freebsd-dev/sys/netinet/tcp_reass.c

3176 lines
88 KiB
C
Raw Normal View History

1994-05-24 10:09:53 +00:00
/*
* Copyright (c) 1982, 1986, 1988, 1990, 1993, 1994, 1995
1994-05-24 10:09:53 +00:00
* The Regents of the University of California. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)tcp_input.c 8.12 (Berkeley) 5/24/95
1999-08-28 01:08:13 +00:00
* $FreeBSD$
1994-05-24 10:09:53 +00:00
*/
#include "opt_ipfw.h" /* for ipfw_fwd */
Initial import of RFC 2385 (TCP-MD5) digest support. This is the first of two commits; bringing in the kernel support first. This can be enabled by compiling a kernel with options TCP_SIGNATURE and FAST_IPSEC. For the uninitiated, this is a TCP option which provides for a means of authenticating TCP sessions which came into being before IPSEC. It is still relevant today, however, as it is used by many commercial router vendors, particularly with BGP, and as such has become a requirement for interconnect at many major Internet points of presence. Several parts of the TCP and IP headers, including the segment payload, are digested with MD5, including a shared secret. The PF_KEY interface is used to manage the secrets using security associations in the SADB. There is a limitation here in that as there is no way to map a TCP flow per-port back to an SPI without polluting tcpcb or using the SPD; the code to do the latter is unstable at this time. Therefore this code only supports per-host keying granularity. Whilst FAST_IPSEC is mutually exclusive with KAME IPSEC (and thus IPv6), TCP_SIGNATURE applies only to IPv4. For the vast majority of prospective users of this feature, this will not pose any problem. This implementation is output-only; that is, the option is honoured when responding to a host initiating a TCP session, but no effort is made [yet] to authenticate inbound traffic. This is, however, sufficient to interwork with Cisco equipment. Tested with a Cisco 2501 running IOS 12.0(27), and Quagga 0.96.4 with local patches. Patches for tcpdump to validate TCP-MD5 sessions are also available from me upon request. Sponsored by: sentex.net
2004-02-11 04:26:04 +00:00
#include "opt_inet.h"
#include "opt_inet6.h"
#include "opt_ipsec.h"
#include "opt_mac.h"
1997-09-16 18:36:06 +00:00
#include "opt_tcpdebug.h"
#include "opt_tcp_input.h"
#include "opt_tcp_sack.h"
1997-09-16 18:36:06 +00:00
1994-05-24 10:09:53 +00:00
#include <sys/param.h>
#include <sys/kernel.h>
#include <sys/mac.h>
1994-05-24 10:09:53 +00:00
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/proc.h> /* for proc0 declaration */
1994-05-24 10:09:53 +00:00
#include <sys/protosw.h>
#include <sys/signalvar.h>
1994-05-24 10:09:53 +00:00
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/sysctl.h>
#include <sys/syslog.h>
#include <sys/systm.h>
1994-05-24 10:09:53 +00:00
#include <machine/cpu.h> /* before tcp_seq.h, for tcp_random18() */
#include <vm/uma.h>
1994-05-24 10:09:53 +00:00
#include <net/if.h>
#include <net/route.h>
#include <netinet/in.h>
#include <netinet/in_pcb.h>
1994-05-24 10:09:53 +00:00
#include <netinet/in_systm.h>
#include <netinet/in_var.h>
1994-05-24 10:09:53 +00:00
#include <netinet/ip.h>
#include <netinet/ip_icmp.h> /* for ICMP_BANDLIM */
#include <netinet/icmp_var.h> /* for ICMP_BANDLIM */
1994-05-24 10:09:53 +00:00
#include <netinet/ip_var.h>
#include <netinet/ip6.h>
#include <netinet/icmp6.h>
#include <netinet6/in6_pcb.h>
#include <netinet6/ip6_var.h>
#include <netinet6/nd6.h>
1994-05-24 10:09:53 +00:00
#include <netinet/tcp.h>
#include <netinet/tcp_fsm.h>
#include <netinet/tcp_seq.h>
#include <netinet/tcp_timer.h>
#include <netinet/tcp_var.h>
#include <netinet6/tcp6_var.h>
1994-05-24 10:09:53 +00:00
#include <netinet/tcpip.h>
#ifdef TCPDEBUG
1994-05-24 10:09:53 +00:00
#include <netinet/tcp_debug.h>
#endif /* TCPDEBUG */
#ifdef FAST_IPSEC
#include <netipsec/ipsec.h>
#include <netipsec/ipsec6.h>
#endif /*FAST_IPSEC*/
#ifdef IPSEC
#include <netinet6/ipsec.h>
#include <netinet6/ipsec6.h>
#include <netkey/key.h>
#endif /*IPSEC*/
#include <machine/in_cksum.h>
static const int tcprexmtthresh = 3;
struct tcpstat tcpstat;
SYSCTL_STRUCT(_net_inet_tcp, TCPCTL_STATS, stats, CTLFLAG_RW,
&tcpstat , tcpstat, "TCP statistics (struct tcpstat, netinet/tcp_var.h)");
static int log_in_vain = 0;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, log_in_vain, CTLFLAG_RW,
&log_in_vain, 0, "Log all incoming TCP connections");
static int blackhole = 0;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, blackhole, CTLFLAG_RW,
&blackhole, 0, "Do not send RST when dropping refused connections");
int tcp_delack_enabled = 1;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, delayed_ack, CTLFLAG_RW,
&tcp_delack_enabled, 0,
"Delay ACK to try and piggyback it onto a data packet");
#ifdef TCP_DROP_SYNFIN
static int drop_synfin = 0;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, drop_synfin, CTLFLAG_RW,
&drop_synfin, 0, "Drop TCP packets with SYN+FIN set");
#endif
static int tcp_do_rfc3042 = 1;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, rfc3042, CTLFLAG_RW,
&tcp_do_rfc3042, 0, "Enable RFC 3042 (Limited Transmit)");
static int tcp_do_rfc3390 = 1;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, rfc3390, CTLFLAG_RW,
&tcp_do_rfc3390, 0,
"Enable RFC 3390 (Increasing TCP's Initial Congestion Window)");
SYSCTL_NODE(_net_inet_tcp, OID_AUTO, reass, CTLFLAG_RW, 0,
"TCP Segment Reassembly Queue");
static int tcp_reass_maxseg = 0;
SYSCTL_INT(_net_inet_tcp_reass, OID_AUTO, maxsegments, CTLFLAG_RDTUN,
&tcp_reass_maxseg, 0,
"Global maximum number of TCP Segments in Reassembly Queue");
int tcp_reass_qsize = 0;
SYSCTL_INT(_net_inet_tcp_reass, OID_AUTO, cursegments, CTLFLAG_RD,
&tcp_reass_qsize, 0,
"Global number of TCP Segments currently in Reassembly Queue");
static int tcp_reass_maxqlen = 48;
SYSCTL_INT(_net_inet_tcp_reass, OID_AUTO, maxqlen, CTLFLAG_RW,
&tcp_reass_maxqlen, 0,
"Maximum number of TCP Segments per individual Reassembly Queue");
static int tcp_reass_overflows = 0;
SYSCTL_INT(_net_inet_tcp_reass, OID_AUTO, overflows, CTLFLAG_RD,
&tcp_reass_overflows, 0,
"Global number of TCP Segment Reassembly Queue Overflows");
static int tcp_sack_recovery_initburst = 3;
SYSCTL_INT(_net_inet_tcp_sack, OID_AUTO,
initburst, CTLFLAG_RW,
&tcp_sack_recovery_initburst, 0,
"Initial Number of Rexmits when sack recovery is set up");
struct inpcbhead tcb;
#define tcb6 tcb /* for KAME src sync over BSD*'s */
struct inpcbinfo tcbinfo;
struct mtx *tcbinfo_mtx;
1994-05-24 10:09:53 +00:00
static void tcp_dooptions(struct tcpcb *, struct tcpopt *, u_char *,
int, int, struct tcphdr *);
2002-03-19 21:25:46 +00:00
static void tcp_pulloutofband(struct socket *,
struct tcphdr *, struct mbuf *, int);
static int tcp_reass(struct tcpcb *, struct tcphdr *, int *,
struct mbuf *);
2002-03-19 21:25:46 +00:00
static void tcp_xmit_timer(struct tcpcb *, int);
static void tcp_newreno_partial_ack(struct tcpcb *, struct tcphdr *);
static int tcp_timewait(struct tcptw *, struct tcpopt *,
struct tcphdr *, struct mbuf *, int);
/* Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. */
#ifdef INET6
#define ND6_HINT(tp) \
do { \
if ((tp) && (tp)->t_inpcb && \
((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0) \
nd6_nud_hint(NULL, NULL, 0); \
} while (0)
#else
#define ND6_HINT(tp)
#endif
1994-05-24 10:09:53 +00:00
/*
* Indicate whether this ack should be delayed. We can delay the ack if
2001-12-13 04:02:09 +00:00
* - there is no delayed ack timer in progress and
* - our last ack wasn't a 0-sized window. We never want to delay
* the ack that opens up a 0-sized window and
* - delayed acks are enabled or
* - this is a half-synchronized T/TCP connection.
*/
#define DELAY_ACK(tp) \
((!callout_active(tp->tt_delack) && \
(tp->t_flags & TF_RXWIN0SENT) == 0) && \
(tcp_delack_enabled || (tp->t_flags & TF_NEEDSYN)))
/* Initialize TCP reassembly queue */
uma_zone_t tcp_reass_zone;
void
tcp_reass_init()
{
tcp_reass_maxseg = nmbclusters / 16;
TUNABLE_INT_FETCH("net.inet.tcp.reass.maxsegments",
&tcp_reass_maxseg);
tcp_reass_zone = uma_zcreate("tcpreass", sizeof (struct tseg_qent),
NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, UMA_ZONE_NOFREE);
uma_zone_set_max(tcp_reass_zone, tcp_reass_maxseg);
}
static int
tcp_reass(tp, th, tlenp, m)
1994-05-24 10:09:53 +00:00
register struct tcpcb *tp;
register struct tcphdr *th;
int *tlenp;
1994-05-24 10:09:53 +00:00
struct mbuf *m;
{
struct tseg_qent *q;
struct tseg_qent *p = NULL;
struct tseg_qent *nq;
struct tseg_qent *te = NULL;
1994-05-24 10:09:53 +00:00
struct socket *so = tp->t_inpcb->inp_socket;
int flags;
/*
* XXX: tcp_reass() is rather inefficient with its data structures
* and should be rewritten (see NetBSD for optimizations). While
* doing that it should move to its own file tcp_reass.c.
*/
1994-05-24 10:09:53 +00:00
/*
* Call with th==0 after become established to
1994-05-24 10:09:53 +00:00
* force pre-ESTABLISHED data up to user socket.
*/
if (th == 0)
1994-05-24 10:09:53 +00:00
goto present;
/*
* Limit the number of segments in the reassembly queue to prevent
* holding on to too many segments (and thus running out of mbufs).
* Make sure to let the missing segment through which caused this
* queue. Always keep one global queue entry spare to be able to
* process the missing segment.
*/
if (th->th_seq != tp->rcv_nxt &&
(tcp_reass_qsize + 1 >= tcp_reass_maxseg ||
tp->t_segqlen >= tcp_reass_maxqlen)) {
tcp_reass_overflows++;
tcpstat.tcps_rcvmemdrop++;
m_freem(m);
return (0);
}
/*
* Allocate a new queue entry. If we can't, or hit the zone limit
* just drop the pkt.
*/
te = uma_zalloc(tcp_reass_zone, M_NOWAIT);
if (te == NULL) {
tcpstat.tcps_rcvmemdrop++;
m_freem(m);
return (0);
}
tp->t_segqlen++;
tcp_reass_qsize++;
1994-05-24 10:09:53 +00:00
/*
* Find a segment which begins after this one does.
*/
LIST_FOREACH(q, &tp->t_segq, tqe_q) {
if (SEQ_GT(q->tqe_th->th_seq, th->th_seq))
1994-05-24 10:09:53 +00:00
break;
p = q;
}
1994-05-24 10:09:53 +00:00
/*
* If there is a preceding segment, it may provide some of
* our data already. If so, drop the data from the incoming
* segment. If it provides all of our data, drop us.
*/
if (p != NULL) {
1994-05-24 10:09:53 +00:00
register int i;
/* conversion to int (in i) handles seq wraparound */
i = p->tqe_th->th_seq + p->tqe_len - th->th_seq;
1994-05-24 10:09:53 +00:00
if (i > 0) {
if (i >= *tlenp) {
1994-05-24 10:09:53 +00:00
tcpstat.tcps_rcvduppack++;
tcpstat.tcps_rcvdupbyte += *tlenp;
1994-05-24 10:09:53 +00:00
m_freem(m);
uma_zfree(tcp_reass_zone, te);
tp->t_segqlen--;
tcp_reass_qsize--;
/*
* Try to present any queued data
* at the left window edge to the user.
* This is needed after the 3-WHS
* completes.
*/
goto present; /* ??? */
1994-05-24 10:09:53 +00:00
}
m_adj(m, i);
*tlenp -= i;
th->th_seq += i;
1994-05-24 10:09:53 +00:00
}
}
tcpstat.tcps_rcvoopack++;
tcpstat.tcps_rcvoobyte += *tlenp;
1994-05-24 10:09:53 +00:00
/*
* While we overlap succeeding segments trim them or,
* if they are completely covered, dequeue them.
*/
while (q) {
register int i = (th->th_seq + *tlenp) - q->tqe_th->th_seq;
1994-05-24 10:09:53 +00:00
if (i <= 0)
break;
if (i < q->tqe_len) {
q->tqe_th->th_seq += i;
q->tqe_len -= i;
m_adj(q->tqe_m, i);
1994-05-24 10:09:53 +00:00
break;
}
nq = LIST_NEXT(q, tqe_q);
LIST_REMOVE(q, tqe_q);
m_freem(q->tqe_m);
uma_zfree(tcp_reass_zone, q);
tp->t_segqlen--;
tcp_reass_qsize--;
q = nq;
1994-05-24 10:09:53 +00:00
}
/* Insert the new segment queue entry into place. */
te->tqe_m = m;
te->tqe_th = th;
te->tqe_len = *tlenp;
if (p == NULL) {
LIST_INSERT_HEAD(&tp->t_segq, te, tqe_q);
} else {
LIST_INSERT_AFTER(p, te, tqe_q);
}
1994-05-24 10:09:53 +00:00
present:
/*
* Present data to user, advancing rcv_nxt through
* completed sequence space.
*/
if (!TCPS_HAVEESTABLISHED(tp->t_state))
1994-05-24 10:09:53 +00:00
return (0);
q = LIST_FIRST(&tp->t_segq);
if (!q || q->tqe_th->th_seq != tp->rcv_nxt)
1994-05-24 10:09:53 +00:00
return (0);
SOCKBUF_LOCK(&so->so_rcv);
1994-05-24 10:09:53 +00:00
do {
tp->rcv_nxt += q->tqe_len;
flags = q->tqe_th->th_flags & TH_FIN;
nq = LIST_NEXT(q, tqe_q);
LIST_REMOVE(q, tqe_q);
/* Unlocked read. */
if (so->so_rcv.sb_state & SBS_CANTRCVMORE)
m_freem(q->tqe_m);
else
sbappendstream_locked(&so->so_rcv, q->tqe_m);
uma_zfree(tcp_reass_zone, q);
tp->t_segqlen--;
tcp_reass_qsize--;
q = nq;
} while (q && q->tqe_th->th_seq == tp->rcv_nxt);
ND6_HINT(tp);
sorwakeup_locked(so);
1994-05-24 10:09:53 +00:00
return (flags);
}
/*
* TCP input routine, follows pages 65-76 of the
* protocol specification dated September, 1981 very closely.
*/
#ifdef INET6
int
tcp6_input(mp, offp, proto)
struct mbuf **mp;
int *offp, proto;
{
register struct mbuf *m = *mp;
struct in6_ifaddr *ia6;
IP6_EXTHDR_CHECK(m, *offp, sizeof(struct tcphdr), IPPROTO_DONE);
/*
* draft-itojun-ipv6-tcp-to-anycast
* better place to put this in?
