freebsd-dev/sys/netinet/tcp_usrreq.c

1119 lines
27 KiB
C
Raw Normal View History

1994-05-24 10:09:53 +00:00
/*
* Copyright (c) 1982, 1986, 1988, 1993
* 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.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 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.
*
* From: @(#)tcp_usrreq.c 8.2 (Berkeley) 1/3/94
1999-08-28 01:08:13 +00:00
* $FreeBSD$
1994-05-24 10:09:53 +00:00
*/
#include "opt_ipsec.h"
#include "opt_inet6.h"
1997-09-16 18:36:06 +00:00
#include "opt_tcpdebug.h"
1994-05-24 10:09:53 +00:00
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/sysctl.h>
1994-05-24 10:09:53 +00:00
#include <sys/mbuf.h>
#ifdef INET6
#include <sys/domain.h>
#endif /* INET6 */
1994-05-24 10:09:53 +00:00
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/protosw.h>
#include <net/if.h>
#include <net/route.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#ifdef INET6
#include <netinet/ip6.h>
#endif
1994-05-24 10:09:53 +00:00
#include <netinet/in_pcb.h>
#ifdef INET6
#include <netinet6/in6_pcb.h>
#endif
#include <netinet/in_var.h>
1994-05-24 10:09:53 +00:00
#include <netinet/ip_var.h>
#ifdef INET6
#include <netinet6/ip6_var.h>
#endif
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 <netinet/tcpip.h>
#ifdef TCPDEBUG
1994-05-24 10:09:53 +00:00
#include <netinet/tcp_debug.h>
#endif
1994-05-24 10:09:53 +00:00
#ifdef IPSEC
#include <netinet6/ipsec.h>
#endif /*IPSEC*/
1994-05-24 10:09:53 +00:00
/*
* TCP protocol interface to socket abstraction.
*/
extern char *tcpstates[]; /* XXX ??? */
1994-05-24 10:09:53 +00:00
static int tcp_attach __P((struct socket *, struct proc *));
static int tcp_connect __P((struct tcpcb *, struct sockaddr *,
struct proc *));
#ifdef INET6
static int tcp6_connect __P((struct tcpcb *, struct sockaddr *,
struct proc *));
#endif /* INET6 */
static struct tcpcb *
tcp_disconnect __P((struct tcpcb *));
static struct tcpcb *
tcp_usrclosed __P((struct tcpcb *));
#ifdef TCPDEBUG
#define TCPDEBUG0 int ostate
#define TCPDEBUG1() ostate = tp ? tp->t_state : 0
#define TCPDEBUG2(req) if (tp && (so->so_options & SO_DEBUG)) \
tcp_trace(TA_USER, ostate, tp, 0, 0, req)
#else
#define TCPDEBUG0
#define TCPDEBUG1()
#define TCPDEBUG2(req)
#endif
/*
* TCP attaches to socket via pru_attach(), reserving space,
* and an internet control block.
*/
static int
tcp_usr_attach(struct socket *so, int proto, struct proc *p)
{
int s = splnet();
int error;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp = 0;
TCPDEBUG0;
TCPDEBUG1();
if (inp) {
error = EISCONN;
goto out;
}
error = tcp_attach(so, p);
if (error)
goto out;
if ((so->so_options & SO_LINGER) && so->so_linger == 0)
so->so_linger = TCP_LINGERTIME;
tp = sototcpcb(so);
out:
TCPDEBUG2(PRU_ATTACH);
splx(s);
return error;
}
/*
* pru_detach() detaches the TCP protocol from the socket.
* If the protocol state is non-embryonic, then can't
* do this directly: have to initiate a pru_disconnect(),
* which may finish later; embryonic TCB's can just
* be discarded here.
*/
static int
tcp_usr_detach(struct socket *so)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
TCPDEBUG0;
if (inp == 0) {
splx(s);
return EINVAL; /* XXX */
}
tp = intotcpcb(inp);
TCPDEBUG1();
tp = tcp_disconnect(tp);
TCPDEBUG2(PRU_DETACH);
splx(s);
return error;
}
#define COMMON_START() TCPDEBUG0; \
do { \
if (inp == 0) { \
splx(s); \
return EINVAL; \
} \
tp = intotcpcb(inp); \
TCPDEBUG1(); \
} while(0)
#define COMMON_END(req) out: TCPDEBUG2(req); splx(s); return error; goto out
/*
* Give the socket an address.
