
|| so->so_laddr.s_addr != ip->ip_src.s_addr)
{
struct socket *tmp;
for (tmp = udb.so_next; tmp != &udb; tmp = tmp->so_next)
{
if ( tmp->so_lport == uh->uh_sport
&& tmp->so_laddr.s_addr == ip->ip_src.s_addr)
{
so = tmp;
break;
}
}
if (tmp == &udb)
so = NULL;
else
{
udpstat.udpps_pcbcachemiss++;
udp_last_so = so;
}
}
if (so == NULL)
{
/*
* If there's no socket for this packet,
* create one
*/
if ((so = socreate()) == NULL)
{
Log2(("NAT: IP(id: %hd) failed to create socket\n", ip->ip_id));
goto bad_free_mbuf;
}
if (udp_attach(pData, so) <= 0)
{
Log2(("NAT: IP(id: %hd) udp_attach errno = %d (%s)\n",
ip->ip_id, errno, strerror(errno)));
sofree(pData, so);
goto bad_free_mbuf;
}
/*
* Setup fields
*/
/* udp_last_so = so; */
so->so_laddr = ip->ip_src;
so->so_lport = uh->uh_sport;
so->so_iptos = ip->ip_tos;
/*
* XXXXX Here, check if it's in udpexec_list,
* and if it is, do the fork_exec() etc.
*/
}
so->so_faddr = ip->ip_dst; /* XXX */
so->so_fport = uh->uh_dport; /* XXX */
Assert(so->so_type == IPPROTO_UDP);
/*
* DNS proxy
*/
if ( pData->fUseDnsProxy
&& CTL_CHECK(ip->ip_dst.s_addr, CTL_DNS)
&& (uh->uh_dport == RT_H2N_U16_C(53)))
{
dnsproxy_query(pData, so, m, iphlen);
goto done_free_mbuf;
}
iphlen += sizeof(struct udphdr);
m->m_len -= iphlen;
m->m_data += iphlen;
ttl = ip->ip_ttl = save_ip.ip_ttl;
ret = setsockopt(so->s, IPPROTO_IP, IP_TTL, (const char*)&ttl, sizeof(ttl));
if (ret < 0)
LogRel(("NAT: Error (%s) occurred while setting TTL(%d) attribute "
"of IP packet to socket %R[natsock]\n", strerror(errno), ip->ip_ttl, so));
if ( sosendto(pData, so, m) == -1
&& ( !soIgnorableErrorCode(errno)
&& errno != ENOTCONN))
{
m->m_len += iphlen;
m->m_data -= iphlen;
*ip = save_ip;
Log2(("NAT: UDP tx errno = %d (%s) on sent to %RTnaipv4\n",
errno, strerror(errno), ip->ip_dst));
#if 0
/* ICMP_SOURCEQUENCH haven't got any effect, the idea here
* inform guest about the exosting NAT resources with assumption that
* that guest reduce traffic. But it doesn't work
*/
if( errno == EAGAIN
|| errno == EWOULDBLOCK
|| errno == EINPROGRESS
|| errno == ENOTCONN)
icmp_error(pData, m, ICMP_SOURCEQUENCH, 0, 1, strerror(errno));
else
#endif
icmp_error(pData, m, ICMP_UNREACH, ICMP_UNREACH_NET, 0, strerror(errno));
so->so_m = NULL;
LogFlowFuncLeave();
return;
}
if (so->so_m)
m_freem(pData, so->so_m); /* used for ICMP if error on sorecvfrom */
/* restore the orig mbuf packet */
m->m_len += iphlen;
m->m_data -= iphlen;
*ip = save_ip;
so->so_m = m; /* ICMP backup */
LogFlowFuncLeave();
return;
bad_free_mbuf:
Log2(("NAT: UDP(id: %hd) datagram to %RTnaipv4 with size(%d) claimed as bad\n",
ip->ip_id, &ip->ip_dst, ip->ip_len));
done_free_mbuf:
/* some services like bootp(built-in), dns(buildt-in) and dhcp don't need sockets
* and create new m'buffers to send them to guest, so we'll free their incomming
* buffers here.
*/
m_freem(pData, m);
LogFlowFuncLeave();
return;
}