*/
ia6 = ip6_getdstifaddr(m);
if (ia6 && (ia6->ia6_flags & IN6_IFF_ANYCAST)) {
struct ip6_hdr *ip6;
ip6 = mtod(m, struct ip6_hdr *);
icmp6_error(m, ICMP6_DST_UNREACH, ICMP6_DST_UNREACH_ADDR,
(caddr_t)&ip6->ip6_dst - (caddr_t)ip6);
return IPPROTO_DONE;
}
tcp_input(m, *offp);
return IPPROTO_DONE;
}
#endif
1994-05-24 10:09:53 +00:00
void
tcp_input(m, off0)
1994-05-24 10:09:53 +00:00
register struct mbuf *m;
int off0;
1994-05-24 10:09:53 +00:00
{
register struct tcphdr *th;
register struct ip *ip = NULL;
register struct ipovly *ipov;
register struct inpcb *inp = NULL;
u_char *optp = NULL;
int optlen = 0;
int len, tlen, off;
int drop_hdrlen;
1994-05-24 10:09:53 +00:00
register struct tcpcb *tp = 0;
register int thflags;
struct socket *so = 0;
1994-05-24 10:09:53 +00:00
int todrop, acked, ourfinisacked, needoutput = 0;
u_long tiwin;
struct tcpopt to; /* options in this segment */
int headlocked = 0;
Convert ipfw to use PFIL_HOOKS. This is change is transparent to userland and preserves the ipfw ABI. The ipfw core packet inspection and filtering functions have not been changed, only how ipfw is invoked is different. However there are many changes how ipfw is and its add-on's are handled: In general ipfw is now called through the PFIL_HOOKS and most associated magic, that was in ip_input() or ip_output() previously, is now done in ipfw_check_[in|out]() in the ipfw PFIL handler. IPDIVERT is entirely handled within the ipfw PFIL handlers. A packet to be diverted is checked if it is fragmented, if yes, ip_reass() gets in for reassembly. If not, or all fragments arrived and the packet is complete, divert_packet is called directly. For 'tee' no reassembly attempt is made and a copy of the packet is sent to the divert socket unmodified. The original packet continues its way through ip_input/output(). ipfw 'forward' is done via m_tag's. The ipfw PFIL handlers tag the packet with the new destination sockaddr_in. A check if the new destination is a local IP address is made and the m_flags are set appropriately. ip_input() and ip_output() have some more work to do here. For ip_input() the m_flags are checked and a packet for us is directly sent to the 'ours' section for further processing. Destination changes on the input path are only tagged and the 'srcrt' flag to ip_forward() is set to disable destination checks and ICMP replies at this stage. The tag is going to be handled on output. ip_output() again checks for m_flags and the 'ours' tag. If found, the packet will be dropped back to the IP netisr where it is going to be picked up by ip_input() again and the directly sent to the 'ours' section. When only the destination changes, the route's 'dst' is overwritten with the new destination from the forward m_tag. Then it jumps back at the route lookup again and skips the firewall check because it has been marked with M_SKIP_FIREWALL. ipfw 'forward' has to be compiled into the kernel with 'option IPFIREWALL_FORWARD' to enable it. DUMMYNET is entirely handled within the ipfw PFIL handlers. A packet for a dummynet pipe or queue is directly sent to dummynet_io(). Dummynet will then inject it back into ip_input/ip_output() after it has served its time. Dummynet packets are tagged and will continue from the next rule when they hit the ipfw PFIL handlers again after re-injection. BRIDGING and IPFW_ETHER are not changed yet and use ipfw_chk() directly as they did before. Later this will be changed to dedicated ETHER PFIL_HOOKS. More detailed changes to the code: conf/files Add netinet/ip_fw_pfil.c. conf/options Add IPFIREWALL_FORWARD option. modules/ipfw/Makefile Add ip_fw_pfil.c. net/bridge.c Disable PFIL_HOOKS if ipfw for bridging is active. Bridging ipfw is still directly invoked to handle layer2 headers and packets would get a double ipfw when run through PFIL_HOOKS as well. netinet/ip_divert.c Removed divert_clone() function. It is no longer used. netinet/ip_dummynet.[ch] Neither the route 'ro' nor the destination 'dst' need to be stored while in dummynet transit. Structure members and associated macros are removed. netinet/ip_fastfwd.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_fw.h Removed 'ro' and 'dst' from struct ip_fw_args. netinet/ip_fw2.c (Re)moved some global variables and the module handling. netinet/ip_fw_pfil.c New file containing the ipfw PFIL handlers and module initialization. netinet/ip_input.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. ip_forward() does not longer require the 'next_hop' struct sockaddr_in argument. Disable early checks if 'srcrt' is set. netinet/ip_output.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_var.h Add ip_reass() as general function. (Used from ipfw PFIL handlers for IPDIVERT.) netinet/raw_ip.c Directly check if ipfw and dummynet control pointers are active. netinet/tcp_input.c Rework the 'ipfw forward' to local code to work with the new way of forward tags. netinet/tcp_sack.c Remove include 'opt_ipfw.h' which is not needed here. sys/mbuf.h Remove m_claim_next() macro which was exclusively for ipfw 'forward' and is no longer needed. Approved by: re (scottl)
2004-08-17 22:05:54 +00:00
#ifdef IPFIREWALL_FORWARD
struct m_tag *fwd_tag;
#endif
int rstreason; /* For badport_bandlim accounting purposes */
struct ip6_hdr *ip6 = NULL;
#ifdef INET6
int isipv6;
#else
const int isipv6 = 0;
#endif
#ifdef TCPDEBUG
/*
* The size of tcp_saveipgen must be the size of the max ip header,
* now IPv6.
*/
u_char tcp_saveipgen[40];
struct tcphdr tcp_savetcp;
short ostate = 0;
#endif
#ifdef INET6
isipv6 = (mtod(m, struct ip *)->ip_v == 6) ? 1 : 0;
#endif
bzero((char *)&to, sizeof(to));
1994-05-24 10:09:53 +00:00
tcpstat.tcps_rcvtotal++;
if (isipv6) {
#ifdef INET6
/* IP6_EXTHDR_CHECK() is already done at tcp6_input() */
ip6 = mtod(m, struct ip6_hdr *);
tlen = sizeof(*ip6) + ntohs(ip6->ip6_plen) - off0;
if (in6_cksum(m, IPPROTO_TCP, off0, tlen)) {
tcpstat.tcps_rcvbadsum++;
goto drop;
}
th = (struct tcphdr *)((caddr_t)ip6 + off0);
/*
* Be proactive about unspecified IPv6 address in source.
* As we use all-zero to indicate unbounded/unconnected pcb,
* unspecified IPv6 address can be used to confuse us.
*
* Note that packets with unspecified IPv6 destination is
* already dropped in ip6_input.
*/
if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
/* XXX stat */
goto drop;
}
#else
th = NULL; /* XXX: avoid compiler warning */
#endif
} else {
/*
* Get IP and TCP header together in first mbuf.
* Note: IP leaves IP header in first mbuf.
*/
if (off0 > sizeof (struct ip)) {
ip_stripoptions(m, (struct mbuf *)0);
off0 = sizeof(struct ip);
}
if (m->m_len < sizeof (struct tcpiphdr)) {
if ((m = m_pullup(m, sizeof (struct tcpiphdr))) == 0) {
tcpstat.tcps_rcvshort++;
return;
}
}
ip = mtod(m, struct ip *);
ipov = (struct ipovly *)ip;
th = (struct tcphdr *)((caddr_t)ip + off0);
tlen = ip->ip_len;
if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) {
if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR)
th->th_sum = m->m_pkthdr.csum_data;
else
th->th_sum = in_pseudo(ip->ip_src.s_addr,
ip->ip_dst.s_addr,
htonl(m->m_pkthdr.csum_data +
ip->ip_len +
IPPROTO_TCP));
th->th_sum ^= 0xffff;
#ifdef TCPDEBUG
ipov->ih_len = (u_short)tlen;
ipov->ih_len = htons(ipov->ih_len);
#endif
} else {
/*
* Checksum extended TCP header and data.
*/
len = sizeof (struct ip) + tlen;
bzero(ipov->ih_x1, sizeof(ipov->ih_x1));
ipov->ih_len = (u_short)tlen;
ipov->ih_len = htons(ipov->ih_len);
th->th_sum = in_cksum(m, len);
}
if (th->th_sum) {
tcpstat.tcps_rcvbadsum++;
goto drop;
}
#ifdef INET6
/* Re-initialization for later version check */
ip->ip_v = IPVERSION;
#endif
}
1994-05-24 10:09:53 +00:00
/*
* Check that TCP offset makes sense,
* pull out TCP options and adjust length. XXX
*/
off = th->th_off << 2;
1994-05-24 10:09:53 +00:00
if (off < sizeof (struct tcphdr) || off > tlen) {
tcpstat.tcps_rcvbadoff++;
goto drop;
}
tlen -= off; /* tlen is used instead of ti->ti_len */
1994-05-24 10:09:53 +00:00
if (off > sizeof (struct tcphdr)) {
if (isipv6) {
#ifdef INET6
IP6_EXTHDR_CHECK(m, off0, off, );
ip6 = mtod(m, struct ip6_hdr *);
th = (struct tcphdr *)((caddr_t)ip6 + off0);
#endif
} else {
if (m->m_len < sizeof(struct ip) + off) {
if ((m = m_pullup(m, sizeof (struct ip) + off))
== 0) {
tcpstat.tcps_rcvshort++;
return;
}
ip = mtod(m, struct ip *);
ipov = (struct ipovly *)ip;
th = (struct tcphdr *)((caddr_t)ip + off0);
1994-05-24 10:09:53 +00:00
}
}
optlen = off - sizeof (struct tcphdr);
optp = (u_char *)(th + 1);
1994-05-24 10:09:53 +00:00
}
thflags = th->th_flags;
1994-05-24 10:09:53 +00:00
#ifdef TCP_DROP_SYNFIN
/*
* If the drop_synfin option is enabled, drop all packets with
* both the SYN and FIN bits set. This prevents e.g. nmap from
* identifying the TCP/IP stack.
*
2001-01-24 16:25:36 +00:00
* This is a violation of the TCP specification.
*/
if (drop_synfin && (thflags & (TH_SYN|TH_FIN)) == (TH_SYN|TH_FIN))
goto drop;
#endif
1994-05-24 10:09:53 +00:00
/*
* Convert TCP protocol specific fields to host format.
*/
th->th_seq = ntohl(th->th_seq);
th->th_ack = ntohl(th->th_ack);
th->th_win = ntohs(th->th_win);
th->th_urp = ntohs(th->th_urp);
1994-05-24 10:09:53 +00:00
/*
* Delay dropping TCP, IP headers, IPv6 ext headers, and TCP options,
* until after ip6_savecontrol() is called and before other functions
* which don't want those proto headers.
* Because ip6_savecontrol() is going to parse the mbuf to
* search for data to be passed up to user-land, it wants mbuf
* parameters to be unchanged.
* XXX: the call of ip6_savecontrol() has been obsoleted based on
* latest version of the advanced API (20020110).
*/
drop_hdrlen = off0 + off;
1994-05-24 10:09:53 +00:00
/*
* Locate pcb for segment.
*/
INP_INFO_WLOCK(&tcbinfo);
headlocked = 1;
1994-05-24 10:09:53 +00:00
findpcb:
Convert ipfw to use PFIL_HOOKS. This is change is transparent to userland and preserves the ipfw ABI. The ipfw core packet inspection and filtering functions have not been changed, only how ipfw is invoked is different. However there are many changes how ipfw is and its add-on's are handled: In general ipfw is now called through the PFIL_HOOKS and most associated magic, that was in ip_input() or ip_output() previously, is now done in ipfw_check_[in|out]() in the ipfw PFIL handler. IPDIVERT is entirely handled within the ipfw PFIL handlers. A packet to be diverted is checked if it is fragmented, if yes, ip_reass() gets in for reassembly. If not, or all fragments arrived and the packet is complete, divert_packet is called directly. For 'tee' no reassembly attempt is made and a copy of the packet is sent to the divert socket unmodified. The original packet continues its way through ip_input/output(). ipfw 'forward' is done via m_tag's. The ipfw PFIL handlers tag the packet with the new destination sockaddr_in. A check if the new destination is a local IP address is made and the m_flags are set appropriately. ip_input() and ip_output() have some more work to do here. For ip_input() the m_flags are checked and a packet for us is directly sent to the 'ours' section for further processing. Destination changes on the input path are only tagged and the 'srcrt' flag to ip_forward() is set to disable destination checks and ICMP replies at this stage. The tag is going to be handled on output. ip_output() again checks for m_flags and the 'ours' tag. If found, the packet will be dropped back to the IP netisr where it is going to be picked up by ip_input() again and the directly sent to the 'ours' section. When only the destination changes, the route's 'dst' is overwritten with the new destination from the forward m_tag. Then it jumps back at the route lookup again and skips the firewall check because it has been marked with M_SKIP_FIREWALL. ipfw 'forward' has to be compiled into the kernel with 'option IPFIREWALL_FORWARD' to enable it. DUMMYNET is entirely handled within the ipfw PFIL handlers. A packet for a dummynet pipe or queue is directly sent to dummynet_io(). Dummynet will then inject it back into ip_input/ip_output() after it has served its time. Dummynet packets are tagged and will continue from the next rule when they hit the ipfw PFIL handlers again after re-injection. BRIDGING and IPFW_ETHER are not changed yet and use ipfw_chk() directly as they did before. Later this will be changed to dedicated ETHER PFIL_HOOKS. More detailed changes to the code: conf/files Add netinet/ip_fw_pfil.c. conf/options Add IPFIREWALL_FORWARD option. modules/ipfw/Makefile Add ip_fw_pfil.c. net/bridge.c Disable PFIL_HOOKS if ipfw for bridging is active. Bridging ipfw is still directly invoked to handle layer2 headers and packets would get a double ipfw when run through PFIL_HOOKS as well. netinet/ip_divert.c Removed divert_clone() function. It is no longer used. netinet/ip_dummynet.[ch] Neither the route 'ro' nor the destination 'dst' need to be stored while in dummynet transit. Structure members and associated macros are removed. netinet/ip_fastfwd.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_fw.h Removed 'ro' and 'dst' from struct ip_fw_args. netinet/ip_fw2.c (Re)moved some global variables and the module handling. netinet/ip_fw_pfil.c New file containing the ipfw PFIL handlers and module initialization. netinet/ip_input.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. ip_forward() does not longer require the 'next_hop' struct sockaddr_in argument. Disable early checks if 'srcrt' is set. netinet/ip_output.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_var.h Add ip_reass() as general function. (Used from ipfw PFIL handlers for IPDIVERT.) netinet/raw_ip.c Directly check if ipfw and dummynet control pointers are active. netinet/tcp_input.c Rework the 'ipfw forward' to local code to work with the new way of forward tags. netinet/tcp_sack.c Remove include 'opt_ipfw.h' which is not needed here. sys/mbuf.h Remove m_claim_next() macro which was exclusively for ipfw 'forward' and is no longer needed. Approved by: re (scottl)
2004-08-17 22:05:54 +00:00
#ifdef IPFIREWALL_FORWARD
/* Grab info from PACKET_TAG_IPFORWARD tag prepended to the chain. */
fwd_tag = m_tag_find(m, PACKET_TAG_IPFORWARD, NULL);
if (fwd_tag != NULL && isipv6 == 0) { /* IPv6 support is not yet */
struct sockaddr_in *next_hop;
next_hop = (struct sockaddr_in *)(fwd_tag+1);
/*
Remove (almost all) global variables that were used to hold packet forwarding state ("annotations") during ip processing. The code is considerably cleaner now. The variables removed by this change are: ip_divert_cookie used by divert sockets ip_fw_fwd_addr used for transparent ip redirection last_pkt used by dynamic pipes in dummynet Removal of the first two has been done by carrying the annotations into volatile structs prepended to the mbuf chains, and adding appropriate code to add/remove annotations in the routines which make use of them, i.e. ip_input(), ip_output(), tcp_input(), bdg_forward(), ether_demux(), ether_output_frame(), div_output(). On passing, remove a bug in divert handling of fragmented packet. Now it is the fragment at offset 0 which sets the divert status of the whole packet, whereas formerly it was the last incoming fragment to decide. Removal of last_pkt required a change in the interface of ip_fw_chk() and dummynet_io(). On passing, use the same mechanism for dummynet annotations and for divert/forward annotations. option IPFIREWALL_FORWARD is effectively useless, the code to implement it is very small and is now in by default to avoid the obfuscation of conditionally compiled code. NOTES: * there is at least one global variable left, sro_fwd, in ip_output(). I am not sure if/how this can be removed. * I have deliberately avoided gratuitous style changes in this commit to avoid cluttering the diffs. Minor stule cleanup will likely be necessary * this commit only focused on the IP layer. I am sure there is a number of global variables used in the TCP and maybe UDP stack. * despite the number of files touched, there are absolutely no API's or data structures changed by this commit (except the interfaces of ip_fw_chk() and dummynet_io(), which are internal anyways), so an MFC is quite safe and unintrusive (and desirable, given the improved readability of the code). MFC after: 10 days
2002-06-22 11:51:02 +00:00
* Transparently forwarded. Pretend to be the destination.
* already got one like this?