*/
static int
tcp_usr_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
struct sockaddr_in *sinp;
COMMON_START();
/*
* Must check for multicast addresses and disallow binding
* to them.
*/
sinp = (struct sockaddr_in *)nam;
if (sinp->sin_family == AF_INET &&
IN_MULTICAST(ntohl(sinp->sin_addr.s_addr))) {
error = EAFNOSUPPORT;
goto out;
}
error = in_pcbbind(inp, nam, p);
if (error)
goto out;
COMMON_END(PRU_BIND);
}
#ifdef INET6
static int
tcp6_usr_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
struct sockaddr_in6 *sin6p;
COMMON_START();
/*
* Must check for multicast addresses and disallow binding
* to them.
*/
sin6p = (struct sockaddr_in6 *)nam;
if (sin6p->sin6_family == AF_INET6 &&
IN6_IS_ADDR_MULTICAST(&sin6p->sin6_addr)) {
error = EAFNOSUPPORT;
goto out;
}
inp->inp_vflag &= ~INP_IPV4;
inp->inp_vflag |= INP_IPV6;
if ((inp->inp_flags & IN6P_BINDV6ONLY) == 0) {
if (IN6_IS_ADDR_UNSPECIFIED(&sin6p->sin6_addr))
inp->inp_vflag |= INP_IPV4;
else if (IN6_IS_ADDR_V4MAPPED(&sin6p->sin6_addr)) {
struct sockaddr_in sin;
in6_sin6_2_sin(&sin, sin6p);
inp->inp_vflag |= INP_IPV4;
inp->inp_vflag &= ~INP_IPV6;
error = in_pcbbind(inp, (struct sockaddr *)&sin, p);
goto out;
}
}
error = in6_pcbbind(inp, nam, p);
if (error)
goto out;
COMMON_END(PRU_BIND);
}
#endif /* INET6 */
/*
* Prepare to accept connections.
*/
static int
tcp_usr_listen(struct socket *so, struct proc *p)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
COMMON_START();
if (inp->inp_lport == 0)
error = in_pcbbind(inp, (struct sockaddr *)0, p);
if (error == 0)
tp->t_state = TCPS_LISTEN;
COMMON_END(PRU_LISTEN);
}
#ifdef INET6
static int
tcp6_usr_listen(struct socket *so, struct proc *p)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
COMMON_START();
if (inp->inp_lport == 0) {
inp->inp_vflag &= ~INP_IPV4;
if ((inp->inp_flags & IN6P_BINDV6ONLY) == 0)
inp->inp_vflag |= INP_IPV4;
error = in6_pcbbind(inp, (struct sockaddr *)0, p);
}
if (error == 0)
tp->t_state = TCPS_LISTEN;
COMMON_END(PRU_LISTEN);
}
#endif /* INET6 */
/*
* Initiate connection to peer.
* Create a template for use in transmissions on this connection.
* Enter SYN_SENT state, and mark socket as connecting.
* Start keep-alive timer, and seed output sequence space.
* Send initial segment on connection.
*/
static int
tcp_usr_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
struct sockaddr_in *sinp;
COMMON_START();
/*
* Must disallow TCP ``connections'' to multicast addresses.
*/
sinp = (struct sockaddr_in *)nam;
if (sinp->sin_family == AF_INET
&& IN_MULTICAST(ntohl(sinp->sin_addr.s_addr))) {
error = EAFNOSUPPORT;
goto out;
}
This Implements the mumbled about "Jail" feature. This is a seriously beefed up chroot kind of thing. The process is jailed along the same lines as a chroot does it, but with additional tough restrictions imposed on what the superuser can do. For all I know, it is safe to hand over the root bit inside a prison to the customer living in that prison, this is what it was developed for in fact: "real virtual servers". Each prison has an ip number associated with it, which all IP communications will be coerced to use and each prison has its own hostname. Needless to say, you need more RAM this way, but the advantage is that each customer can run their own particular version of apache and not stomp on the toes of their neighbors. It generally does what one would expect, but setting up a jail still takes a little knowledge. A few notes: I have no scripts for setting up a jail, don't ask me for them. The IP number should be an alias on one of the interfaces. mount a /proc in each jail, it will make ps more useable. /proc/<pid>/status tells the hostname of the prison for jailed processes. Quotas are only sensible if you have a mountpoint per prison. There are no privisions for stopping resource-hogging. Some "#ifdef INET" and similar may be missing (send patches!) If somebody wants to take it from here and develop it into more of a "virtual machine" they should be most welcome! Tools, comments, patches & documentation most welcome. Have fun... Sponsored by: http://www.rndassociates.com/ Run for almost a year by: http://www.servetheweb.com/
1999-04-28 11:38:52 +00:00
prison_remote_ip(p, 0, &sinp->sin_addr.s_addr);
if ((error = tcp_connect(tp, nam, p)) != 0)
goto out;
error = tcp_output(tp);
COMMON_END(PRU_CONNECT);
}
#ifdef INET6
static int
tcp6_usr_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
struct sockaddr_in6 *sin6p;
COMMON_START();
/*
* Must disallow TCP ``connections'' to multicast addresses.
*/
sin6p = (struct sockaddr_in6 *)nam;
if (sin6p->sin6_family == AF_INET6
&& IN6_IS_ADDR_MULTICAST(&sin6p->sin6_addr)) {
error = EAFNOSUPPORT;
goto out;
}
if ((inp->inp_flags & IN6P_BINDV6ONLY) == 0 &&
IN6_IS_ADDR_V4MAPPED(&sin6p->sin6_addr)) {
struct sockaddr_in sin;
in6_sin6_2_sin(&sin, sin6p);
inp->inp_vflag |= INP_IPV4;
inp->inp_vflag &= ~INP_IPV6;
if ((error = tcp_connect(tp, (struct sockaddr *)&sin, p)) != 0)
goto out;
error = tcp_output(tp);
goto out;
}
inp->inp_vflag &= ~INP_IPV4;
inp->inp_vflag |= INP_IPV6;
if ((error = tcp6_connect(tp, nam, p)) != 0)
goto out;
error = tcp_output(tp);
COMMON_END(PRU_CONNECT);
}
#endif /* INET6 */
/*
* Initiate disconnect from peer.
* If connection never passed embryonic stage, just drop;
* else if don't need to let data drain, then can just drop anyways,
* else have to begin TCP shutdown process: mark socket disconnecting,
* drain unread data, state switch to reflect user close, and
* send segment (e.g. FIN) to peer. Socket will be really disconnected
* when peer sends FIN and acks ours.
*
* SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB.
*/
static int
tcp_usr_disconnect(struct socket *so)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
COMMON_START();
tp = tcp_disconnect(tp);
COMMON_END(PRU_DISCONNECT);
}
/*
* Accept a connection. Essentially all the work is
* done at higher levels; just return the address
* of the peer, storing through addr.
*/
static int
tcp_usr_accept(struct socket *so, struct sockaddr **nam)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
COMMON_START();
in_setpeeraddr(so, nam);
COMMON_END(PRU_ACCEPT);
}
#ifdef INET6
static int
tcp6_usr_accept(struct socket *so, struct sockaddr **nam)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
COMMON_START();
in6_mapped_peeraddr(so, nam);
COMMON_END(PRU_ACCEPT);
}
#endif /* INET6 */
/*
* Mark the connection as being incapable of further output.
*/
static int
tcp_usr_shutdown(struct socket *so)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
COMMON_START();
socantsendmore(so);
tp = tcp_usrclosed(tp);
if (tp)
error = tcp_output(tp);
COMMON_END(PRU_SHUTDOWN);
}
/*
* After a receive, possibly send window update to peer.
*/
static int
tcp_usr_rcvd(struct socket *so, int flags)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
COMMON_START();
tcp_output(tp);
COMMON_END(PRU_RCVD);
}
/*
* Do a send by putting data in output queue and updating urgent
* marker if URG set. Possibly send more data. Unlike the other
* pru_*() routines, the mbuf chains are our responsibility. We
* must either enqueue them or free them. The other pru_* routines
* generally are caller-frees.