*/
Convert ipfw to use PFIL_HOOKS. This is change is transparent to userland and preserves the ipfw ABI. The ipfw core packet inspection and filtering functions have not been changed, only how ipfw is invoked is different. However there are many changes how ipfw is and its add-on's are handled: In general ipfw is now called through the PFIL_HOOKS and most associated magic, that was in ip_input() or ip_output() previously, is now done in ipfw_check_[in|out]() in the ipfw PFIL handler. IPDIVERT is entirely handled within the ipfw PFIL handlers. A packet to be diverted is checked if it is fragmented, if yes, ip_reass() gets in for reassembly. If not, or all fragments arrived and the packet is complete, divert_packet is called directly. For 'tee' no reassembly attempt is made and a copy of the packet is sent to the divert socket unmodified. The original packet continues its way through ip_input/output(). ipfw 'forward' is done via m_tag's. The ipfw PFIL handlers tag the packet with the new destination sockaddr_in. A check if the new destination is a local IP address is made and the m_flags are set appropriately. ip_input() and ip_output() have some more work to do here. For ip_input() the m_flags are checked and a packet for us is directly sent to the 'ours' section for further processing. Destination changes on the input path are only tagged and the 'srcrt' flag to ip_forward() is set to disable destination checks and ICMP replies at this stage. The tag is going to be handled on output. ip_output() again checks for m_flags and the 'ours' tag. If found, the packet will be dropped back to the IP netisr where it is going to be picked up by ip_input() again and the directly sent to the 'ours' section. When only the destination changes, the route's 'dst' is overwritten with the new destination from the forward m_tag. Then it jumps back at the route lookup again and skips the firewall check because it has been marked with M_SKIP_FIREWALL. ipfw 'forward' has to be compiled into the kernel with 'option IPFIREWALL_FORWARD' to enable it. DUMMYNET is entirely handled within the ipfw PFIL handlers. A packet for a dummynet pipe or queue is directly sent to dummynet_io(). Dummynet will then inject it back into ip_input/ip_output() after it has served its time. Dummynet packets are tagged and will continue from the next rule when they hit the ipfw PFIL handlers again after re-injection. BRIDGING and IPFW_ETHER are not changed yet and use ipfw_chk() directly as they did before. Later this will be changed to dedicated ETHER PFIL_HOOKS. More detailed changes to the code: conf/files Add netinet/ip_fw_pfil.c. conf/options Add IPFIREWALL_FORWARD option. modules/ipfw/Makefile Add ip_fw_pfil.c. net/bridge.c Disable PFIL_HOOKS if ipfw for bridging is active. Bridging ipfw is still directly invoked to handle layer2 headers and packets would get a double ipfw when run through PFIL_HOOKS as well. netinet/ip_divert.c Removed divert_clone() function. It is no longer used. netinet/ip_dummynet.[ch] Neither the route 'ro' nor the destination 'dst' need to be stored while in dummynet transit. Structure members and associated macros are removed. netinet/ip_fastfwd.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_fw.h Removed 'ro' and 'dst' from struct ip_fw_args. netinet/ip_fw2.c (Re)moved some global variables and the module handling. netinet/ip_fw_pfil.c New file containing the ipfw PFIL handlers and module initialization. netinet/ip_input.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. ip_forward() does not longer require the 'next_hop' struct sockaddr_in argument. Disable early checks if 'srcrt' is set. netinet/ip_output.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_var.h Add ip_reass() as general function. (Used from ipfw PFIL handlers for IPDIVERT.) netinet/raw_ip.c Directly check if ipfw and dummynet control pointers are active. netinet/tcp_input.c Rework the 'ipfw forward' to local code to work with the new way of forward tags. netinet/tcp_sack.c Remove include 'opt_ipfw.h' which is not needed here. sys/mbuf.h Remove m_claim_next() macro which was exclusively for ipfw 'forward' and is no longer needed. Approved by: re (scottl)
2004-08-17 22:05:54 +00:00
inp = in_pcblookup_hash(&tcbinfo,
ip->ip_src, th->th_sport,
ip->ip_dst, th->th_dport,
0, m->m_pkthdr.rcvif);
if (!inp) {
Convert ipfw to use PFIL_HOOKS. This is change is transparent to userland and preserves the ipfw ABI. The ipfw core packet inspection and filtering functions have not been changed, only how ipfw is invoked is different. However there are many changes how ipfw is and its add-on's are handled: In general ipfw is now called through the PFIL_HOOKS and most associated magic, that was in ip_input() or ip_output() previously, is now done in ipfw_check_[in|out]() in the ipfw PFIL handler. IPDIVERT is entirely handled within the ipfw PFIL handlers. A packet to be diverted is checked if it is fragmented, if yes, ip_reass() gets in for reassembly. If not, or all fragments arrived and the packet is complete, divert_packet is called directly. For 'tee' no reassembly attempt is made and a copy of the packet is sent to the divert socket unmodified. The original packet continues its way through ip_input/output(). ipfw 'forward' is done via m_tag's. The ipfw PFIL handlers tag the packet with the new destination sockaddr_in. A check if the new destination is a local IP address is made and the m_flags are set appropriately. ip_input() and ip_output() have some more work to do here. For ip_input() the m_flags are checked and a packet for us is directly sent to the 'ours' section for further processing. Destination changes on the input path are only tagged and the 'srcrt' flag to ip_forward() is set to disable destination checks and ICMP replies at this stage. The tag is going to be handled on output. ip_output() again checks for m_flags and the 'ours' tag. If found, the packet will be dropped back to the IP netisr where it is going to be picked up by ip_input() again and the directly sent to the 'ours' section. When only the destination changes, the route's 'dst' is overwritten with the new destination from the forward m_tag. Then it jumps back at the route lookup again and skips the firewall check because it has been marked with M_SKIP_FIREWALL. ipfw 'forward' has to be compiled into the kernel with 'option IPFIREWALL_FORWARD' to enable it. DUMMYNET is entirely handled within the ipfw PFIL handlers. A packet for a dummynet pipe or queue is directly sent to dummynet_io(). Dummynet will then inject it back into ip_input/ip_output() after it has served its time. Dummynet packets are tagged and will continue from the next rule when they hit the ipfw PFIL handlers again after re-injection. BRIDGING and IPFW_ETHER are not changed yet and use ipfw_chk() directly as they did before. Later this will be changed to dedicated ETHER PFIL_HOOKS. More detailed changes to the code: conf/files Add netinet/ip_fw_pfil.c. conf/options Add IPFIREWALL_FORWARD option. modules/ipfw/Makefile Add ip_fw_pfil.c. net/bridge.c Disable PFIL_HOOKS if ipfw for bridging is active. Bridging ipfw is still directly invoked to handle layer2 headers and packets would get a double ipfw when run through PFIL_HOOKS as well. netinet/ip_divert.c Removed divert_clone() function. It is no longer used. netinet/ip_dummynet.[ch] Neither the route 'ro' nor the destination 'dst' need to be stored while in dummynet transit. Structure members and associated macros are removed. netinet/ip_fastfwd.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_fw.h Removed 'ro' and 'dst' from struct ip_fw_args. netinet/ip_fw2.c (Re)moved some global variables and the module handling. netinet/ip_fw_pfil.c New file containing the ipfw PFIL handlers and module initialization. netinet/ip_input.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. ip_forward() does not longer require the 'next_hop' struct sockaddr_in argument. Disable early checks if 'srcrt' is set. netinet/ip_output.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_var.h Add ip_reass() as general function. (Used from ipfw PFIL handlers for IPDIVERT.) netinet/raw_ip.c Directly check if ipfw and dummynet control pointers are active. netinet/tcp_input.c Rework the 'ipfw forward' to local code to work with the new way of forward tags. netinet/tcp_sack.c Remove include 'opt_ipfw.h' which is not needed here. sys/mbuf.h Remove m_claim_next() macro which was exclusively for ipfw 'forward' and is no longer needed. Approved by: re (scottl)
2004-08-17 22:05:54 +00:00
/* It's new. Try to find the ambushing socket. */
inp = in_pcblookup_hash(&tcbinfo,
ip->ip_src, th->th_sport,
next_hop->sin_addr,
next_hop->sin_port ?
ntohs(next_hop->sin_port) :
th->th_dport,
1, m->m_pkthdr.rcvif);
}
Convert ipfw to use PFIL_HOOKS. This is change is transparent to userland and preserves the ipfw ABI. The ipfw core packet inspection and filtering functions have not been changed, only how ipfw is invoked is different. However there are many changes how ipfw is and its add-on's are handled: In general ipfw is now called through the PFIL_HOOKS and most associated magic, that was in ip_input() or ip_output() previously, is now done in ipfw_check_[in|out]() in the ipfw PFIL handler. IPDIVERT is entirely handled within the ipfw PFIL handlers. A packet to be diverted is checked if it is fragmented, if yes, ip_reass() gets in for reassembly. If not, or all fragments arrived and the packet is complete, divert_packet is called directly. For 'tee' no reassembly attempt is made and a copy of the packet is sent to the divert socket unmodified. The original packet continues its way through ip_input/output(). ipfw 'forward' is done via m_tag's. The ipfw PFIL handlers tag the packet with the new destination sockaddr_in. A check if the new destination is a local IP address is made and the m_flags are set appropriately. ip_input() and ip_output() have some more work to do here. For ip_input() the m_flags are checked and a packet for us is directly sent to the 'ours' section for further processing. Destination changes on the input path are only tagged and the 'srcrt' flag to ip_forward() is set to disable destination checks and ICMP replies at this stage. The tag is going to be handled on output. ip_output() again checks for m_flags and the 'ours' tag. If found, the packet will be dropped back to the IP netisr where it is going to be picked up by ip_input() again and the directly sent to the 'ours' section. When only the destination changes, the route's 'dst' is overwritten with the new destination from the forward m_tag. Then it jumps back at the route lookup again and skips the firewall check because it has been marked with M_SKIP_FIREWALL. ipfw 'forward' has to be compiled into the kernel with 'option IPFIREWALL_FORWARD' to enable it. DUMMYNET is entirely handled within the ipfw PFIL handlers. A packet for a dummynet pipe or queue is directly sent to dummynet_io(). Dummynet will then inject it back into ip_input/ip_output() after it has served its time. Dummynet packets are tagged and will continue from the next rule when they hit the ipfw PFIL handlers again after re-injection. BRIDGING and IPFW_ETHER are not changed yet and use ipfw_chk() directly as they did before. Later this will be changed to dedicated ETHER PFIL_HOOKS. More detailed changes to the code: conf/files Add netinet/ip_fw_pfil.c. conf/options Add IPFIREWALL_FORWARD option. modules/ipfw/Makefile Add ip_fw_pfil.c. net/bridge.c Disable PFIL_HOOKS if ipfw for bridging is active. Bridging ipfw is still directly invoked to handle layer2 headers and packets would get a double ipfw when run through PFIL_HOOKS as well. netinet/ip_divert.c Removed divert_clone() function. It is no longer used. netinet/ip_dummynet.[ch] Neither the route 'ro' nor the destination 'dst' need to be stored while in dummynet transit. Structure members and associated macros are removed. netinet/ip_fastfwd.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_fw.h Removed 'ro' and 'dst' from struct ip_fw_args. netinet/ip_fw2.c (Re)moved some global variables and the module handling. netinet/ip_fw_pfil.c New file containing the ipfw PFIL handlers and module initialization. netinet/ip_input.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. ip_forward() does not longer require the 'next_hop' struct sockaddr_in argument. Disable early checks if 'srcrt' is set. netinet/ip_output.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_var.h Add ip_reass() as general function. (Used from ipfw PFIL handlers for IPDIVERT.) netinet/raw_ip.c Directly check if ipfw and dummynet control pointers are active. netinet/tcp_input.c Rework the 'ipfw forward' to local code to work with the new way of forward tags. netinet/tcp_sack.c Remove include 'opt_ipfw.h' which is not needed here. sys/mbuf.h Remove m_claim_next() macro which was exclusively for ipfw 'forward' and is no longer needed. Approved by: re (scottl)
2004-08-17 22:05:54 +00:00
/* Remove the tag from the packet. We don't need it anymore. */
m_tag_delete(m, fwd_tag);
} else {
Convert ipfw to use PFIL_HOOKS. This is change is transparent to userland and preserves the ipfw ABI. The ipfw core packet inspection and filtering functions have not been changed, only how ipfw is invoked is different. However there are many changes how ipfw is and its add-on's are handled: In general ipfw is now called through the PFIL_HOOKS and most associated magic, that was in ip_input() or ip_output() previously, is now done in ipfw_check_[in|out]() in the ipfw PFIL handler. IPDIVERT is entirely handled within the ipfw PFIL handlers. A packet to be diverted is checked if it is fragmented, if yes, ip_reass() gets in for reassembly. If not, or all fragments arrived and the packet is complete, divert_packet is called directly. For 'tee' no reassembly attempt is made and a copy of the packet is sent to the divert socket unmodified. The original packet continues its way through ip_input/output(). ipfw 'forward' is done via m_tag's. The ipfw PFIL handlers tag the packet with the new destination sockaddr_in. A check if the new destination is a local IP address is made and the m_flags are set appropriately. ip_input() and ip_output() have some more work to do here. For ip_input() the m_flags are checked and a packet for us is directly sent to the 'ours' section for further processing. Destination changes on the input path are only tagged and the 'srcrt' flag to ip_forward() is set to disable destination checks and ICMP replies at this stage. The tag is going to be handled on output. ip_output() again checks for m_flags and the 'ours' tag. If found, the packet will be dropped back to the IP netisr where it is going to be picked up by ip_input() again and the directly sent to the 'ours' section. When only the destination changes, the route's 'dst' is overwritten with the new destination from the forward m_tag. Then it jumps back at the route lookup again and skips the firewall check because it has been marked with M_SKIP_FIREWALL. ipfw 'forward' has to be compiled into the kernel with 'option IPFIREWALL_FORWARD' to enable it. DUMMYNET is entirely handled within the ipfw PFIL handlers. A packet for a dummynet pipe or queue is directly sent to dummynet_io(). Dummynet will then inject it back into ip_input/ip_output() after it has served its time. Dummynet packets are tagged and will continue from the next rule when they hit the ipfw PFIL handlers again after re-injection. BRIDGING and IPFW_ETHER are not changed yet and use ipfw_chk() directly as they did before. Later this will be changed to dedicated ETHER PFIL_HOOKS. More detailed changes to the code: conf/files Add netinet/ip_fw_pfil.c. conf/options Add IPFIREWALL_FORWARD option. modules/ipfw/Makefile Add ip_fw_pfil.c. net/bridge.c Disable PFIL_HOOKS if ipfw for bridging is active. Bridging ipfw is still directly invoked to handle layer2 headers and packets would get a double ipfw when run through PFIL_HOOKS as well. netinet/ip_divert.c Removed divert_clone() function. It is no longer used. netinet/ip_dummynet.[ch] Neither the route 'ro' nor the destination 'dst' need to be stored while in dummynet transit. Structure members and associated macros are removed. netinet/ip_fastfwd.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_fw.h Removed 'ro' and 'dst' from struct ip_fw_args. netinet/ip_fw2.c (Re)moved some global variables and the module handling. netinet/ip_fw_pfil.c New file containing the ipfw PFIL handlers and module initialization. netinet/ip_input.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. ip_forward() does not longer require the 'next_hop' struct sockaddr_in argument. Disable early checks if 'srcrt' is set. netinet/ip_output.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_var.h Add ip_reass() as general function. (Used from ipfw PFIL handlers for IPDIVERT.) netinet/raw_ip.c Directly check if ipfw and dummynet control pointers are active. netinet/tcp_input.c Rework the 'ipfw forward' to local code to work with the new way of forward tags. netinet/tcp_sack.c Remove include 'opt_ipfw.h' which is not needed here. sys/mbuf.h Remove m_claim_next() macro which was exclusively for ipfw 'forward' and is no longer needed. Approved by: re (scottl)
2004-08-17 22:05:54 +00:00
#endif /* IPFIREWALL_FORWARD */
if (isipv6) {
#ifdef INET6
inp = in6_pcblookup_hash(&tcbinfo,
&ip6->ip6_src, th->th_sport,
&ip6->ip6_dst, th->th_dport,
1, m->m_pkthdr.rcvif);
#endif
} else
inp = in_pcblookup_hash(&tcbinfo,
ip->ip_src, th->th_sport,
ip->ip_dst, th->th_dport,
1, m->m_pkthdr.rcvif);
Convert ipfw to use PFIL_HOOKS. This is change is transparent to userland and preserves the ipfw ABI. The ipfw core packet inspection and filtering functions have not been changed, only how ipfw is invoked is different. However there are many changes how ipfw is and its add-on's are handled: In general ipfw is now called through the PFIL_HOOKS and most associated magic, that was in ip_input() or ip_output() previously, is now done in ipfw_check_[in|out]() in the ipfw PFIL handler. IPDIVERT is entirely handled within the ipfw PFIL handlers. A packet to be diverted is checked if it is fragmented, if yes, ip_reass() gets in for reassembly. If not, or all fragments arrived and the packet is complete, divert_packet is called directly. For 'tee' no reassembly attempt is made and a copy of the packet is sent to the divert socket unmodified. The original packet continues its way through ip_input/output(). ipfw 'forward' is done via m_tag's. The ipfw PFIL handlers tag the packet with the new destination sockaddr_in. A check if the new destination is a local IP address is made and the m_flags are set appropriately. ip_input() and ip_output() have some more work to do here. For ip_input() the m_flags are checked and a packet for us is directly sent to the 'ours' section for further processing. Destination changes on the input path are only tagged and the 'srcrt' flag to ip_forward() is set to disable destination checks and ICMP replies at this stage. The tag is going to be handled on output. ip_output() again checks for m_flags and the 'ours' tag. If found, the packet will be dropped back to the IP netisr where it is going to be picked up by ip_input() again and the directly sent to the 'ours' section. When only the destination changes, the route's 'dst' is overwritten with the new destination from the forward m_tag. Then it jumps back at the route lookup again and skips the firewall check because it has been marked with M_SKIP_FIREWALL. ipfw 'forward' has to be compiled into the kernel with 'option IPFIREWALL_FORWARD' to enable it. DUMMYNET is entirely handled within the ipfw PFIL handlers. A packet for a dummynet pipe or queue is directly sent to dummynet_io(). Dummynet will then inject it back into ip_input/ip_output() after it has served its time. Dummynet packets are tagged and will continue from the next rule when they hit the ipfw PFIL handlers again after re-injection. BRIDGING and IPFW_ETHER are not changed yet and use ipfw_chk() directly as they did before. Later this will be changed to dedicated ETHER PFIL_HOOKS. More detailed changes to the code: conf/files Add netinet/ip_fw_pfil.c. conf/options Add IPFIREWALL_FORWARD option. modules/ipfw/Makefile Add ip_fw_pfil.c. net/bridge.c Disable PFIL_HOOKS if ipfw for bridging is active. Bridging ipfw is still directly invoked to handle layer2 headers and packets would get a double ipfw when run through PFIL_HOOKS as well. netinet/ip_divert.c Removed divert_clone() function. It is no longer used. netinet/ip_dummynet.[ch] Neither the route 'ro' nor the destination 'dst' need to be stored while in dummynet transit. Structure members and associated macros are removed. netinet/ip_fastfwd.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_fw.h Removed 'ro' and 'dst' from struct ip_fw_args. netinet/ip_fw2.c (Re)moved some global variables and the module handling. netinet/ip_fw_pfil.c New file containing the ipfw PFIL handlers and module initialization. netinet/ip_input.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. ip_forward() does not longer require the 'next_hop' struct sockaddr_in argument. Disable early checks if 'srcrt' is set. netinet/ip_output.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_var.h Add ip_reass() as general function. (Used from ipfw PFIL handlers for IPDIVERT.) netinet/raw_ip.c Directly check if ipfw and dummynet control pointers are active. netinet/tcp_input.c Rework the 'ipfw forward' to local code to work with the new way of forward tags. netinet/tcp_sack.c Remove include 'opt_ipfw.h' which is not needed here. sys/mbuf.h Remove m_claim_next() macro which was exclusively for ipfw 'forward' and is no longer needed. Approved by: re (scottl)
2004-08-17 22:05:54 +00:00
#ifdef IPFIREWALL_FORWARD
}
Convert ipfw to use PFIL_HOOKS. This is change is transparent to userland and preserves the ipfw ABI. The ipfw core packet inspection and filtering functions have not been changed, only how ipfw is invoked is different. However there are many changes how ipfw is and its add-on's are handled: In general ipfw is now called through the PFIL_HOOKS and most associated magic, that was in ip_input() or ip_output() previously, is now done in ipfw_check_[in|out]() in the ipfw PFIL handler. IPDIVERT is entirely handled within the ipfw PFIL handlers. A packet to be diverted is checked if it is fragmented, if yes, ip_reass() gets in for reassembly. If not, or all fragments arrived and the packet is complete, divert_packet is called directly. For 'tee' no reassembly attempt is made and a copy of the packet is sent to the divert socket unmodified. The original packet continues its way through ip_input/output(). ipfw 'forward' is done via m_tag's. The ipfw PFIL handlers tag the packet with the new destination sockaddr_in. A check if the new destination is a local IP address is made and the m_flags are set appropriately. ip_input() and ip_output() have some more work to do here. For ip_input() the m_flags are checked and a packet for us is directly sent to the 'ours' section for further processing. Destination changes on the input path are only tagged and the 'srcrt' flag to ip_forward() is set to disable destination checks and ICMP replies at this stage. The tag is going to be handled on output. ip_output() again checks for m_flags and the 'ours' tag. If found, the packet will be dropped back to the IP netisr where it is going to be picked up by ip_input() again and the directly sent to the 'ours' section. When only the destination changes, the route's 'dst' is overwritten with the new destination from the forward m_tag. Then it jumps back at the route lookup again and skips the firewall check because it has been marked with M_SKIP_FIREWALL. ipfw 'forward' has to be compiled into the kernel with 'option IPFIREWALL_FORWARD' to enable it. DUMMYNET is entirely handled within the ipfw PFIL handlers. A packet for a dummynet pipe or queue is directly sent to dummynet_io(). Dummynet will then inject it back into ip_input/ip_output() after it has served its time. Dummynet packets are tagged and will continue from the next rule when they hit the ipfw PFIL handlers again after re-injection. BRIDGING and IPFW_ETHER are not changed yet and use ipfw_chk() directly as they did before. Later this will be changed to dedicated ETHER PFIL_HOOKS. More detailed changes to the code: conf/files Add netinet/ip_fw_pfil.c. conf/options Add IPFIREWALL_FORWARD option. modules/ipfw/Makefile Add ip_fw_pfil.c. net/bridge.c Disable PFIL_HOOKS if ipfw for bridging is active. Bridging ipfw is still directly invoked to handle layer2 headers and packets would get a double ipfw when run through PFIL_HOOKS as well. netinet/ip_divert.c Removed divert_clone() function. It is no longer used. netinet/ip_dummynet.[ch] Neither the route 'ro' nor the destination 'dst' need to be stored while in dummynet transit. Structure members and associated macros are removed. netinet/ip_fastfwd.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_fw.h Removed 'ro' and 'dst' from struct ip_fw_args. netinet/ip_fw2.c (Re)moved some global variables and the module handling. netinet/ip_fw_pfil.c New file containing the ipfw PFIL handlers and module initialization. netinet/ip_input.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. ip_forward() does not longer require the 'next_hop' struct sockaddr_in argument. Disable early checks if 'srcrt' is set. netinet/ip_output.c Removed all direct ipfw handling code and replace it with the new 'ipfw forward' handling code. netinet/ip_var.h Add ip_reass() as general function. (Used from ipfw PFIL handlers for IPDIVERT.) netinet/raw_ip.c Directly check if ipfw and dummynet control pointers are active. netinet/tcp_input.c Rework the 'ipfw forward' to local code to work with the new way of forward tags. netinet/tcp_sack.c Remove include 'opt_ipfw.h' which is not needed here. sys/mbuf.h Remove m_claim_next() macro which was exclusively for ipfw 'forward' and is no longer needed. Approved by: re (scottl)
2004-08-17 22:05:54 +00:00
#endif /* IPFIREWALL_FORWARD */
#if defined(IPSEC) || defined(FAST_IPSEC)
#ifdef INET6
if (isipv6) {
if (inp != NULL && ipsec6_in_reject(m, inp)) {
#ifdef IPSEC
ipsec6stat.in_polvio++;
#endif
goto drop;
}
} else
#endif /* INET6 */
if (inp != NULL && ipsec4_in_reject(m, inp)) {
#ifdef IPSEC
ipsecstat.in_polvio++;
#endif
goto drop;
}
#endif /*IPSEC || FAST_IPSEC*/
1994-05-24 10:09:53 +00:00
/*
* 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 == NULL) {
if (log_in_vain) {
#ifdef INET6
char dbuf[INET6_ADDRSTRLEN+2], sbuf[INET6_ADDRSTRLEN+2];
#else
char dbuf[4*sizeof "123"], sbuf[4*sizeof "123"];
#endif
if (isipv6) {
#ifdef INET6
strcpy(dbuf, "[");
strcpy(sbuf, "[");
strcat(dbuf, ip6_sprintf(&ip6->ip6_dst));
strcat(sbuf, ip6_sprintf(&ip6->ip6_src));
strcat(dbuf, "]");
strcat(sbuf, "]");
#endif
} else {
strcpy(dbuf, inet_ntoa(ip->ip_dst));
strcpy(sbuf, inet_ntoa(ip->ip_src));
}
switch (log_in_vain) {
case 1:
if ((thflags & TH_SYN) == 0)
break;
/* FALLTHROUGH */
case 2:
log(LOG_INFO,
"Connection attempt to TCP %s:%d "
"from %s:%d flags:0x%02x\n",
dbuf, ntohs(th->th_dport), sbuf,
ntohs(th->th_sport), thflags);
break;
default:
break;
}
}
if (blackhole) {
switch (blackhole) {
case 1:
if (thflags & TH_SYN)
goto drop;
break;
case 2:
goto drop;
default:
goto drop;
}
}
rstreason = BANDLIM_RST_CLOSEDPORT;
goto dropwithreset;
}
INP_LOCK(inp);
if (inp->inp_vflag & INP_TIMEWAIT) {
/*
* The only option of relevance is TOF_CC, and only if
* present in a SYN segment. See tcp_timewait().