*/
static int
tcp_usr_send(struct socket *so, int flags, struct mbuf *m,
struct sockaddr *nam, struct mbuf *control, struct proc *p)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
#ifdef INET6
int isipv6;
#endif
TCPDEBUG0;
if (inp == NULL) {
/*
* OOPS! we lost a race, the TCP session got reset after
* we checked SS_CANTSENDMORE, eg: while doing uiomove or a
* network interrupt in the non-splnet() section of sosend().
*/
if (m)
m_freem(m);
if (control)
m_freem(control);
error = ECONNRESET; /* XXX EPIPE? */
tp = NULL;
TCPDEBUG1();
goto out;
}
#ifdef INET6
isipv6 = nam && nam->sa_family == AF_INET6;
#endif /* INET6 */
tp = intotcpcb(inp);
TCPDEBUG1();
if (control) {
/* TCP doesn't do control messages (rights, creds, etc) */
if (control->m_len) {
m_freem(control);
if (m)
m_freem(m);
error = EINVAL;
goto out;
}
m_freem(control); /* empty control, just free it */
}
if(!(flags & PRUS_OOB)) {
sbappend(&so->so_snd, m);
if (nam && tp->t_state < TCPS_SYN_SENT) {
/*
* Do implied connect if not yet connected,
* initialize window to default value, and
* initialize maxseg/maxopd using peer's cached
* MSS.
*/
#ifdef INET6
if (isipv6)
error = tcp6_connect(tp, nam, p);
else
#endif /* INET6 */
error = tcp_connect(tp, nam, p);
if (error)
goto out;
tp->snd_wnd = TTCP_CLIENT_SND_WND;
tcp_mss(tp, -1);
}
if (flags & PRUS_EOF) {
/*
* Close the send side of the connection after
* the data is sent.
*/
socantsendmore(so);
tp = tcp_usrclosed(tp);
}
if (tp != NULL) {
if (flags & PRUS_MORETOCOME)
tp->t_flags |= TF_MORETOCOME;
error = tcp_output(tp);
if (flags & PRUS_MORETOCOME)
tp->t_flags &= ~TF_MORETOCOME;
}
} else {
if (sbspace(&so->so_snd) < -512) {
m_freem(m);
error = ENOBUFS;
goto out;
}
/*
* According to RFC961 (Assigned Protocols),
* the urgent pointer points to the last octet
* of urgent data. We continue, however,
* to consider it to indicate the first octet
* of data past the urgent section.
* Otherwise, snd_up should be one lower.
*/
sbappend(&so->so_snd, m);
if (nam && tp->t_state < TCPS_SYN_SENT) {
/*
* Do implied connect if not yet connected,
* initialize window to default value, and
* initialize maxseg/maxopd using peer's cached
* MSS.
*/
#ifdef INET6
if (isipv6)
error = tcp6_connect(tp, nam, p);
else
#endif /* INET6 */
error = tcp_connect(tp, nam, p);
if (error)
goto out;
tp->snd_wnd = TTCP_CLIENT_SND_WND;
tcp_mss(tp, -1);
}
tp->snd_up = tp->snd_una + so->so_snd.sb_cc;
tp->t_force = 1;
error = tcp_output(tp);
tp->t_force = 0;
}
COMMON_END((flags & PRUS_OOB) ? PRU_SENDOOB :
((flags & PRUS_EOF) ? PRU_SEND_EOF : PRU_SEND));
}
/*
* Abort the TCP.
*/
static int
tcp_usr_abort(struct socket *so)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
COMMON_START();
tp = tcp_drop(tp, ECONNABORTED);
COMMON_END(PRU_ABORT);
}
/*
* Receive out-of-band data.