*/
if (thflags & TH_SYN)
tcp_dooptions((struct tcpcb *)NULL, &to, optp, optlen, 1, th);
if (tcp_timewait((struct tcptw *)inp->inp_ppcb,
&to, th, m, tlen))
goto findpcb;
/*
* tcp_timewait unlocks inp.
*/
INP_INFO_WUNLOCK(&tcbinfo);
return;
}
1994-05-24 10:09:53 +00:00
tp = intotcpcb(inp);
if (tp == 0) {
INP_UNLOCK(inp);
rstreason = BANDLIM_RST_CLOSEDPORT;
goto dropwithreset;
}
1994-05-24 10:09:53 +00:00
if (tp->t_state == TCPS_CLOSED)
goto drop;
1995-05-30 08:16:23 +00:00
1994-05-24 10:09:53 +00:00
/* Unscale the window into a 32-bit value. */
if ((thflags & TH_SYN) == 0)
tiwin = th->th_win << tp->snd_scale;
1994-05-24 10:09:53 +00:00
else
tiwin = th->th_win;
#ifdef MAC
INP_LOCK_ASSERT(inp);
if (mac_check_inpcb_deliver(inp, m))
goto drop;
#endif
so = inp->inp_socket;
#ifdef TCPDEBUG
if (so->so_options & SO_DEBUG) {
ostate = tp->t_state;
if (isipv6)
bcopy((char *)ip6, (char *)tcp_saveipgen, sizeof(*ip6));
else
bcopy((char *)ip, (char *)tcp_saveipgen, sizeof(*ip));
tcp_savetcp = *th;
}
#endif
if (so->so_options & SO_ACCEPTCONN) {
struct in_conninfo inc;
#ifdef INET6
inc.inc_isipv6 = isipv6;
#endif
if (isipv6) {
inc.inc6_faddr = ip6->ip6_src;
inc.inc6_laddr = ip6->ip6_dst;
} else {
inc.inc_faddr = ip->ip_src;
inc.inc_laddr = ip->ip_dst;
}
inc.inc_fport = th->th_sport;
inc.inc_lport = th->th_dport;
/*
* If the state is LISTEN then ignore segment if it contains
* a 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.
*
* If the state is SYN_RECEIVED (syncache) and seg contains
* an ACK, but not for our SYN/ACK, send a RST. If the seg
* contains a RST, check the sequence number to see if it
* is a valid reset segment.
*/
if ((thflags & (TH_RST|TH_ACK|TH_SYN)) != TH_SYN) {
if ((thflags & (TH_RST|TH_ACK|TH_SYN)) == TH_ACK) {
if (!syncache_expand(&inc, th, &so, m)) {
/*
* No syncache entry, or ACK was not
* for our SYN/ACK. Send a RST.
*/
tcpstat.tcps_badsyn++;
rstreason = BANDLIM_RST_OPENPORT;
goto dropwithreset;
}
if (so == NULL) {
/*
* Could not complete 3-way handshake,
* connection is being closed down, and
* syncache will free mbuf.
*/
INP_UNLOCK(inp);
INP_INFO_WUNLOCK(&tcbinfo);
return;
}
/*
* Socket is created in state SYN_RECEIVED.
* Continue processing segment.
*/
INP_UNLOCK(inp);
inp = sotoinpcb(so);
INP_LOCK(inp);
tp = intotcpcb(inp);
/*
* This is what would have happened in
* tcp_output() when the SYN,ACK was sent.
*/
tp->snd_up = tp->snd_una;
tp->snd_max = tp->snd_nxt = tp->iss + 1;
tp->last_ack_sent = tp->rcv_nxt;
2003-02-19 21:33:46 +00:00
/*
* RFC1323: The window in SYN & SYN/ACK
* segments is never scaled.
*/
tp->snd_wnd = tiwin; /* unscaled */
goto after_listen;
}
if (thflags & TH_RST) {
syncache_chkrst(&inc, th);
goto drop;
}
if (thflags & TH_ACK) {
syncache_badack(&inc);
tcpstat.tcps_badsyn++;
rstreason = BANDLIM_RST_OPENPORT;
goto dropwithreset;
}
goto drop;
}
/*
* Segment's flags are (SYN) or (SYN|FIN).
*/
#ifdef INET6
/*
* If deprecated address is forbidden,
* we do not accept SYN to deprecated interface
* address to prevent any new inbound connection from
* getting established.
* When we do not accept SYN, we send a TCP RST,
* with deprecated source address (instead of dropping
* it). We compromise it as it is much better for peer
* to send a RST, and RST will be the final packet
* for the exchange.
*
* If we do not forbid deprecated addresses, we accept
* the SYN packet. RFC2462 does not suggest dropping
* SYN in this case.
* If we decipher RFC2462 5.5.4, it says like this:
* 1. use of deprecated addr with existing
* communication is okay - "SHOULD continue to be
* used"
* 2. use of it with new communication:
* (2a) "SHOULD NOT be used if alternate address
* with sufficient scope is available"
* (2b) nothing mentioned otherwise.
* Here we fall into (2b) case as we have no choice in
* our source address selection - we must obey the peer.
*
* The wording in RFC2462 is confusing, and there are
* multiple description text for deprecated address
* handling - worse, they are not exactly the same.
* I believe 5.5.4 is the best one, so we follow 5.5.4.
*/
if (isipv6 && !ip6_use_deprecated) {
struct in6_ifaddr *ia6;
if ((ia6 = ip6_getdstifaddr(m)) &&
(ia6->ia6_flags & IN6_IFF_DEPRECATED)) {
INP_UNLOCK(inp);
tp = NULL;
rstreason = BANDLIM_RST_OPENPORT;
goto dropwithreset;
}
}
#endif
/*
* If it is from this socket, drop it, it must be forged.
* Don't bother responding if the destination was a broadcast.
*/
if (th->th_dport == th->th_sport) {
if (isipv6) {
if (IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst,
&ip6->ip6_src))
goto drop;
} else {
if (ip->ip_dst.s_addr == ip->ip_src.s_addr)
goto drop;
}
}
/*
* RFC1122 4.2.3.10, p. 104: discard bcast/mcast SYN
*
* Note that it is quite possible to receive unicast
* link-layer packets with a broadcast IP address. Use
* in_broadcast() to find them.
*/
if (m->m_flags & (M_BCAST|M_MCAST))
goto drop;
if (isipv6) {
if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
IN6_IS_ADDR_MULTICAST(&ip6->ip6_src))
goto drop;
} else {
if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
IN_MULTICAST(ntohl(ip->ip_src.s_addr)) ||
ip->ip_src.s_addr == htonl(INADDR_BROADCAST) ||
in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif))
goto drop;
}
/*
* SYN appears to be valid; create compressed TCP state
* for syncache, or perform t/tcp connection.
*/
if (so->so_qlen <= so->so_qlimit) {
#ifdef TCPDEBUG
if (so->so_options & SO_DEBUG)
tcp_trace(TA_INPUT, ostate, tp,
(void *)tcp_saveipgen, &tcp_savetcp, 0);
#endif
tcp_dooptions(tp, &to, optp, optlen, 1, th);
if (!syncache_add(&inc, &to, th, &so, m))
goto drop;
if (so == NULL) {
Improved connection establishment performance by doing local port lookups via a hashed port list. In the new scheme, in_pcblookup() goes away and is replaced by a new routine, in_pcblookup_local() for doing the local port check. Note that this implementation is space inefficient in that the PCB struct is now too large to fit into 128 bytes. I might deal with this in the future by using the new zone allocator, but I wanted these changes to be extensively tested in their current form first. Also: 1) Fixed off-by-one errors in the port lookup loops in in_pcbbind(). 2) Got rid of some unneeded rehashing. Adding a new routine, in_pcbinshash() to do the initialial hash insertion. 3) Renamed in_pcblookuphash() to in_pcblookup_hash() for easier readability. 4) Added a new routine, in_pcbremlists() to remove the PCB from the various hash lists. 5) Added/deleted comments where appropriate. 6) Removed unnecessary splnet() locking. In general, the PCB functions should be called at splnet()...there are unfortunately a few exceptions, however. 7) Reorganized a few structs for better cache line behavior. 8) Killed my TCP_ACK_HACK kludge. It may come back in a different form in the future, however. These changes have been tested on wcarchive for more than a month. In tests done here, connection establishment overhead is reduced by more than 50 times, thus getting rid of one of the major networking scalability problems. Still to do: make tcp_fastimo/tcp_slowtimo scale well for systems with a large number of connections. tcp_fastimo is easy; tcp_slowtimo is difficult. WARNING: Anything that knows about inpcb and tcpcb structs will have to be recompiled; at the very least, this includes netstat(1).
1998-01-27 09:15:13 +00:00
/*
* Entry added to syncache, mbuf used to
* send SYN,ACK packet.
Improved connection establishment performance by doing local port lookups via a hashed port list. In the new scheme, in_pcblookup() goes away and is replaced by a new routine, in_pcblookup_local() for doing the local port check. Note that this implementation is space inefficient in that the PCB struct is now too large to fit into 128 bytes. I might deal with this in the future by using the new zone allocator, but I wanted these changes to be extensively tested in their current form first. Also: 1) Fixed off-by-one errors in the port lookup loops in in_pcbbind(). 2) Got rid of some unneeded rehashing. Adding a new routine, in_pcbinshash() to do the initialial hash insertion. 3) Renamed in_pcblookuphash() to in_pcblookup_hash() for easier readability. 4) Added a new routine, in_pcbremlists() to remove the PCB from the various hash lists. 5) Added/deleted comments where appropriate. 6) Removed unnecessary splnet() locking. In general, the PCB functions should be called at splnet()...there are unfortunately a few exceptions, however. 7) Reorganized a few structs for better cache line behavior. 8) Killed my TCP_ACK_HACK kludge. It may come back in a different form in the future, however. These changes have been tested on wcarchive for more than a month. In tests done here, connection establishment overhead is reduced by more than 50 times, thus getting rid of one of the major networking scalability problems. Still to do: make tcp_fastimo/tcp_slowtimo scale well for systems with a large number of connections. tcp_fastimo is easy; tcp_slowtimo is difficult. WARNING: Anything that knows about inpcb and tcpcb structs will have to be recompiled; at the very least, this includes netstat(1).
1998-01-27 09:15:13 +00:00
*/
KASSERT(headlocked, ("headlocked"));
INP_UNLOCK(inp);
INP_INFO_WUNLOCK(&tcbinfo);
return;
}
/*
* Segment passed TAO tests.
*/
INP_UNLOCK(inp);
inp = sotoinpcb(so);
INP_LOCK(inp);
tp = intotcpcb(inp);
tp->snd_wnd = tiwin;
tp->t_starttime = ticks;
tp->t_state = TCPS_ESTABLISHED;
/*
* T/TCP logic:
* If there is a FIN or if there is data, then
* delay SYN,ACK(SYN) in the hope of piggy-backing
* it on a response segment. Otherwise must send
* ACK now in case the other side is slow starting.
*/
if (thflags & TH_FIN || tlen != 0)
tp->t_flags |= (TF_DELACK | TF_NEEDSYN);
else
tp->t_flags |= (TF_ACKNOW | TF_NEEDSYN);
tcpstat.tcps_connects++;
soisconnected(so);
goto trimthenstep6;
1994-05-24 10:09:53 +00:00
}
goto drop;
}
after_listen:
KASSERT(headlocked, ("tcp_input(): after_listen head is not locked"));
INP_LOCK_ASSERT(inp);
/* XXX temp debugging */
/* should not happen - syncache should pick up these connections */
if (tp->t_state == TCPS_LISTEN)
panic("tcp_input: TCPS_LISTEN");
1994-05-24 10:09:53 +00:00
Limiters and sanity checks for TCP MSS (maximum segement size) resource exhaustion attacks. For network link optimization TCP can adjust its MSS and thus packet size according to the observed path MTU. This is done dynamically based on feedback from the remote host and network components along the packet path. This information can be abused to pretend an extremely low path MTU. The resource exhaustion works in two ways: o during tcp connection setup the advertized local MSS is exchanged between the endpoints. The remote endpoint can set this arbitrarily low (except for a minimum MTU of 64 octets enforced in the BSD code). When the local host is sending data it is forced to send many small IP packets instead of a large one. For example instead of the normal TCP payload size of 1448 it forces TCP payload size of 12 (MTU 64) and thus we have a 120 times increase in workload and packets. On fast links this quickly saturates the local CPU and may also hit pps processing limites of network components along the path. This type of attack is particularly effective for servers where the attacker can download large files (WWW and FTP). We mitigate it by enforcing a minimum MTU settable by sysctl net.inet.tcp.minmss defaulting to 256 octets. o the local host is reveiving data on a TCP connection from the remote host. The local host has no control over the packet size the remote host is sending. The remote host may chose to do what is described in the first attack and send the data in packets with an TCP payload of at least one byte. For each packet the tcp_input() function will be entered, the packet is processed and a sowakeup() is signalled to the connected process. For example an attack with 2 Mbit/s gives 4716 packets per second and the same amount of sowakeup()s to the process (and context switches). This type of attack is particularly effective for servers where the attacker can upload large amounts of data. Normally this is the case with WWW server where large POSTs can be made. We mitigate this by calculating the average MSS payload per second. If it goes below 'net.inet.tcp.minmss' and the pps rate is above 'net.inet.tcp.minmssoverload' defaulting to 1000 this particular TCP connection is resetted and dropped. MITRE CVE: CAN-2004-0002 Reviewed by: sam (mentor) MFC after: 1 day
2004-01-08 17:40:07 +00:00
/*
* This is the second part of the MSS DoS prevention code (after
* minmss on the sending side) and it deals with too many too small
* tcp packets in a too short timeframe (1 second).
*
* For every full second we count the number of received packets
* and bytes. If we get a lot of packets per second for this connection
* (tcp_minmssoverload) we take a closer look at it and compute the
* average packet size for the past second. If that is less than
* tcp_minmss we get too many packets with very small payload which
* is not good and burdens our system (and every packet generates
* a wakeup to the process connected to our socket). We can reasonable
* expect this to be small packet DoS attack to exhaust our CPU
* cycles.
*
* Care has to be taken for the minimum packet overload value. This
* value defines the minimum number of packets per second before we
* start to worry. This must not be too low to avoid killing for
* example interactive connections with many small packets like
* telnet or SSH.
*
* Setting either tcp_minmssoverload or tcp_minmss to "0" disables
* this check.
*
* Account for packet if payload packet, skip over ACK, etc.
*/
if (tcp_minmss && tcp_minmssoverload &&
tp->t_state == TCPS_ESTABLISHED && tlen > 0) {
if (tp->rcv_second > ticks) {
tp->rcv_pps++;
tp->rcv_byps += tlen + off;
if (tp->rcv_pps > tcp_minmssoverload) {
if ((tp->rcv_byps / tp->rcv_pps) < tcp_minmss) {
printf("too many small tcp packets from "
"%s:%u, av. %lubyte/packet, "
"dropping connection\n",
#ifdef INET6
isipv6 ?
ip6_sprintf(&inp->inp_inc.inc6_faddr) :
#endif
inet_ntoa(inp->inp_inc.inc_faddr),
inp->inp_inc.inc_fport,
tp->rcv_byps / tp->rcv_pps);
tp = tcp_drop(tp, ECONNRESET);
tcpstat.tcps_minmssdrops++;
goto drop;
}
}
} else {
tp->rcv_second = ticks + hz;
tp->rcv_pps = 1;
tp->rcv_byps = tlen + off;
}
}
1994-05-24 10:09:53 +00:00
/*
* Segment received on connection.
* Reset idle time and keep-alive timer.
*/
tp->t_rcvtime = ticks;
if (TCPS_HAVEESTABLISHED(tp->t_state))
callout_reset(tp->tt_keep, tcp_keepidle, tcp_timer_keep, tp);
1994-05-24 10:09:53 +00:00
/*
* Process options only when we get SYN/ACK back. The SYN case
* for incoming connections is handled in tcp_syncache.
* XXX this is traditional behavior, may need to be cleaned up.
1994-05-24 10:09:53 +00:00
*/
tcp_dooptions(tp, &to, optp, optlen, thflags & TH_SYN, th);
if (thflags & TH_SYN) {
if (to.to_flags & TOF_SCALE) {
tp->t_flags |= TF_RCVD_SCALE;
tp->requested_s_scale = to.to_requested_s_scale;
}
if (to.to_flags & TOF_TS) {
tp->t_flags |= TF_RCVD_TSTMP;
tp->ts_recent = to.to_tsval;
tp->ts_recent_age = ticks;
}
if (to.to_flags & TOF_MSS)
tcp_mss(tp, to.to_mss);
if (tp->sack_enable) {
if (!(to.to_flags & TOF_SACK))
tp->sack_enable = 0;
else
tp->t_flags |= TF_SACK_PERMIT;
}
}
if (tp->sack_enable) {
/* Delete stale (cumulatively acked) SACK holes */
tcp_del_sackholes(tp, th);
tp->rcv_laststart = th->th_seq; /* last recv'd segment*/
tp->rcv_lastend = th->th_seq + tlen;
}
1994-05-24 10:09:53 +00:00
1995-05-30 08:16:23 +00:00
/*
1994-05-24 10:09:53 +00:00
* 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.
* Make sure that the hidden state-flags are also off.
* Since we check for TCPS_ESTABLISHED above, it can only
* be TH_NEEDSYN.
1994-05-24 10:09:53 +00:00
*/
if (tp->t_state == TCPS_ESTABLISHED &&
(thflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ACK)) == TH_ACK &&
((tp->t_flags & (TF_NEEDSYN|TF_NEEDFIN)) == 0) &&
((to.to_flags & TOF_TS) == 0 ||
TSTMP_GEQ(to.to_tsval, tp->ts_recent)) &&
th->th_seq == tp->rcv_nxt && tiwin && tiwin == tp->snd_wnd &&
tp->snd_nxt == tp->snd_max) {
1994-05-24 10:09:53 +00:00
1995-05-30 08:16:23 +00:00
/*
1994-05-24 10:09:53 +00:00
* If last ACK falls within this segment's sequence numbers,
* record the timestamp.
* NOTE that the test is modified according to the latest
* proposal of the tcplw@cray.com list (Braden 1993/04/26).
1994-05-24 10:09:53 +00:00
*/
if ((to.to_flags & TOF_TS) != 0 &&
SEQ_LEQ(th->th_seq, tp->last_ack_sent)) {
tp->ts_recent_age = ticks;
tp->ts_recent = to.to_tsval;
1994-05-24 10:09:53 +00:00
}
if (tlen == 0) {
if (SEQ_GT(th->th_ack, tp->snd_una) &&
SEQ_LEQ(th->th_ack, tp->snd_max) &&
tp->snd_cwnd >= tp->snd_wnd &&
((!tcp_do_newreno && !tp->sack_enable &&
tp->t_dupacks < tcprexmtthresh) ||
((tcp_do_newreno || tp->sack_enable) &&
!IN_FASTRECOVERY(tp)))) {
KASSERT(headlocked, ("headlocked"));
INP_INFO_WUNLOCK(&tcbinfo);
1994-05-24 10:09:53 +00:00
/*
* this is a pure ack for outstanding data.
*/
++tcpstat.tcps_predack;
/*
* "bad retransmit" recovery
*/
if (tp->t_rxtshift == 1 &&
ticks < tp->t_badrxtwin) {
++tcpstat.tcps_sndrexmitbad;
tp->snd_cwnd = tp->snd_cwnd_prev;
tp->snd_ssthresh =
tp->snd_ssthresh_prev;
tp->snd_recover = tp->snd_recover_prev;
if (tp->t_flags & TF_WASFRECOVERY)
ENTER_FASTRECOVERY(tp);
tp->snd_nxt = tp->snd_max;
tp->t_badrxtwin = 0;
}
/*
* Recalculate the transmit timer / rtt.
*
* Some boxes send broken timestamp replies
* during the SYN+ACK phase, ignore
* timestamps of 0 or we could calculate a
* huge RTT and blow up the retransmit timer.
*/
if ((to.to_flags & TOF_TS) != 0 &&
to.to_tsecr) {
tcp_xmit_timer(tp,
ticks - to.to_tsecr + 1);
} else if (tp->t_rtttime &&
SEQ_GT(th->th_ack, tp->t_rtseq)) {
tcp_xmit_timer(tp,
ticks - tp->t_rtttime);
}
tcp_xmit_bandwidth_limit(tp, th->th_ack);
acked = th->th_ack - tp->snd_una;
1994-05-24 10:09:53 +00:00
tcpstat.tcps_rcvackpack++;
tcpstat.tcps_rcvackbyte += acked;
sbdrop(&so->so_snd, acked);
if (SEQ_GT(tp->snd_una, tp->snd_recover) &&
SEQ_LEQ(th->th_ack, tp->snd_recover))
tp->snd_recover = th->th_ack - 1;
tp->snd_una = th->th_ack;
/*
* pull snd_wl2 up to prevent seq wrap relative
* to th_ack.
*/
tp->snd_wl2 = th->th_ack;
tp->t_dupacks = 0;
1994-05-24 10:09:53 +00:00
m_freem(m);
ND6_HINT(tp); /* some progress has been done */
1994-05-24 10:09:53 +00:00
/*
* 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.
#ifdef TCPDEBUG
if (so->so_options & SO_DEBUG)
tcp_trace(TA_INPUT, ostate, tp,
(void *)tcp_saveipgen,
&tcp_savetcp, 0);
#endif
1994-05-24 10:09:53 +00:00
*/
if (tp->snd_una == tp->snd_max)
callout_stop(tp->tt_rexmt);
else if (!callout_active(tp->tt_persist))
callout_reset(tp->tt_rexmt,
tp->t_rxtcur,
tcp_timer_rexmt, tp);
1994-05-24 10:09:53 +00:00
sowwakeup(so);
1994-05-24 10:09:53 +00:00
if (so->so_snd.sb_cc)
(void) tcp_output(tp);
goto check_delack;
1994-05-24 10:09:53 +00:00
}
} else if (th->th_ack == tp->snd_una &&
LIST_EMPTY(&tp->t_segq) &&
tlen <= sbspace(&so->so_rcv)) {
KASSERT(headlocked, ("headlocked"));
INP_INFO_WUNLOCK(&tcbinfo);
1994-05-24 10:09:53 +00:00
/*
* this is a pure, in-sequence data packet
* with nothing on the reassembly queue and
* we have enough buffer space to take it.
*/
/* Clean receiver SACK report if present */
if (tp->sack_enable && tp->rcv_numsacks)
tcp_clean_sackreport(tp);
1994-05-24 10:09:53 +00:00
++tcpstat.tcps_preddat;
tp->rcv_nxt += tlen;
/*
* Pull snd_wl1 up to prevent seq wrap relative to
* th_seq.
*/
tp->snd_wl1 = th->th_seq;
/*
* Pull rcv_up up to prevent seq wrap relative to
* rcv_nxt.
*/
tp->rcv_up = tp->rcv_nxt;
1994-05-24 10:09:53 +00:00
tcpstat.tcps_rcvpack++;
tcpstat.tcps_rcvbyte += tlen;
ND6_HINT(tp); /* some progress has been done */
1994-05-24 10:09:53 +00:00
/*
#ifdef TCPDEBUG
if (so->so_options & SO_DEBUG)
tcp_trace(TA_INPUT, ostate, tp,
(void *)tcp_saveipgen, &tcp_savetcp, 0);
#endif
* Add data to socket buffer.
1994-05-24 10:09:53 +00:00
*/
/* Unlocked read. */
SOCKBUF_LOCK(&so->so_rcv);
if (so->so_rcv.sb_state & SBS_CANTRCVMORE) {
m_freem(m);
} else {
m_adj(m, drop_hdrlen); /* delayed header drop */
sbappendstream_locked(&so->so_rcv, m);
}
sorwakeup_locked(so);
if (DELAY_ACK(tp)) {
tp->t_flags |= TF_DELACK;
} else {
tp->t_flags |= TF_ACKNOW;
tcp_output(tp);
}
goto check_delack;
1994-05-24 10:09:53 +00:00
}
}
/*
* 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 = imax(win, (int)(tp->rcv_adv - tp->rcv_nxt));
1994-05-24 10:09:53 +00:00
}
switch (tp->t_state) {
/*
* If the state is SYN_RECEIVED:
* if seg contains an ACK, but not for our SYN/ACK, send a RST.
*/
case TCPS_SYN_RECEIVED:
if ((thflags & TH_ACK) &&
(SEQ_LEQ(th->th_ack, tp->snd_una) ||
SEQ_GT(th->th_ack, tp->snd_max))) {
rstreason = BANDLIM_RST_OPENPORT;
goto dropwithreset;
}
break;
1994-05-24 10:09:53 +00:00
/*
* 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 ((thflags & TH_ACK) &&
(SEQ_LEQ(th->th_ack, tp->iss) ||
SEQ_GT(th->th_ack, tp->snd_max))) {
rstreason = BANDLIM_UNLIMITED;
goto dropwithreset;
}
if (thflags & TH_RST) {
if (thflags & TH_ACK)
1994-05-24 10:09:53 +00:00
tp = tcp_drop(tp, ECONNREFUSED);
goto drop;
}
if ((thflags & TH_SYN) == 0)
1994-05-24 10:09:53 +00:00
goto drop;
tp->snd_wnd = th->th_win; /* initial send window */
tp->irs = th->th_seq;
1994-05-24 10:09:53 +00:00
tcp_rcvseqinit(tp);
if (thflags & TH_ACK) {
tcpstat.tcps_connects++;
soisconnected(so);
#ifdef MAC
SOCK_LOCK(so);
mac_set_socket_peer_from_mbuf(m, so);
SOCK_UNLOCK(so);
#endif
/* 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;
}
tp->rcv_adv += tp->rcv_wnd;
tp->snd_una++; /* SYN is acked */
/*
* If there's data, delay ACK; if there's also a FIN
* ACKNOW will be turned on later.
*/
if (DELAY_ACK(tp) && tlen != 0)
callout_reset(tp->tt_delack, tcp_delacktime,
tcp_timer_delack, tp);
else
tp->t_flags |= TF_ACKNOW;
/*
* Received <SYN,ACK> in SYN_SENT[*] state.
* Transitions:
* SYN_SENT --> ESTABLISHED
* SYN_SENT* --> FIN_WAIT_1
*/
tp->t_starttime = ticks;
if (tp->t_flags & TF_NEEDFIN) {
tp->t_state = TCPS_FIN_WAIT_1;
tp->t_flags &= ~TF_NEEDFIN;
thflags &= ~TH_SYN;
} else {
tp->t_state = TCPS_ESTABLISHED;
callout_reset(tp->tt_keep, tcp_keepidle,
tcp_timer_keep, tp);
}
} else {
/*
* Received initial SYN in SYN-SENT[*] state =>
* simultaneous open. If segment contains CC option
* and there is a cached CC, apply TAO test.
* If it succeeds, connection is * half-synchronized.
* Otherwise, do 3-way handshake:
* SYN-SENT -> SYN-RECEIVED
* SYN-SENT* -> SYN-RECEIVED*
* If there was no CC option, clear cached CC value.
*/
tp->t_flags |= TF_ACKNOW;
callout_stop(tp->tt_rexmt);
tp->t_state = TCPS_SYN_RECEIVED;
1995-05-30 08:16:23 +00:00
}
1994-05-24 10:09:53 +00:00
trimthenstep6:
KASSERT(headlocked,
("tcp_input(): trimthenstep6 head is not locked"));
INP_LOCK_ASSERT(inp);
1994-05-24 10:09:53 +00:00
/*
* Advance th->th_seq to correspond to first data byte.
1994-05-24 10:09:53 +00:00
* If data, trim to stay within window,
* dropping FIN if necessary.
*/
th->th_seq++;
if (tlen > tp->rcv_wnd) {
todrop = tlen - tp->rcv_wnd;
1994-05-24 10:09:53 +00:00
m_adj(m, -todrop);
tlen = tp->rcv_wnd;
thflags &= ~TH_FIN;
1994-05-24 10:09:53 +00:00
tcpstat.tcps_rcvpackafterwin++;
tcpstat.tcps_rcvbyteafterwin += todrop;
}
tp->snd_wl1 = th->th_seq - 1;
tp->rcv_up = th->th_seq;
/*
* Client side of transaction: already sent SYN and data.
* If the remote host used T/TCP to validate the SYN,
* our data will be ACK'd; if so, enter normal data segment
* processing in the middle of step 5, ack processing.
* Otherwise, goto step 6.
1995-05-30 08:16:23 +00:00
*/
if (thflags & TH_ACK)
goto process_ACK;
1994-05-24 10:09:53 +00:00
goto step6;
/*
* If the state is LAST_ACK or CLOSING or TIME_WAIT:
* do normal processing.
*
* NB: Leftover from RFC1644 T/TCP. Cases to be reused later.
1995-05-30 08:16:23 +00:00
*/
case TCPS_LAST_ACK:
case TCPS_CLOSING:
case TCPS_TIME_WAIT:
KASSERT(tp->t_state != TCPS_TIME_WAIT, ("timewait"));
break; /* continue normal processing */
1994-05-24 10:09:53 +00:00
}
/*
* States other than LISTEN or SYN_SENT.
* First check the RST flag and sequence number since reset segments
* are exempt from the timestamp and connection count tests. This
* fixes a bug introduced by the Stevens, vol. 2, p. 960 bugfix
* below which allowed reset segments in half the sequence space
* to fall though and be processed (which gives forged reset
* segments with a random sequence number a 50 percent chance of
* killing a connection).
* Then check timestamp, if present.
* Then check the connection count, if present.
1995-05-30 08:16:23 +00:00
* Then check that at least some bytes of segment are within
1994-05-24 10:09:53 +00:00
* receive window. If segment begins before rcv_nxt,
* drop leading data (and SYN); if nothing left, just ack.
*
*
* If the RST bit is set, check the sequence number to see
* if this is a valid reset segment.
* RFC 793 page 37:
* In all states except SYN-SENT, all reset (RST) segments
* are validated by checking their SEQ-fields. A reset is
* valid if its sequence number is in the window.
* Note: this does not take into account delayed ACKs, so
* we should test against last_ack_sent instead of rcv_nxt.
* The sequence number in the reset segment is normally an
* echo of our outgoing acknowlegement numbers, but some hosts
* send a reset with the sequence number at the rightmost edge
* of our receive window, and we have to handle this case.
* Note 2: Paul Watson's paper "Slipping in the Window" has shown
* that brute force RST attacks are possible. To combat this,
* we use a much stricter check while in the ESTABLISHED state,
* only accepting RSTs where the sequence number is equal to
* last_ack_sent. In all other states (the states in which a
* RST is more likely), the more permissive check is used.
* If we have multiple segments in flight, the intial reset
* segment sequence numbers will be to the left of last_ack_sent,
* but they will eventually catch up.
* In any case, it never made sense to trim reset segments to
* fit the receive window since RFC 1122 says:
* 4.2.2.12 RST Segment: RFC-793 Section 3.4
*
* A TCP SHOULD allow a received RST segment to include data.
*
* DISCUSSION
* It has been suggested that a RST segment could contain
* ASCII text that encoded and explained the cause of the
* RST. No standard has yet been established for such
* data.
*
* If the reset segment passes the sequence number test 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_WAIT_2, CLOSE_WAIT STATES:
* Inform user that connection was reset, and close tcb.
* CLOSING, LAST_ACK STATES:
* Close the tcb.
* TIME_WAIT STATE:
* Drop the segment - see Stevens, vol. 2, p. 964 and
* RFC 1337.
*/
if (thflags & TH_RST) {
if (SEQ_GEQ(th->th_seq, tp->last_ack_sent) &&
SEQ_LT(th->th_seq, tp->last_ack_sent + tp->rcv_wnd)) {
switch (tp->t_state) {
case TCPS_SYN_RECEIVED:
so->so_error = ECONNREFUSED;
goto close;
case TCPS_ESTABLISHED:
if (tp->last_ack_sent != th->th_seq) {
tcpstat.tcps_badrst++;
goto drop;
}
case TCPS_FIN_WAIT_1:
case TCPS_FIN_WAIT_2:
case TCPS_CLOSE_WAIT:
so->so_error = ECONNRESET;
close:
tp->t_state = TCPS_CLOSED;
tcpstat.tcps_drops++;
tp = tcp_close(tp);
break;
case TCPS_CLOSING:
case TCPS_LAST_ACK:
tp = tcp_close(tp);
break;
case TCPS_TIME_WAIT:
KASSERT(tp->t_state != TCPS_TIME_WAIT,
("timewait"));
break;
}
}
goto drop;
}
/*
1994-05-24 10:09:53 +00:00
* RFC 1323 PAWS: If we have a timestamp reply on this segment
* and it's less than ts_recent, drop it.