*/
static int
tcp_usr_rcvoob(struct socket *so, struct mbuf *m, int flags)
{
int s = splnet();
int error = 0;
struct inpcb *inp = sotoinpcb(so);
struct tcpcb *tp;
COMMON_START();
if ((so->so_oobmark == 0 &&
(so->so_state & SS_RCVATMARK) == 0) ||
so->so_options & SO_OOBINLINE ||
tp->t_oobflags & TCPOOB_HADDATA) {
error = EINVAL;
goto out;
}
if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) {
error = EWOULDBLOCK;
goto out;
}
m->m_len = 1;
*mtod(m, caddr_t) = tp->t_iobc;
if ((flags & MSG_PEEK) == 0)
tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA);
COMMON_END(PRU_RCVOOB);
}
/* xxx - should be const */
struct pr_usrreqs tcp_usrreqs = {
tcp_usr_abort, tcp_usr_accept, tcp_usr_attach, tcp_usr_bind,
tcp_usr_connect, pru_connect2_notsupp, in_control, tcp_usr_detach,
tcp_usr_disconnect, tcp_usr_listen, in_setpeeraddr, tcp_usr_rcvd,
tcp_usr_rcvoob, tcp_usr_send, pru_sense_null, tcp_usr_shutdown,
in_setsockaddr, sosend, soreceive, sopoll
};
1994-05-24 10:09:53 +00:00
#ifdef INET6
struct pr_usrreqs tcp6_usrreqs = {
tcp_usr_abort, tcp6_usr_accept, tcp_usr_attach, tcp6_usr_bind,
tcp6_usr_connect, pru_connect2_notsupp, in6_control, tcp_usr_detach,
tcp_usr_disconnect, tcp6_usr_listen, in6_mapped_peeraddr, tcp_usr_rcvd,
tcp_usr_rcvoob, tcp_usr_send, pru_sense_null, tcp_usr_shutdown,
in6_mapped_sockaddr, sosend, soreceive, sopoll
};
#endif /* INET6 */
/*
* Common subroutine to open a TCP connection to remote host specified
* by struct sockaddr_in in mbuf *nam. Call in_pcbbind to assign a local
* port number if needed. Call in_pcbladdr to do the routing and to choose
* a local host address (interface). If there is an existing incarnation
* of the same connection in TIME-WAIT state and if the remote host was
* sending CC options and if the connection duration was < MSL, then
* truncate the previous TIME-WAIT state and proceed.
* Initialize connection parameters and enter SYN-SENT state.
*/
static int
tcp_connect(tp, nam, p)
register struct tcpcb *tp;
struct sockaddr *nam;
struct proc *p;
{
struct inpcb *inp = tp->t_inpcb, *oinp;
struct socket *so = inp->inp_socket;
struct tcpcb *otp;
struct sockaddr_in *sin = (struct sockaddr_in *)nam;
struct sockaddr_in *ifaddr;
struct rmxp_tao *taop;
struct rmxp_tao tao_noncached;
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
int error;
if (inp->inp_lport == 0) {
error = in_pcbbind(inp, (struct sockaddr *)0, p);
if (error)
return error;
}
1995-05-30 08:16:23 +00:00
/*
* Cannot simply call in_pcbconnect, because there might be an
* earlier incarnation of this same connection still in
* TIME_WAIT state, creating an ADDRINUSE error.
*/
error = in_pcbladdr(inp, nam, &ifaddr);
1995-06-11 19:33:05 +00:00
if (error)
return error;
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
oinp = in_pcblookup_hash(inp->inp_pcbinfo,
sin->sin_addr, sin->sin_port,
inp->inp_laddr.s_addr != INADDR_ANY ? inp->inp_laddr
: ifaddr->sin_addr,
inp->inp_lport, 0, NULL);
if (oinp) {
if (oinp != inp && (otp = intotcpcb(oinp)) != NULL &&
otp->t_state == TCPS_TIME_WAIT &&
(ticks - otp->t_starttime) < tcp_msl &&
(otp->t_flags & TF_RCVD_CC))
otp = tcp_close(otp);
else
return EADDRINUSE;
}
if (inp->inp_laddr.s_addr == INADDR_ANY)
inp->inp_laddr = ifaddr->sin_addr;
inp->inp_faddr = sin->sin_addr;
inp->inp_fport = sin->sin_port;
in_pcbrehash(inp);
tp->t_template = tcp_template(tp);
if (tp->t_template == 0) {
in_pcbdisconnect(inp);
return ENOBUFS;
}
/* Compute window scaling to request. */
while (tp->request_r_scale < TCP_MAX_WINSHIFT &&
(TCP_MAXWIN << tp->request_r_scale) < so->so_rcv.sb_hiwat)
tp->request_r_scale++;
soisconnecting(so);
tcpstat.tcps_connattempt++;
tp->t_state = TCPS_SYN_SENT;
callout_reset(tp->tt_keep, tcp_keepinit, tcp_timer_keep, tp);
tp->iss = tcp_iss; tcp_iss += TCP_ISSINCR/2;
tcp_sendseqinit(tp);
/*
* Generate a CC value for this connection and
* check whether CC or CCnew should be used.
*/
if ((taop = tcp_gettaocache(tp->t_inpcb)) == NULL) {
taop = &tao_noncached;
bzero(taop, sizeof(*taop));
}
tp->cc_send = CC_INC(tcp_ccgen);
if (taop->tao_ccsent != 0 &&
CC_GEQ(tp->cc_send, taop->tao_ccsent)) {
taop->tao_ccsent = tp->cc_send;
} else {
taop->tao_ccsent = 0;
tp->t_flags |= TF_SENDCCNEW;
}
return 0;
}
#ifdef INET6
static int
tcp6_connect(tp, nam, p)
register struct tcpcb *tp;
struct sockaddr *nam;
struct proc *p;
{
struct inpcb *inp = tp->t_inpcb, *oinp;
struct socket *so = inp->inp_socket;
struct tcpcb *otp;
struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam;
struct in6_addr *addr6;
struct rmxp_tao *taop;
struct rmxp_tao tao_noncached;
int error;
if (inp->inp_lport == 0) {
error = in6_pcbbind(inp, (struct sockaddr *)0, p);
if (error)
return error;
}
/*
* Cannot simply call in_pcbconnect, because there might be an
* earlier incarnation of this same connection still in
* TIME_WAIT state, creating an ADDRINUSE error.
*/
error = in6_pcbladdr(inp, nam, &addr6);
if (error)
return error;
oinp = in6_pcblookup_hash(inp->inp_pcbinfo,
&sin6->sin6_addr, sin6->sin6_port,
IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)
? addr6
: &inp->in6p_laddr,
inp->inp_lport, 0, NULL);
if (oinp) {
if (oinp != inp && (otp = intotcpcb(oinp)) != NULL &&
otp->t_state == TCPS_TIME_WAIT &&
(ticks - otp->t_starttime) < tcp_msl &&
(otp->t_flags & TF_RCVD_CC))
otp = tcp_close(otp);
else
return EADDRINUSE;
}
if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr))
inp->in6p_laddr = *addr6;
inp->in6p_faddr = sin6->sin6_addr;
inp->inp_fport = sin6->sin6_port;
if ((sin6->sin6_flowinfo & IPV6_FLOWINFO_MASK) != NULL)
inp->in6p_flowinfo = sin6->sin6_flowinfo;
in_pcbrehash(inp);
tp->t_template = tcp_template(tp);
if (tp->t_template == 0) {
in6_pcbdisconnect(inp);
return ENOBUFS;
}
/* Compute window scaling to request. */
while (tp->request_r_scale < TCP_MAX_WINSHIFT &&
(TCP_MAXWIN << tp->request_r_scale) < so->so_rcv.sb_hiwat)
tp->request_r_scale++;
soisconnecting(so);
tcpstat.tcps_connattempt++;
tp->t_state = TCPS_SYN_SENT;
callout_reset(tp->tt_keep, tcp_keepinit, tcp_timer_keep, tp);
tp->iss = tcp_iss; tcp_iss += TCP_ISSINCR/2;
tcp_sendseqinit(tp);
/*
* Generate a CC value for this connection and
* check whether CC or CCnew should be used.
*/
if ((taop = tcp_gettaocache(tp->t_inpcb)) == NULL) {
taop = &tao_noncached;
bzero(taop, sizeof(*taop));
}
tp->cc_send = CC_INC(tcp_ccgen);
if (taop->tao_ccsent != 0 &&
CC_GEQ(tp->cc_send, taop->tao_ccsent)) {
taop->tao_ccsent = tp->cc_send;
} else {
taop->tao_ccsent = 0;
tp->t_flags |= TF_SENDCCNEW;
}
return 0;
}
#endif /* INET6 */
/*
* The new sockopt interface makes it possible for us to block in the
* copyin/out step (if we take a page fault). Taking a page fault at
* splnet() is probably a Bad Thing. (Since sockets and pcbs both now
* use TSM, there probably isn't any need for this function to run at
* splnet() any more. This needs more examination.)