*/
if ((to.to_flags & TOF_TS) != 0 && tp->ts_recent &&
TSTMP_LT(to.to_tsval, tp->ts_recent)) {
1994-05-24 10:09:53 +00:00
/* Check to see if ts_recent is over 24 days old. */
if ((int)(ticks - tp->ts_recent_age) > TCP_PAWS_IDLE) {
1994-05-24 10:09:53 +00:00
/*
* 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 {
tcpstat.tcps_rcvduppack++;
tcpstat.tcps_rcvdupbyte += tlen;
1994-05-24 10:09:53 +00:00
tcpstat.tcps_pawsdrop++;
if (tlen)
2002-12-17 00:24:48 +00:00
goto dropafterack;
goto drop;
1994-05-24 10:09:53 +00:00
}
}
/*
* In the SYN-RECEIVED state, validate that the packet belongs to
* this connection before trimming the data to fit the receive
* window. Check the sequence number versus IRS since we know
* the sequence numbers haven't wrapped. This is a partial fix
* for the "LAND" DoS attack.
*/
if (tp->t_state == TCPS_SYN_RECEIVED && SEQ_LT(th->th_seq, tp->irs)) {
rstreason = BANDLIM_RST_OPENPORT;
goto dropwithreset;
}
todrop = tp->rcv_nxt - th->th_seq;
1994-05-24 10:09:53 +00:00
if (todrop > 0) {
if (thflags & TH_SYN) {
thflags &= ~TH_SYN;
th->th_seq++;
if (th->th_urp > 1)
th->th_urp--;
1994-05-24 10:09:53 +00:00
else
thflags &= ~TH_URG;
1994-05-24 10:09:53 +00:00
todrop--;
}
/*
* Following if statement from Stevens, vol. 2, p. 960.
*/
if (todrop > tlen
|| (todrop == tlen && (thflags & TH_FIN) == 0)) {
1994-05-24 10:09:53 +00:00
/*
* 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.
1994-05-24 10:09:53 +00:00
*/
thflags &= ~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 = tlen;
tcpstat.tcps_rcvduppack++;
tcpstat.tcps_rcvdupbyte += todrop;
1994-05-24 10:09:53 +00:00
} else {
tcpstat.tcps_rcvpartduppack++;
tcpstat.tcps_rcvpartdupbyte += todrop;
}
drop_hdrlen += todrop; /* drop from the top afterwards */
th->th_seq += todrop;
tlen -= todrop;
if (th->th_urp > todrop)
th->th_urp -= todrop;
1994-05-24 10:09:53 +00:00
else {
thflags &= ~TH_URG;
th->th_urp = 0;
1994-05-24 10:09:53 +00:00
}
}
/*
* 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 && tlen) {
1994-05-24 10:09:53 +00:00
tp = tcp_close(tp);
tcpstat.tcps_rcvafterclose++;
rstreason = BANDLIM_UNLIMITED;
1994-05-24 10:09:53 +00:00
goto dropwithreset;
}
/*
* If segment ends after window, drop trailing data
* (and PUSH and FIN); if nothing left, just ACK.
*/
todrop = (th->th_seq+tlen) - (tp->rcv_nxt+tp->rcv_wnd);
1994-05-24 10:09:53 +00:00
if (todrop > 0) {
tcpstat.tcps_rcvpackafterwin++;
if (todrop >= tlen) {
tcpstat.tcps_rcvbyteafterwin += tlen;
1994-05-24 10:09:53 +00:00
/*
* 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.
*/
KASSERT(tp->t_state != TCPS_TIME_WAIT, ("timewait"));
if (thflags & TH_SYN &&
1994-05-24 10:09:53 +00:00
tp->t_state == TCPS_TIME_WAIT &&
SEQ_GT(th->th_seq, tp->rcv_nxt)) {
1994-05-24 10:09:53 +00:00
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 && th->th_seq == tp->rcv_nxt) {
1994-05-24 10:09:53 +00:00
tp->t_flags |= TF_ACKNOW;
tcpstat.tcps_rcvwinprobe++;
} else
goto dropafterack;
} else
tcpstat.tcps_rcvbyteafterwin += todrop;
m_adj(m, -todrop);
tlen -= todrop;
thflags &= ~(TH_PUSH|TH_FIN);
1994-05-24 10:09:53 +00:00
}
/*
* If last ACK falls within this segment's sequence numbers,
* record its timestamp.
* NOTE that the test is modified according to the latest
* proposal of the tcplw@cray.com list (Braden 1993/04/26).
1994-05-24 10:09:53 +00:00
*/
if ((to.to_flags & TOF_TS) != 0 &&
SEQ_LEQ(th->th_seq, tp->last_ack_sent)) {
tp->ts_recent_age = ticks;
tp->ts_recent = to.to_tsval;
1994-05-24 10:09:53 +00:00
}
/*
* If a SYN is in the window, then this is an
* error and we send an RST and drop the connection.
*/
if (thflags & TH_SYN) {
1994-05-24 10:09:53 +00:00
tp = tcp_drop(tp, ECONNRESET);
rstreason = BANDLIM_UNLIMITED;
goto drop;
1994-05-24 10:09:53 +00:00
}
/*
* If the ACK bit is off: if in SYN-RECEIVED state or SENDSYN
* flag is on (half-synchronized state), then queue data for
* later processing; else drop segment and return.
*/
if ((thflags & TH_ACK) == 0) {
if (tp->t_state == TCPS_SYN_RECEIVED ||
(tp->t_flags & TF_NEEDSYN))
goto step6;
else
goto drop;
}
1995-05-30 08:16:23 +00:00
1994-05-24 10:09:53 +00:00
/*
* Ack processing.
*/
switch (tp->t_state) {
/*
* In SYN_RECEIVED state, the ack ACKs our SYN, so enter
* ESTABLISHED state and continue processing.
* The ACK was checked above.
1994-05-24 10:09:53 +00:00
*/
case TCPS_SYN_RECEIVED:
1994-05-24 10:09:53 +00:00
tcpstat.tcps_connects++;
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;
}
/*
1995-05-30 08:16:23 +00:00
* Make transitions:
* SYN-RECEIVED -> ESTABLISHED
* SYN-RECEIVED* -> FIN-WAIT-1
*/
tp->t_starttime = ticks;
if (tp->t_flags & TF_NEEDFIN) {
tp->t_state = TCPS_FIN_WAIT_1;
tp->t_flags &= ~TF_NEEDFIN;
} else {
tp->t_state = TCPS_ESTABLISHED;
callout_reset(tp->tt_keep, tcp_keepidle,
tcp_timer_keep, tp);
}
1995-05-30 08:16:23 +00:00
/*
* If segment contains data or ACK, will call tcp_reass()
* later; if not, do so now to pass queued data to user.
*/
if (tlen == 0 && (thflags & TH_FIN) == 0)
(void) tcp_reass(tp, (struct tcphdr *)0, 0,
(struct mbuf *)0);
tp->snd_wl1 = th->th_seq - 1;
/* FALLTHROUGH */
1994-05-24 10:09:53 +00:00
/*
* In ESTABLISHED state: drop duplicate ACKs; ACK out of range
* ACKs. If the ack is in the range
* tp->snd_una < th->th_ack <= tp->snd_max
* then advance tp->snd_una to th->th_ack and drop
1994-05-24 10:09:53 +00:00
* 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:
KASSERT(tp->t_state != TCPS_TIME_WAIT, ("timewait"));
if (SEQ_LEQ(th->th_ack, tp->snd_una)) {
if (tlen == 0 && tiwin == tp->snd_wnd) {
1994-05-24 10:09:53 +00:00
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
1995-05-30 08:16:23 +00:00
* network (they're now cached at the receiver)
1994-05-24 10:09:53 +00:00
* so bump cwnd by the amount in the receiver
* to keep a constant cwnd packets in the
* network.
*/
if (!callout_active(tp->tt_rexmt) ||
th->th_ack != tp->snd_una)
1994-05-24 10:09:53 +00:00
tp->t_dupacks = 0;
else if (++tp->t_dupacks > tcprexmtthresh ||
((tcp_do_newreno || tp->sack_enable) &&
IN_FASTRECOVERY(tp))) {
tp->snd_cwnd += tp->t_maxseg;
(void) tcp_output(tp);
goto drop;
} else if (tp->t_dupacks == tcprexmtthresh) {
1994-05-24 10:09:53 +00:00
tcp_seq onxt = tp->snd_nxt;
u_int win;
/*
* If we're doing sack, check to
* see if we're already in sack
* recovery. If we're not doing sack,
* check to see if we're in newreno
* recovery.
*/
if (tp->sack_enable) {
if (IN_FASTRECOVERY(tp)) {
tp->t_dupacks = 0;
break;
}
} else if (tcp_do_newreno) {
if (SEQ_LEQ(th->th_ack,
tp->snd_recover)) {
tp->t_dupacks = 0;
break;
}
}
win = min(tp->snd_wnd, tp->snd_cwnd) /
2 / tp->t_maxseg;
1994-05-24 10:09:53 +00:00
if (win < 2)
win = 2;
tp->snd_ssthresh = win * tp->t_maxseg;
ENTER_FASTRECOVERY(tp);
tp->snd_recover = tp->snd_max;
callout_stop(tp->tt_rexmt);
tp->t_rtttime = 0;
if (tp->sack_enable) {
tcpstat.tcps_sack_recovery_episode++;
tp->sack_newdata = tp->snd_nxt;
tp->snd_cwnd =
tp->t_maxseg * tcp_sack_recovery_initburst;
(void) tcp_output(tp);
tp->snd_cwnd +=
tp->snd_ssthresh;
goto drop;
}
tp->snd_nxt = th->th_ack;
1994-05-24 10:09:53 +00:00
tp->snd_cwnd = tp->t_maxseg;
(void) tcp_output(tp);
KASSERT(tp->snd_limited <= 2,
("tp->snd_limited too big"));
1994-05-24 10:09:53 +00:00
tp->snd_cwnd = tp->snd_ssthresh +
tp->t_maxseg *
(tp->t_dupacks - tp->snd_limited);
1994-05-24 10:09:53 +00:00
if (SEQ_GT(onxt, tp->snd_nxt))
tp->snd_nxt = onxt;
goto drop;
} else if (tcp_do_rfc3042) {
u_long oldcwnd = tp->snd_cwnd;
tcp_seq oldsndmax = tp->snd_max;
u_int sent;
KASSERT(tp->t_dupacks == 1 ||
tp->t_dupacks == 2,
("dupacks not 1 or 2"));
if (tp->t_dupacks == 1)
tp->snd_limited = 0;
tp->snd_cwnd =
(tp->snd_nxt - tp->snd_una) +
(tp->t_dupacks - tp->snd_limited) *
tp->t_maxseg;
(void) tcp_output(tp);
sent = tp->snd_max - oldsndmax;
if (sent > tp->t_maxseg) {
KASSERT((tp->t_dupacks == 2 &&
tp->snd_limited == 0) ||
(sent == tp->t_maxseg + 1 &&
tp->t_flags & TF_SENTFIN),
("sent too much"));
tp->snd_limited = 2;
} else if (sent > 0)
++tp->snd_limited;
tp->snd_cwnd = oldcwnd;
goto drop;
1994-05-24 10:09:53 +00:00
}
} else
tp->t_dupacks = 0;
break;
}
KASSERT(SEQ_GT(th->th_ack, tp->snd_una), ("th_ack <= snd_una"));
1994-05-24 10:09:53 +00:00
/*
* If the congestion window was inflated to account
* for the other side's cached packets, retract it.
*/
if (tcp_do_newreno || tp->sack_enable) {
if (IN_FASTRECOVERY(tp)) {
if (SEQ_LT(th->th_ack, tp->snd_recover)) {
if (tp->sack_enable)
tcp_sack_partialack(tp, th);
else
tcp_newreno_partial_ack(tp, th);
} else {
/*
* Out of fast recovery.
* Window inflation should have left us
* with approximately snd_ssthresh
* outstanding data.
* But in case we would be inclined to
* send a burst, better to do it via
* the slow start mechanism.
*/
if (SEQ_GT(th->th_ack +
tp->snd_ssthresh,
tp->snd_max))
tp->snd_cwnd = tp->snd_max -
th->th_ack +
tp->t_maxseg;
else
tp->snd_cwnd = tp->snd_ssthresh;
}
}
} else {
if (tp->t_dupacks >= tcprexmtthresh &&
tp->snd_cwnd > tp->snd_ssthresh)
tp->snd_cwnd = tp->snd_ssthresh;
}
tp->t_dupacks = 0;
if (SEQ_GT(th->th_ack, tp->snd_max)) {
1994-05-24 10:09:53 +00:00
tcpstat.tcps_rcvacktoomuch++;
goto dropafterack;
}
/*
* If we reach this point, ACK is not a duplicate,
* i.e., it ACKs something we sent.
*/
if (tp->t_flags & TF_NEEDSYN) {
1995-05-30 08:16:23 +00:00
/*
* T/TCP: Connection was half-synchronized, and our
* SYN has been ACK'd (so connection is now fully
* synchronized). Go to non-starred state,
* increment snd_una for ACK of SYN, and check if
* we can do window scaling.
*/
tp->t_flags &= ~TF_NEEDSYN;
tp->snd_una++;
/* 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;
}
}
process_ACK:
KASSERT(headlocked,
("tcp_input(): process_ACK head is not locked"));
INP_LOCK_ASSERT(inp);
acked = th->th_ack - tp->snd_una;
1994-05-24 10:09:53 +00:00
tcpstat.tcps_rcvackpack++;
tcpstat.tcps_rcvackbyte += acked;
/*
* If we just performed our first retransmit, and the ACK
* arrives within our recovery window, then it was a mistake
* to do the retransmit in the first place. Recover our
* original cwnd and ssthresh, and proceed to transmit where
* we left off.
*/
if (tp->t_rxtshift == 1 && ticks < tp->t_badrxtwin) {
++tcpstat.tcps_sndrexmitbad;
tp->snd_cwnd = tp->snd_cwnd_prev;
tp->snd_ssthresh = tp->snd_ssthresh_prev;
tp->snd_recover = tp->snd_recover_prev;
if (tp->t_flags & TF_WASFRECOVERY)
ENTER_FASTRECOVERY(tp);
tp->snd_nxt = tp->snd_max;
tp->t_badrxtwin = 0; /* XXX probably not required */
}
1994-05-24 10:09:53 +00:00
/*
* 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.
*
* Some boxes send broken timestamp replies
* during the SYN+ACK phase, ignore
* timestamps of 0 or we could calculate a
* huge RTT and blow up the retransmit timer.
1994-05-24 10:09:53 +00:00
*/
if ((to.to_flags & TOF_TS) != 0 &&
to.to_tsecr) {
tcp_xmit_timer(tp, ticks - to.to_tsecr + 1);
} else if (tp->t_rtttime && SEQ_GT(th->th_ack, tp->t_rtseq)) {
tcp_xmit_timer(tp, ticks - tp->t_rtttime);
}
tcp_xmit_bandwidth_limit(tp, th->th_ack);
1994-05-24 10:09:53 +00:00
/*
* 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 (th->th_ack == tp->snd_max) {
callout_stop(tp->tt_rexmt);
1994-05-24 10:09:53 +00:00
needoutput = 1;
} else if (!callout_active(tp->tt_persist))
callout_reset(tp->tt_rexmt, tp->t_rxtcur,
tcp_timer_rexmt, tp);
/*
* If no data (only SYN) was ACK'd,
* skip rest of ACK processing.
*/
if (acked == 0)
goto step6;
1994-05-24 10:09:53 +00:00
/*
* 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).
1994-05-24 10:09:53 +00:00
*/
if ((!tcp_do_newreno && !tp->sack_enable) ||
!IN_FASTRECOVERY(tp)) {
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);
1994-05-24 10:09:53 +00:00
}
SOCKBUF_LOCK(&so->so_snd);
1994-05-24 10:09:53 +00:00
if (acked > so->so_snd.sb_cc) {
tp->snd_wnd -= so->so_snd.sb_cc;
sbdrop_locked(&so->so_snd, (int)so->so_snd.sb_cc);
1994-05-24 10:09:53 +00:00
ourfinisacked = 1;
} else {
sbdrop_locked(&so->so_snd, acked);
1994-05-24 10:09:53 +00:00
tp->snd_wnd -= acked;
ourfinisacked = 0;
}
sowwakeup_locked(so);
/* detect una wraparound */
if ((tcp_do_newreno || tp->sack_enable) &&
!IN_FASTRECOVERY(tp) &&
SEQ_GT(tp->snd_una, tp->snd_recover) &&
SEQ_LEQ(th->th_ack, tp->snd_recover))
tp->snd_recover = th->th_ack - 1;
if ((tcp_do_newreno || tp->sack_enable) &&
IN_FASTRECOVERY(tp) &&
SEQ_GEQ(th->th_ack, tp->snd_recover))
EXIT_FASTRECOVERY(tp);
tp->snd_una = th->th_ack;
if (tp->sack_enable) {
if (SEQ_GT(tp->snd_una, tp->snd_recover))
tp->snd_recover = tp->snd_una;
}
1994-05-24 10:09:53 +00:00
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.
*/
/* XXXjl
* we should release the tp also, and use a
* compressed state.
*/
if (so->so_rcv.sb_state & SBS_CANTRCVMORE) {
soisdisconnected(so);
callout_reset(tp->tt_2msl, tcp_maxidle,
tcp_timer_2msl, tp);
}
1994-05-24 10:09:53 +00:00
tp->t_state = TCPS_FIN_WAIT_2;
}
break;
/*
1994-05-24 10:09:53 +00:00
* 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) {
KASSERT(headlocked, ("headlocked"));
tcp_twstart(tp);
INP_INFO_WUNLOCK(&tcbinfo);
m_freem(m);
return;
1994-05-24 10:09:53 +00:00
}
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:
KASSERT(tp->t_state != TCPS_TIME_WAIT, ("timewait"));
callout_reset(tp->tt_2msl, 2 * tcp_msl,
tcp_timer_2msl, tp);
1994-05-24 10:09:53 +00:00
goto dropafterack;
}
}
step6:
KASSERT(headlocked, ("tcp_input(): step6 head is not locked"));
INP_LOCK_ASSERT(inp);
1994-05-24 10:09:53 +00:00
/*
* Update window information.
* Don't look at window if no ACK: TAC's send garbage on first SYN.
*/
if ((thflags & TH_ACK) &&
(SEQ_LT(tp->snd_wl1, th->th_seq) ||
(tp->snd_wl1 == th->th_seq && (SEQ_LT(tp->snd_wl2, th->th_ack) ||
(tp->snd_wl2 == th->th_ack && tiwin > tp->snd_wnd))))) {
1994-05-24 10:09:53 +00:00
/* keep track of pure window updates */
if (tlen == 0 &&
tp->snd_wl2 == th->th_ack && tiwin > tp->snd_wnd)
1994-05-24 10:09:53 +00:00
tcpstat.tcps_rcvwinupd++;
tp->snd_wnd = tiwin;
tp->snd_wl1 = th->th_seq;
tp->snd_wl2 = th->th_ack;
1994-05-24 10:09:53 +00:00
if (tp->snd_wnd > tp->max_sndwnd)
tp->max_sndwnd = tp->snd_wnd;
needoutput = 1;
}
/*
* Process segments with URG.
*/
if ((thflags & TH_URG) && th->th_urp &&
1994-05-24 10:09:53 +00:00
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 (th->th_urp + so->so_rcv.sb_cc > sb_max) {
th->th_urp = 0; /* XXX */
thflags &= ~TH_URG; /* XXX */
1994-05-24 10:09:53 +00:00
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
1995-05-30 08:16:23 +00:00
* a FIN has been received from the remote side.
1994-05-24 10:09:53 +00:00
* 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
1995-05-30 08:16:23 +00:00
* of data past the urgent section as the original
1994-05-24 10:09:53 +00:00
* spec states (in one of two places).
*/
if (SEQ_GT(th->th_seq+th->th_urp, tp->rcv_up)) {
tp->rcv_up = th->th_seq + th->th_urp;
SOCKBUF_LOCK(&so->so_rcv);
1994-05-24 10:09:53 +00:00
so->so_oobmark = so->so_rcv.sb_cc +
(tp->rcv_up - tp->rcv_nxt) - 1;
if (so->so_oobmark == 0)
so->so_rcv.sb_state |= SBS_RCVATMARK;
SOCKBUF_UNLOCK(&so->so_rcv);
1994-05-24 10:09:53 +00:00
sohasoutofband(so);
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 (th->th_urp <= (u_long)tlen &&
!(so->so_options & SO_OOBINLINE)) {
/* hdr drop is delayed */
tcp_pulloutofband(so, th, m, drop_hdrlen);
}
} else {
1994-05-24 10:09:53 +00:00
/*
* 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;
}
1994-05-24 10:09:53 +00:00
dodata: /* XXX */
KASSERT(headlocked, ("tcp_input(): dodata head is not locked"));
INP_LOCK_ASSERT(inp);
1994-05-24 10:09:53 +00:00
/*
* 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 ((tlen || (thflags & TH_FIN)) &&
1994-05-24 10:09:53 +00:00
TCPS_HAVERCVDFIN(tp->t_state) == 0) {
m_adj(m, drop_hdrlen); /* delayed header drop */
/*
* Insert segment which includes th into TCP reassembly queue
* with control block tp. Set thflags to whether reassembly now
* includes a segment with FIN. This handles the common case
* inline (segment is the next to be received on an established
* connection, and the queue is empty), avoiding linkage into
* and removal from the queue and repetition of various
* conversions.
* Set DELACK for segments received in order, but ack
* immediately when segments are out of order (so
* fast retransmit can work).
*/
if (th->th_seq == tp->rcv_nxt &&
LIST_EMPTY(&tp->t_segq) &&
TCPS_HAVEESTABLISHED(tp->t_state)) {
if (DELAY_ACK(tp))
tp->t_flags |= TF_DELACK;
else
tp->t_flags |= TF_ACKNOW;
tp->rcv_nxt += tlen;
thflags = th->th_flags & TH_FIN;
tcpstat.tcps_rcvpack++;
tcpstat.tcps_rcvbyte += tlen;
ND6_HINT(tp);
/* Unlocked read. */
SOCKBUF_LOCK(&so->so_rcv);
if (so->so_rcv.sb_state & SBS_CANTRCVMORE)
m_freem(m);
else
sbappendstream_locked(&so->so_rcv, m);
sorwakeup_locked(so);
} else {
thflags = tcp_reass(tp, th, &tlen, m);
tp->t_flags |= TF_ACKNOW;
}
if (tp->sack_enable)
tcp_update_sack_list(tp);
1994-05-24 10:09:53 +00:00
/*
* 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 {
m_freem(m);
thflags &= ~TH_FIN;
1994-05-24 10:09:53 +00:00
}
/*
* If FIN is received ACK the FIN and let the user know
* that the connection is closing.
*/
if (thflags & TH_FIN) {
1994-05-24 10:09:53 +00:00
if (TCPS_HAVERCVDFIN(tp->t_state) == 0) {
socantrcvmore(so);
/*
* If connection is half-synchronized
* (ie NEEDSYN flag on) then delay ACK,
* so it may be piggybacked when SYN is sent.
* Otherwise, since we received a FIN then no
* more input can be expected, send ACK now.
*/
if (tp->t_flags & TF_NEEDSYN)
tp->t_flags |= TF_DELACK;
1995-05-30 08:16:23 +00:00
else
tp->t_flags |= TF_ACKNOW;
1994-05-24 10:09:53 +00:00
tp->rcv_nxt++;
}
switch (tp->t_state) {
/*
1994-05-24 10:09:53 +00:00
* In SYN_RECEIVED and ESTABLISHED STATES
* enter the CLOSE_WAIT state.
*/
case TCPS_SYN_RECEIVED:
tp->t_starttime = ticks;
/*FALLTHROUGH*/
1994-05-24 10:09:53 +00:00
case TCPS_ESTABLISHED:
tp->t_state = TCPS_CLOSE_WAIT;
break;
/*
1994-05-24 10:09:53 +00:00
* 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;
/*
1994-05-24 10:09:53 +00:00
* In FIN_WAIT_2 state enter the TIME_WAIT state,
1995-05-30 08:16:23 +00:00
* starting the time-wait timer, turning off the other
1994-05-24 10:09:53 +00:00
* standard timers.
*/
case TCPS_FIN_WAIT_2:
KASSERT(headlocked == 1, ("headlocked should be 1"));
tcp_twstart(tp);
INP_INFO_WUNLOCK(&tcbinfo);
return;
1994-05-24 10:09:53 +00:00
/*
* In TIME_WAIT state restart the 2 MSL time_wait timer.
*/
case TCPS_TIME_WAIT:
KASSERT(tp->t_state != TCPS_TIME_WAIT, ("timewait"));
callout_reset(tp->tt_2msl, 2 * tcp_msl,
tcp_timer_2msl, tp);
1994-05-24 10:09:53 +00:00
break;
}
}
INP_INFO_WUNLOCK(&tcbinfo);
#ifdef TCPDEBUG
if (so->so_options & SO_DEBUG)
tcp_trace(TA_INPUT, ostate, tp, (void *)tcp_saveipgen,
&tcp_savetcp, 0);
#endif
1994-05-24 10:09:53 +00:00
/*
* Return any desired output.
*/
if (needoutput || (tp->t_flags & TF_ACKNOW))
(void) tcp_output(tp);
check_delack:
INP_LOCK_ASSERT(inp);
if (tp->t_flags & TF_DELACK) {
tp->t_flags &= ~TF_DELACK;
callout_reset(tp->tt_delack, tcp_delacktime,
tcp_timer_delack, tp);
}
INP_UNLOCK(inp);
1994-05-24 10:09:53 +00:00
return;
dropafterack:
/*
* Generate an ACK dropping incoming segment if it occupies
* sequence space, where the ACK reflects our state.
*
* We can now skip the test for the RST flag since all
* paths to this code happen after packets containing
* RST have been dropped.
*
* In the SYN-RECEIVED state, don't send an ACK unless the
* segment we received passes the SYN-RECEIVED ACK test.
* If it fails send a RST. This breaks the loop in the
* "LAND" DoS attack, and also prevents an ACK storm
* between two listening ports that have been sent forged
* SYN segments, each with the source address of the other.
1994-05-24 10:09:53 +00:00
*/
if (tp->t_state == TCPS_SYN_RECEIVED && (thflags & TH_ACK) &&
(SEQ_GT(tp->snd_una, th->th_ack) ||
SEQ_GT(th->th_ack, tp->snd_max)) ) {
rstreason = BANDLIM_RST_OPENPORT;
goto dropwithreset;
}
#ifdef TCPDEBUG
if (so->so_options & SO_DEBUG)
tcp_trace(TA_DROP, ostate, tp, (void *)tcp_saveipgen,
&tcp_savetcp, 0);
#endif
KASSERT(headlocked, ("headlocked should be 1"));
INP_INFO_WUNLOCK(&tcbinfo);
1994-05-24 10:09:53 +00:00
m_freem(m);
tp->t_flags |= TF_ACKNOW;
(void) tcp_output(tp);
INP_UNLOCK(inp);
1994-05-24 10:09:53 +00:00
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 ((thflags & TH_RST) || m->m_flags & (M_BCAST|M_MCAST))
1994-05-24 10:09:53 +00:00
goto drop;
if (isipv6) {
if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
IN6_IS_ADDR_MULTICAST(&ip6->ip6_src))
goto drop;
} else {
if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
IN_MULTICAST(ntohl(ip->ip_src.s_addr)) ||
ip->ip_src.s_addr == htonl(INADDR_BROADCAST) ||
in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif))
goto drop;
}
/* IPv6 anycast check is done at tcp6_input() */
/*
* Perform bandwidth limiting.
*/
if (badport_bandlim(rstreason) < 0)
goto drop;
#ifdef TCPDEBUG
if (tp == 0 || (tp->t_inpcb->inp_socket->so_options & SO_DEBUG))
tcp_trace(TA_DROP, ostate, tp, (void *)tcp_saveipgen,
&tcp_savetcp, 0);
#endif
if (thflags & TH_ACK)
/* mtod() below is safe as long as hdr dropping is delayed */
tcp_respond(tp, mtod(m, void *), th, m, (tcp_seq)0, th->th_ack,
TH_RST);
1994-05-24 10:09:53 +00:00
else {
if (thflags & TH_SYN)
tlen++;
/* mtod() below is safe as long as hdr dropping is delayed */
tcp_respond(tp, mtod(m, void *), th, m, th->th_seq+tlen,
(tcp_seq)0, TH_RST|TH_ACK);
1994-05-24 10:09:53 +00:00
}
if (tp)
INP_UNLOCK(inp);
if (headlocked)
INP_INFO_WUNLOCK(&tcbinfo);
1994-05-24 10:09:53 +00:00
return;
drop:
/*
* Drop space held by incoming segment and return.
*/
#ifdef TCPDEBUG
if (tp == 0 || (tp->t_inpcb->inp_socket->so_options & SO_DEBUG))
tcp_trace(TA_DROP, ostate, tp, (void *)tcp_saveipgen,
&tcp_savetcp, 0);
#endif
if (tp)
INP_UNLOCK(inp);
1994-05-24 10:09:53 +00:00
m_freem(m);
if (headlocked)
INP_INFO_WUNLOCK(&tcbinfo);
1994-05-24 10:09:53 +00:00
return;
}
/*
* Parse TCP options and place in tcpopt.
*/
static void
tcp_dooptions(tp, to, cp, cnt, is_syn, th)
struct tcpcb *tp;
struct tcpopt *to;
u_char *cp;
1994-05-24 10:09:53 +00:00
int cnt;
2003-04-21 16:27:46 +00:00
int is_syn;
struct tcphdr *th;
1994-05-24 10:09:53 +00:00
{
int opt, optlen;
to->to_flags = 0;
1994-05-24 10:09:53 +00:00
for (; cnt > 0; cnt -= optlen, cp += optlen) {
opt = cp[0];
if (opt == TCPOPT_EOL)
break;
if (opt == TCPOPT_NOP)
optlen = 1;
else {
if (cnt < 2)
break;
1994-05-24 10:09:53 +00:00
optlen = cp[1];
if (optlen < 2 || optlen > cnt)
1994-05-24 10:09:53 +00:00
break;
}
switch (opt) {
case TCPOPT_MAXSEG:
if (optlen != TCPOLEN_MAXSEG)
continue;
if (!is_syn)
1994-05-24 10:09:53 +00:00
continue;
to->to_flags |= TOF_MSS;
bcopy((char *)cp + 2,
(char *)&to->to_mss, sizeof(to->to_mss));
to->to_mss = ntohs(to->to_mss);
1994-05-24 10:09:53 +00:00
break;
case TCPOPT_WINDOW:
if (optlen != TCPOLEN_WINDOW)
continue;
if (! is_syn)
1994-05-24 10:09:53 +00:00
continue;
to->to_flags |= TOF_SCALE;
to->to_requested_s_scale = min(cp[2], TCP_MAX_WINSHIFT);
1994-05-24 10:09:53 +00:00
break;
case TCPOPT_TIMESTAMP:
if (optlen != TCPOLEN_TIMESTAMP)
continue;
to->to_flags |= TOF_TS;
bcopy((char *)cp + 2,
(char *)&to->to_tsval, sizeof(to->to_tsval));
to->to_tsval = ntohl(to->to_tsval);
bcopy((char *)cp + 6,
(char *)&to->to_tsecr, sizeof(to->to_tsecr));
to->to_tsecr = ntohl(to->to_tsecr);
1994-05-24 10:09:53 +00:00
break;
Initial import of RFC 2385 (TCP-MD5) digest support. This is the first of two commits; bringing in the kernel support first. This can be enabled by compiling a kernel with options TCP_SIGNATURE and FAST_IPSEC. For the uninitiated, this is a TCP option which provides for a means of authenticating TCP sessions which came into being before IPSEC. It is still relevant today, however, as it is used by many commercial router vendors, particularly with BGP, and as such has become a requirement for interconnect at many major Internet points of presence. Several parts of the TCP and IP headers, including the segment payload, are digested with MD5, including a shared secret. The PF_KEY interface is used to manage the secrets using security associations in the SADB. There is a limitation here in that as there is no way to map a TCP flow per-port back to an SPI without polluting tcpcb or using the SPD; the code to do the latter is unstable at this time. Therefore this code only supports per-host keying granularity. Whilst FAST_IPSEC is mutually exclusive with KAME IPSEC (and thus IPv6), TCP_SIGNATURE applies only to IPv4. For the vast majority of prospective users of this feature, this will not pose any problem. This implementation is output-only; that is, the option is honoured when responding to a host initiating a TCP session, but no effort is made [yet] to authenticate inbound traffic. This is, however, sufficient to interwork with Cisco equipment. Tested with a Cisco 2501 running IOS 12.0(27), and Quagga 0.96.4 with local patches. Patches for tcpdump to validate TCP-MD5 sessions are also available from me upon request. Sponsored by: sentex.net
2004-02-11 04:26:04 +00:00
#ifdef TCP_SIGNATURE
/*
* XXX In order to reply to a host which has set the
* TCP_SIGNATURE option in its initial SYN, we have to
* record the fact that the option was observed here
* for the syncache code to perform the correct response.
*/
case TCPOPT_SIGNATURE:
if (optlen != TCPOLEN_SIGNATURE)
continue;
to->to_flags |= (TOF_SIGNATURE | TOF_SIGLEN);
break;
2004-02-13 18:21:45 +00:00
#endif
case TCPOPT_SACK_PERMITTED:
if (!tcp_do_sack ||
optlen != TCPOLEN_SACK_PERMITTED)
continue;
if (is_syn) {
/* MUST only be set on SYN */
to->to_flags |= TOF_SACK;
}
break;
case TCPOPT_SACK:
if (!tp || tcp_sack_option(tp, th, cp, optlen))
continue;
break;
default:
continue;
1994-05-24 10:09:53 +00:00
}
}
}
/*
* Pull out of band byte out of a segment so
* it doesn't appear in the user's data queue.
* It is still reflected in the segment length for
* sequencing purposes.
*/
static void
tcp_pulloutofband(so, th, m, off)
1994-05-24 10:09:53 +00:00
struct socket *so;
struct tcphdr *th;
1994-05-24 10:09:53 +00:00
register struct mbuf *m;
int off; /* delayed to be droped hdrlen */
1994-05-24 10:09:53 +00:00
{
int cnt = off + th->th_urp - 1;
1995-05-30 08:16:23 +00:00
1994-05-24 10:09:53 +00:00
while (cnt >= 0) {
if (m->m_len > cnt) {
char *cp = mtod(m, caddr_t) + cnt;
struct tcpcb *tp = sototcpcb(so);
tp->t_iobc = *cp;
tp->t_oobflags |= TCPOOB_HAVEDATA;
bcopy(cp+1, cp, (unsigned)(m->m_len - cnt - 1));
m->m_len--;
if (m->m_flags & M_PKTHDR)
m->m_pkthdr.len--;
1994-05-24 10:09:53 +00:00
return;
}
cnt -= m->m_len;
m = m->m_next;
if (m == 0)
break;
}
panic("tcp_pulloutofband");
}
/*
* Collect new round-trip time estimate
* and update averages and current timeout.
*/
static void
1994-05-24 10:09:53 +00:00
tcp_xmit_timer(tp, rtt)
register struct tcpcb *tp;
int rtt;
1994-05-24 10:09:53 +00:00
{
register int delta;
tcpstat.tcps_rttupdated++;
tp->t_rttupdated++;
if (tp->t_srtt != 0) {
/*
* srtt is stored as fixed point with 5 bits after the
* binary point (i.e., scaled by 8). The following magic
* is equivalent to the smoothing algorithm in rfc793 with
* an alpha of .875 (srtt = rtt/8 + srtt*7/8 in fixed
* point). Adjust rtt to origin 0.
*/
delta = ((rtt - 1) << TCP_DELTA_SHIFT)
- (tp->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT));
if ((tp->t_srtt += delta) <= 0)
tp->t_srtt = 1;
/*
* We accumulate a smoothed rtt variance (actually, a
* smoothed mean difference), then set the retransmit
* timer to smoothed rtt + 4 times the smoothed variance.
* rttvar is stored as fixed point with 4 bits after the
* binary point (scaled by 16). The following is
* equivalent to rfc793 smoothing with an alpha of .75
* (rttvar = rttvar*3/4 + |delta| / 4). This replaces
* rfc793's wired-in beta.
*/
if (delta < 0)
delta = -delta;
delta -= tp->t_rttvar >> (TCP_RTTVAR_SHIFT - TCP_DELTA_SHIFT);
if ((tp->t_rttvar += delta) <= 0)
tp->t_rttvar = 1;
if (tp->t_rttbest > tp->t_srtt + tp->t_rttvar)
tp->t_rttbest = tp->t_srtt + tp->t_rttvar;
} else {
/*
* No rtt measurement yet - use the unsmoothed rtt.
* Set the variance to half the rtt (so our first
* retransmit happens at 3*rtt).
*/
tp->t_srtt = rtt << TCP_RTT_SHIFT;
tp->t_rttvar = rtt << (TCP_RTTVAR_SHIFT - 1);
tp->t_rttbest = tp->t_srtt + tp->t_rttvar;
}
tp->t_rtttime = 0;
1994-05-24 10:09:53 +00:00
tp->t_rxtshift = 0;
/*
* the retransmit should happen at rtt + 4 * rttvar.