*/
1994-05-24 10:09:53 +00:00
int
tcp_ctloutput(so, sopt)
1994-05-24 10:09:53 +00:00
struct socket *so;
struct sockopt *sopt;
1994-05-24 10:09:53 +00:00
{
int error, opt, optval, s;
struct inpcb *inp;
struct tcpcb *tp;
1994-05-24 10:09:53 +00:00
error = 0;
s = splnet(); /* XXX */
1994-05-24 10:09:53 +00:00
inp = sotoinpcb(so);
if (inp == NULL) {
splx(s);
return (ECONNRESET);
}
if (sopt->sopt_level != IPPROTO_TCP) {
#ifdef INET6
if (INP_CHECK_SOCKAF(so, AF_INET6))
error = ip6_ctloutput(so, sopt);
else
#endif /* INET6 */
error = ip_ctloutput(so, sopt);
1994-05-24 10:09:53 +00:00
splx(s);
return (error);
}
tp = intotcpcb(inp);
switch (sopt->sopt_dir) {
case SOPT_SET:
switch (sopt->sopt_name) {
1994-05-24 10:09:53 +00:00
case TCP_NODELAY:
case TCP_NOOPT:
case TCP_NOPUSH:
error = sooptcopyin(sopt, &optval, sizeof optval,
sizeof optval);
if (error)
break;
switch (sopt->sopt_name) {
case TCP_NODELAY:
opt = TF_NODELAY;
break;
case TCP_NOOPT:
opt = TF_NOOPT;
break;
case TCP_NOPUSH:
opt = TF_NOPUSH;
break;
default:
opt = 0; /* dead code to fool gcc */
break;
}
if (optval)
tp->t_flags |= opt;
1994-05-24 10:09:53 +00:00
else
tp->t_flags &= ~opt;
1994-05-24 10:09:53 +00:00
break;
case TCP_MAXSEG:
error = sooptcopyin(sopt, &optval, sizeof optval,
sizeof optval);
if (error)
break;
1994-05-24 10:09:53 +00:00
if (optval > 0 && optval <= tp->t_maxseg)
tp->t_maxseg = optval;
else
error = EINVAL;
break;
1994-05-24 10:09:53 +00:00
default:
error = ENOPROTOOPT;
break;
}
break;
case SOPT_GET:
switch (sopt->sopt_name) {
1994-05-24 10:09:53 +00:00
case TCP_NODELAY:
optval = tp->t_flags & TF_NODELAY;
1994-05-24 10:09:53 +00:00
break;
case TCP_MAXSEG:
optval = tp->t_maxseg;
1994-05-24 10:09:53 +00:00
break;
case TCP_NOOPT:
optval = tp->t_flags & TF_NOOPT;
break;
case TCP_NOPUSH:
optval = tp->t_flags & TF_NOPUSH;
break;
1994-05-24 10:09:53 +00:00
default:
error = ENOPROTOOPT;
break;
}
if (error == 0)
error = sooptcopyout(sopt, &optval, sizeof optval);
1994-05-24 10:09:53 +00:00
break;
}
splx(s);
return (error);
}
/*
* tcp_sendspace and tcp_recvspace are the default send and receive window
* sizes, respectively. These are obsolescent (this information should
* be set by the route).
*/
u_long tcp_sendspace = 1024*16;
SYSCTL_INT(_net_inet_tcp, TCPCTL_SENDSPACE, sendspace, CTLFLAG_RW,
&tcp_sendspace , 0, "Maximum outgoing TCP datagram size");
u_long tcp_recvspace = 1024*16;
SYSCTL_INT(_net_inet_tcp, TCPCTL_RECVSPACE, recvspace, CTLFLAG_RW,
&tcp_recvspace , 0, "Maximum incoming TCP datagram size");
1994-05-24 10:09:53 +00:00
/*
* Attach TCP protocol to socket, allocating
* internet protocol control block, tcp control block,
* bufer space, and entering LISTEN state if to accept connections.