* Because of the way we do the smoothing, srtt and rttvar
* will each average +1/2 tick of bias. When we compute
* the retransmit timer, we want 1/2 tick of rounding and
* 1 extra tick because of +-1/2 tick uncertainty in the
* firing of the timer. The bias will give us exactly the
* 1.5 tick we need. But, because the bias is
* statistical, we have to test that we don't drop below
* the minimum feasible timer (which is 2 ticks).
*/
TCPT_RANGESET(tp->t_rxtcur, TCP_REXMTVAL(tp),
max(tp->t_rttmin, rtt + 2), TCPTV_REXMTMAX);
1995-05-30 08:16:23 +00:00
1994-05-24 10:09:53 +00:00
/*
* We received an ack for a packet that wasn't retransmitted;
* it is probably safe to discard any error indications we've
* received recently. This isn't quite right, but close enough
* for now (a route might have failed after we sent a segment,
* and the return path might not be symmetrical).
*/
tp->t_softerror = 0;
}
/*
* Determine a reasonable value for maxseg size.
* If the route is known, check route for mtu.
* If none, use an mss that can be handled on the outgoing
* interface without forcing IP to fragment; if bigger than
* an mbuf cluster (MCLBYTES), round down to nearest multiple of MCLBYTES
* to utilize large mbufs. If no route is found, route has no mtu,
* or the destination isn't local, use a default, hopefully conservative
* size (usually 512 or the default IP max size, but no more than the mtu
* of the interface), as we can't discover anything about intervening
* gateways or networks. We also initialize the congestion/slow start
* window to be a single segment if the destination isn't local.
* While looking at the routing entry, we also initialize other path-dependent
* parameters from pre-set or cached values in the routing entry.
*
* Also take into account the space needed for options that we
* send regularly. Make maxseg shorter by that amount to assure
* that we can send maxseg amount of data even when the options
* are present. Store the upper limit of the length of options plus
* data in maxopd.
*
*
* In case of T/TCP, we call this routine during implicit connection
* setup as well (offer = -1), to initialize maxseg from the cached
* MSS of our peer.
*
* NOTE that this routine is only called when we process an incoming
* segment. Outgoing SYN/ACK MSS settings are handled in tcp_mssopt().
1994-05-24 10:09:53 +00:00
*/
void
1994-05-24 10:09:53 +00:00
tcp_mss(tp, offer)
struct tcpcb *tp;
int offer;
1994-05-24 10:09:53 +00:00
{
int rtt, mss;
1994-05-24 10:09:53 +00:00
u_long bufsize;
u_long maxmtu;
struct inpcb *inp = tp->t_inpcb;
1994-05-24 10:09:53 +00:00
struct socket *so;
struct hc_metrics_lite metrics;
int origoffer = offer;
#ifdef INET6
int isipv6 = ((inp->inp_vflag & INP_IPV6) != 0) ? 1 : 0;
size_t min_protoh = isipv6 ?
sizeof (struct ip6_hdr) + sizeof (struct tcphdr) :
sizeof (struct tcpiphdr);
#else
const size_t min_protoh = sizeof(struct tcpiphdr);
#endif
/* initialize */
#ifdef INET6
if (isipv6) {
maxmtu = tcp_maxmtu6(&inp->inp_inc);
tp->t_maxopd = tp->t_maxseg = tcp_v6mssdflt;
} else
#endif
{
maxmtu = tcp_maxmtu(&inp->inp_inc);
tp->t_maxopd = tp->t_maxseg = tcp_mssdflt;
1994-05-24 10:09:53 +00:00
}
so = inp->inp_socket;
/*
* no route to sender, stay with default mss and return
*/
if (maxmtu == 0)
return;
/* what have we got? */
switch (offer) {
case 0:
/*
* Offer == 0 means that there was no MSS on the SYN
* segment, in this case we use tcp_mssdflt.
*/
offer =
#ifdef INET6
isipv6 ? tcp_v6mssdflt :
#endif
tcp_mssdflt;
break;
case -1:
/*
* Offer == -1 means that we didn't receive SYN yet.
*/
/* FALLTHROUGH */
default:
Limiters and sanity checks for TCP MSS (maximum segement size) resource exhaustion attacks. For network link optimization TCP can adjust its MSS and thus packet size according to the observed path MTU. This is done dynamically based on feedback from the remote host and network components along the packet path. This information can be abused to pretend an extremely low path MTU. The resource exhaustion works in two ways: o during tcp connection setup the advertized local MSS is exchanged between the endpoints. The remote endpoint can set this arbitrarily low (except for a minimum MTU of 64 octets enforced in the BSD code). When the local host is sending data it is forced to send many small IP packets instead of a large one. For example instead of the normal TCP payload size of 1448 it forces TCP payload size of 12 (MTU 64) and thus we have a 120 times increase in workload and packets. On fast links this quickly saturates the local CPU and may also hit pps processing limites of network components along the path. This type of attack is particularly effective for servers where the attacker can download large files (WWW and FTP). We mitigate it by enforcing a minimum MTU settable by sysctl net.inet.tcp.minmss defaulting to 256 octets. o the local host is reveiving data on a TCP connection from the remote host. The local host has no control over the packet size the remote host is sending. The remote host may chose to do what is described in the first attack and send the data in packets with an TCP payload of at least one byte. For each packet the tcp_input() function will be entered, the packet is processed and a sowakeup() is signalled to the connected process. For example an attack with 2 Mbit/s gives 4716 packets per second and the same amount of sowakeup()s to the process (and context switches). This type of attack is particularly effective for servers where the attacker can upload large amounts of data. Normally this is the case with WWW server where large POSTs can be made. We mitigate this by calculating the average MSS payload per second. If it goes below 'net.inet.tcp.minmss' and the pps rate is above 'net.inet.tcp.minmssoverload' defaulting to 1000 this particular TCP connection is resetted and dropped. MITRE CVE: CAN-2004-0002 Reviewed by: sam (mentor) MFC after: 1 day
2004-01-08 17:40:07 +00:00
/*
* Prevent DoS attack with too small MSS. Round up
* to at least minmss.
*/
offer = max(offer, tcp_minmss);
/*
* Sanity check: make sure that maxopd will be large
* enough to allow some data on segments even if the
* all the option space is used (40bytes). Otherwise
* funny things may happen in tcp_output.
*/
offer = max(offer, 64);
}
1994-05-24 10:09:53 +00:00
/*
* rmx information is now retrieved from tcp_hostcache
1994-05-24 10:09:53 +00:00
*/
tcp_hc_get(&inp->inp_inc, &metrics);
1994-05-24 10:09:53 +00:00
/*
* if there's a discovered mtu int tcp hostcache, use it
* else, use the link mtu.
1994-05-24 10:09:53 +00:00
*/
if (metrics.rmx_mtu)
mss = min(metrics.rmx_mtu, maxmtu) - min_protoh;
else {
#ifdef INET6
if (isipv6) {
mss = maxmtu - min_protoh;
if (!path_mtu_discovery &&
!in6_localaddr(&inp->in6p_faddr))
mss = min(mss, tcp_v6mssdflt);
} else
#endif
{
mss = maxmtu - min_protoh;
if (!path_mtu_discovery &&
!in_localaddr(inp->inp_faddr))
mss = min(mss, tcp_mssdflt);
}
}
mss = min(mss, offer);
/*
* maxopd stores the maximum length of data AND options
* in a segment; maxseg is the amount of data in a normal
* segment. We need to store this value (maxopd) apart
* from maxseg, because now every segment carries options
* and thus we normally have somewhat less data in segments.
*/
tp->t_maxopd = mss;
/*
* origoffer==-1 indicates, that no segments were received yet.
* In this case we just guess.
*/
if ((tp->t_flags & (TF_REQ_TSTMP|TF_NOOPT)) == TF_REQ_TSTMP &&
(origoffer == -1 ||
(tp->t_flags & TF_RCVD_TSTMP) == TF_RCVD_TSTMP))
mss -= TCPOLEN_TSTAMP_APPA;
tp->t_maxseg = mss;
1994-05-24 10:09:53 +00:00
#if (MCLBYTES & (MCLBYTES - 1)) == 0
if (mss > MCLBYTES)
mss &= ~(MCLBYTES-1);
#else
if (mss > MCLBYTES)
mss = mss / MCLBYTES * MCLBYTES;
#endif
tp->t_maxseg = mss;
1994-05-24 10:09:53 +00:00
/*
* If there's a pipesize, change the socket buffer to that size,
* don't change if sb_hiwat is different than default (then it
* has been changed on purpose with setsockopt).
* Make the socket buffers an integral number of mss units;
* if the mss is larger than the socket buffer, decrease the mss.
1994-05-24 10:09:53 +00:00
*/
SOCKBUF_LOCK(&so->so_snd);
if ((so->so_snd.sb_hiwat == tcp_sendspace) && metrics.rmx_sendpipe)
bufsize = metrics.rmx_sendpipe;
else
bufsize = so->so_snd.sb_hiwat;
if (bufsize < mss)
mss = bufsize;
else {
bufsize = roundup(bufsize, mss);
if (bufsize > sb_max)
bufsize = sb_max;
if (bufsize > so->so_snd.sb_hiwat)
(void)sbreserve_locked(&so->so_snd, bufsize, so, NULL);
}
SOCKBUF_UNLOCK(&so->so_snd);
tp->t_maxseg = mss;
1994-05-24 10:09:53 +00:00
SOCKBUF_LOCK(&so->so_rcv);
if ((so->so_rcv.sb_hiwat == tcp_recvspace) && metrics.rmx_recvpipe)
bufsize = metrics.rmx_recvpipe;
else
bufsize = so->so_rcv.sb_hiwat;
if (bufsize > mss) {
bufsize = roundup(bufsize, mss);
if (bufsize > sb_max)
bufsize = sb_max;
if (bufsize > so->so_rcv.sb_hiwat)
(void)sbreserve_locked(&so->so_rcv, bufsize, so, NULL);
1994-05-24 10:09:53 +00:00
}
SOCKBUF_UNLOCK(&so->so_rcv);
/*
* While we're here, check the others too
*/
if (tp->t_srtt == 0 && (rtt = metrics.rmx_rtt)) {
tp->t_srtt = rtt;
tp->t_rttbest = tp->t_srtt + TCP_RTT_SCALE;
tcpstat.tcps_usedrtt++;
if (metrics.rmx_rttvar) {
tp->t_rttvar = metrics.rmx_rttvar;
tcpstat.tcps_usedrttvar++;
} else {
/* default variation is +- 1 rtt */
tp->t_rttvar =
tp->t_srtt * TCP_RTTVAR_SCALE / TCP_RTT_SCALE;
}
TCPT_RANGESET(tp->t_rxtcur,
((tp->t_srtt >> 2) + tp->t_rttvar) >> 1,
tp->t_rttmin, TCPTV_REXMTMAX);
}
if (metrics.rmx_ssthresh) {
/*
* There's some sort of gateway or interface
* buffer limit on the path. Use this to set
* the slow start threshhold, but set the
* threshold to no less than 2*mss.
*/
tp->snd_ssthresh = max(2 * mss, metrics.rmx_ssthresh);
tcpstat.tcps_usedssthresh++;
}
if (metrics.rmx_bandwidth)
tp->snd_bandwidth = metrics.rmx_bandwidth;
/*
* Set the slow-start flight size depending on whether this
* is a local network or not.
*
* Extend this so we cache the cwnd too and retrieve it here.
* Make cwnd even bigger than RFC3390 suggests but only if we
* have previous experience with the remote host. Be careful
* not make cwnd bigger than remote receive window or our own
* send socket buffer. Maybe put some additional upper bound
* on the retrieved cwnd. Should do incremental updates to
* hostcache when cwnd collapses so next connection doesn't
* overloads the path again.
*
* RFC3390 says only do this if SYN or SYN/ACK didn't got lost.
* We currently check only in syncache_socket for that.
*/
#define TCP_METRICS_CWND
#ifdef TCP_METRICS_CWND
if (metrics.rmx_cwnd)
tp->snd_cwnd = max(mss,
min(metrics.rmx_cwnd / 2,
min(tp->snd_wnd, so->so_snd.sb_hiwat)));
else
#endif
if (tcp_do_rfc3390)
tp->snd_cwnd = min(4 * mss, max(2 * mss, 4380));
#ifdef INET6
else if ((isipv6 && in6_localaddr(&inp->in6p_faddr)) ||
(!isipv6 && in_localaddr(inp->inp_faddr)))
#else
else if (in_localaddr(inp->inp_faddr))
#endif
tp->snd_cwnd = mss * ss_fltsz_local;
else
tp->snd_cwnd = mss * ss_fltsz;
}
/*
* Determine the MSS option to send on an outgoing SYN.
*/
int
tcp_mssopt(inc)
struct in_conninfo *inc;
{
int mss = 0;
u_long maxmtu = 0;
u_long thcmtu = 0;
size_t min_protoh;
#ifdef INET6
int isipv6 = inc->inc_isipv6 ? 1 : 0;
#endif
KASSERT(inc != NULL, ("tcp_mssopt with NULL in_conninfo pointer"));
#ifdef INET6
if (isipv6) {
mss = tcp_v6mssdflt;
maxmtu = tcp_maxmtu6(inc);
thcmtu = tcp_hc_getmtu(inc); /* IPv4 and IPv6 */
min_protoh = sizeof(struct ip6_hdr) + sizeof(struct tcphdr);
} else
#endif
{
mss = tcp_mssdflt;
maxmtu = tcp_maxmtu(inc);
thcmtu = tcp_hc_getmtu(inc); /* IPv4 and IPv6 */
min_protoh = sizeof(struct tcpiphdr);
}
if (maxmtu && thcmtu)
mss = min(maxmtu, thcmtu) - min_protoh;
else if (maxmtu || thcmtu)
mss = max(maxmtu, thcmtu) - min_protoh;
return (mss);
1994-05-24 10:09:53 +00:00
}
/*
* On a partial ack arrives, force the retransmission of the
* next unacknowledged segment. Do not clear tp->t_dupacks.
* By setting snd_nxt to ti_ack, this forces retransmission timer to
* be started again.
*/
static void
tcp_newreno_partial_ack(tp, th)
struct tcpcb *tp;
struct tcphdr *th;
{
tcp_seq onxt = tp->snd_nxt;
u_long ocwnd = tp->snd_cwnd;
callout_stop(tp->tt_rexmt);
tp->t_rtttime = 0;
tp->snd_nxt = th->th_ack;
/*
* Set snd_cwnd to one segment beyond acknowledged offset.
* (tp->snd_una has not yet been updated when this function is called.)
*/
tp->snd_cwnd = tp->t_maxseg + (th->th_ack - tp->snd_una);
tp->t_flags |= TF_ACKNOW;
(void) tcp_output(tp);
tp->snd_cwnd = ocwnd;
if (SEQ_GT(onxt, tp->snd_nxt))
tp->snd_nxt = onxt;
/*
* Partial window deflation. Relies on fact that tp->snd_una
* not updated yet.
*/
tp->snd_cwnd -= (th->th_ack - tp->snd_una - tp->t_maxseg);
}
/*
* Returns 1 if the TIME_WAIT state was killed and we should start over,
* looking for a pcb in the listen state. Returns 0 otherwise.
*/
static int
tcp_timewait(tw, to, th, m, tlen)
struct tcptw *tw;
struct tcpopt *to;
struct tcphdr *th;
struct mbuf *m;
int tlen;
{
int thflags;
tcp_seq seq;
#ifdef INET6
int isipv6 = (mtod(m, struct ip *)->ip_v == 6) ? 1 : 0;
#else
const int isipv6 = 0;
#endif
thflags = th->th_flags;
/*
* NOTE: for FIN_WAIT_2 (to be added later),
* must validate sequence number before accepting RST
*/
/*
* If the segment contains RST:
* Drop the segment - see Stevens, vol. 2, p. 964 and
* RFC 1337.
*/
if (thflags & TH_RST)
goto drop;
#if 0
/* PAWS not needed at the moment */
/*
* RFC 1323 PAWS: If we have a timestamp reply on this segment
* and it's less than ts_recent, drop it.
*/
if ((to.to_flags & TOF_TS) != 0 && tp->ts_recent &&
TSTMP_LT(to.to_tsval, tp->ts_recent)) {
if ((thflags & TH_ACK) == 0)
goto drop;
goto ack;
}
/*
* ts_recent is never updated because we never accept new segments.
*/
#endif
/*
* 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 ((thflags & TH_SYN) && SEQ_GT(th->th_seq, tw->rcv_nxt)) {
(void) tcp_twclose(tw, 0);
return (1);
}
/*
* Drop the the segment if it does not contain an ACK.
*/
if ((thflags & TH_ACK) == 0)
goto drop;
/*
* Reset the 2MSL timer if this is a duplicate FIN.
*/
if (thflags & TH_FIN) {
seq = th->th_seq + tlen + (thflags & TH_SYN ? 1 : 0);
if (seq + 1 == tw->rcv_nxt)
tcp_timer_2msl_reset(tw, 2 * tcp_msl);
}
/*
* Acknowledge the segment if it has data or is not a duplicate ACK.
*/
if (thflags != TH_ACK || tlen != 0 ||
th->th_seq != tw->rcv_nxt || th->th_ack != tw->snd_nxt)
tcp_twrespond(tw, TH_ACK);
goto drop;
/*
* Generate a RST, dropping incoming segment.
* Make ACK acceptable to originator of segment.
* Don't bother to respond if destination was broadcast/multicast.
*/
if (m->m_flags & (M_BCAST|M_MCAST))
goto drop;
if (isipv6) {
struct ip6_hdr *ip6;
/* IPv6 anycast check is done at tcp6_input() */
ip6 = mtod(m, struct ip6_hdr *);
if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst) ||
IN6_IS_ADDR_MULTICAST(&ip6->ip6_src))
goto drop;
} else {
struct ip *ip;
ip = mtod(m, struct ip *);
if (IN_MULTICAST(ntohl(ip->ip_dst.s_addr)) ||
IN_MULTICAST(ntohl(ip->ip_src.s_addr)) ||
ip->ip_src.s_addr == htonl(INADDR_BROADCAST) ||
in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif))
goto drop;
}
if (thflags & TH_ACK) {
tcp_respond(NULL,
mtod(m, void *), th, m, 0, th->th_ack, TH_RST);
} else {
seq = th->th_seq + (thflags & TH_SYN ? 1 : 0);
tcp_respond(NULL,
mtod(m, void *), th, m, seq, 0, TH_RST|TH_ACK);
}
INP_UNLOCK(tw->tw_inpcb);
return (0);
drop:
INP_UNLOCK(tw->tw_inpcb);
m_freem(m);
return (0);
}