*/
static int
tcp_attach(so, p)
1994-05-24 10:09:53 +00:00
struct socket *so;
struct proc *p;
1994-05-24 10:09:53 +00:00
{
register struct tcpcb *tp;
struct inpcb *inp;
int error;
#ifdef INET6
int isipv6 = INP_CHECK_SOCKAF(so, AF_INET6) != NULL;
#endif
1994-05-24 10:09:53 +00:00
if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
error = soreserve(so, tcp_sendspace, tcp_recvspace);
if (error)
return (error);
}
error = in_pcballoc(so, &tcbinfo, p);
1994-05-24 10:09:53 +00:00
if (error)
return (error);
inp = sotoinpcb(so);
#ifdef IPSEC
error = ipsec_init_policy(so, &inp->inp_sp);
if (error) {
#ifdef INET6
if (isipv6)
in6_pcbdetach(inp);
else
#endif
in_pcbdetach(inp);
return (error);
}
#endif /*IPSEC*/
#ifdef INET6
if (isipv6) {
inp->inp_vflag |= INP_IPV6;
inp->in6p_hops = -1; /* use kernel default */
}
else
#endif
inp->inp_vflag |= INP_IPV4;
1994-05-24 10:09:53 +00:00
tp = tcp_newtcpcb(inp);
if (tp == 0) {
int nofd = so->so_state & SS_NOFDREF; /* XXX */
so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */
#ifdef INET6
if (isipv6)
in6_pcbdetach(inp);
else
#endif
1994-05-24 10:09:53 +00:00
in_pcbdetach(inp);
so->so_state |= nofd;
return (ENOBUFS);
}
tp->t_state = TCPS_CLOSED;
return (0);
}
/*
* Initiate (or continue) disconnect.
* If embryonic state, just send reset (once).
* If in ``let data drain'' option and linger null, just drop.
* Otherwise (hard), mark socket disconnecting and drop
* current input data; switch states based on user close, and
* send segment to peer (with FIN).
*/
static struct tcpcb *
1994-05-24 10:09:53 +00:00
tcp_disconnect(tp)
register struct tcpcb *tp;
{
struct socket *so = tp->t_inpcb->inp_socket;
if (tp->t_state < TCPS_ESTABLISHED)
tp = tcp_close(tp);
else if ((so->so_options & SO_LINGER) && so->so_linger == 0)
tp = tcp_drop(tp, 0);
else {
soisdisconnecting(so);
sbflush(&so->so_rcv);
tp = tcp_usrclosed(tp);
if (tp)
(void) tcp_output(tp);
}
return (tp);
}
/*
* User issued close, and wish to trail through shutdown states:
* if never received SYN, just forget it. If got a SYN from peer,
* but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN.
* If already got a FIN from peer, then almost done; go to LAST_ACK
* state. In all other cases, have already sent FIN to peer (e.g.
* after PRU_SHUTDOWN), and just have to play tedious game waiting
* for peer to send FIN or not respond to keep-alives, etc.
* We can let the user exit from the close as soon as the FIN is acked.
*/
static struct tcpcb *
1994-05-24 10:09:53 +00:00
tcp_usrclosed(tp)
register struct tcpcb *tp;
{
switch (tp->t_state) {
case TCPS_CLOSED:
case TCPS_LISTEN:
tp->t_state = TCPS_CLOSED;
tp = tcp_close(tp);
break;
case TCPS_SYN_SENT:
case TCPS_SYN_RECEIVED:
tp->t_flags |= TF_NEEDFIN;
break;
1994-05-24 10:09:53 +00:00
case TCPS_ESTABLISHED:
tp->t_state = TCPS_FIN_WAIT_1;
break;
case TCPS_CLOSE_WAIT:
tp->t_state = TCPS_LAST_ACK;
break;
}
if (tp && tp->t_state >= TCPS_FIN_WAIT_2) {
1994-05-24 10:09:53 +00:00
soisdisconnected(tp->t_inpcb->inp_socket);
/* To prevent the connection hanging in FIN_WAIT_2 forever. */
if (tp->t_state == TCPS_FIN_WAIT_2)
callout_reset(tp->tt_2msl, tcp_maxidle,
tcp_timer_2msl, tp);
}
1994-05-24 10:09:53 +00:00
return (tp);
}