d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
fa046d8774
freebsd-dev/sys/netinet6/udp6_usrreq.c
Robert Watson fa046d8774 Decompose the current single inpcbinfo lock into two locks: 
- The existing ipi_lock continues to protect the global inpcb list and
  inpcb counter.  This lock is now relegated to a small number of
  allocation and free operations, and occasional operations that walk
  all connections (including, awkwardly, certain UDP multicast receive
  operations -- something to revisit).

- A new ipi_hash_lock protects the two inpcbinfo hash tables for
  looking up connections and bound sockets, manipulated using new
  INP_HASH_*() macros.  This lock, combined with inpcb locks, protects
  the 4-tuple address space.

Unlike the current ipi_lock, ipi_hash_lock follows the individual inpcb
connection locks, so may be acquired while manipulating a connection on
which a lock is already held, avoiding the need to acquire the inpcbinfo
lock preemptively when a binding change might later be required.  As a
result, however, lookup operations necessarily go through a reference
acquire while holding the lookup lock, later acquiring an inpcb lock --
if required.

A new function in_pcblookup() looks up connections, and accepts flags
indicating how to return the inpcb.  Due to lock order changes, callers
no longer need acquire locks before performing a lookup: the lookup
routine will acquire the ipi_hash_lock as needed.  In the future, it will
also be able to use alternative lookup and locking strategies
transparently to callers, such as pcbgroup lookup.  New lookup flags are,
supplementing the existing INPLOOKUP_WILDCARD flag:

  INPLOOKUP_RLOCKPCB - Acquire a read lock on the returned inpcb
  INPLOOKUP_WLOCKPCB - Acquire a write lock on the returned inpcb

Callers must pass exactly one of these flags (for the time being).

Some notes:

- All protocols are updated to work within the new regime; especially,
  TCP, UDPv4, and UDPv6.  pcbinfo ipi_lock acquisitions are largely
  eliminated, and global hash lock hold times are dramatically reduced
  compared to previous locking.
- The TCP syncache still relies on the pcbinfo lock, something that we
  may want to revisit.
- Support for reverting to the FreeBSD 7.x locking strategy in TCP input
  is no longer available -- hash lookup locks are now held only very
  briefly during inpcb lookup, rather than for potentially extended
  periods.  However, the pcbinfo ipi_lock will still be acquired if a
  connection state might change such that a connection is added or
  removed.
- Raw IP sockets continue to use the pcbinfo ipi_lock for protection,
  due to maintaining their own hash tables.
- The interface in6_pcblookup_hash_locked() is maintained, which allows
  callers to acquire hash locks and perform one or more lookups atomically
  with 4-tuple allocation: this is required only for TCPv6, as there is no
  in6_pcbconnect_setup(), which there should be.
- UDPv6 locking remains significantly more conservative than UDPv4
  locking, which relates to source address selection.  This needs
  attention, as it likely significantly reduces parallelism in this code
  for multithreaded socket use (such as in BIND).
- In the UDPv4 and UDPv6 multicast cases, we need to revisit locking
  somewhat, as they relied on ipi_lock to stablise 4-tuple matches, which
  is no longer sufficient.  A second check once the inpcb lock is held
  should do the trick, keeping the general case from requiring the inpcb
  lock for every inpcb visited.
- This work reminds us that we need to revisit locking of the v4/v6 flags,
  which may be accessed lock-free both before and after this change.
- Right now, a single lock name is used for the pcbhash lock -- this is
  undesirable, and probably another argument is required to take care of
  this (or a char array name field in the pcbinfo?).

This is not an MFC candidate for 8.x due to its impact on lookup and
locking semantics.  It's possible some of these issues could be worked
around with compatibility wrappers, if necessary.

Reviewed by:    bz
Sponsored by:   Juniper Networks, Inc.
2011-05-30 09:43:55 +00:00

1117 lines
28 KiB
C
Raw Blame History

/*-
* Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
* Copyright (c) 2010-2011 Juniper Networks, Inc.
* All rights reserved.
*
* Portions of this software were developed by Robert N. M. Watson under
* contract to Juniper Networks, Inc.
*
* 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. Neither the name of the project 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 PROJECT 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 PROJECT 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.
*
* $KAME: udp6_usrreq.c,v 1.27 2001/05/21 05:45:10 jinmei Exp $
* $KAME: udp6_output.c,v 1.31 2001/05/21 16:39:15 jinmei Exp $
*/
/*-
* Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
* 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.
*
* @(#)udp_usrreq.c 8.6 (Berkeley) 5/23/95
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_inet.h"
#include "opt_inet6.h"
#include "opt_ipsec.h"
#include <sys/param.h>
#include <sys/jail.h>
#include <sys/kernel.h>
#include <sys/lock.h>
#include <sys/mbuf.h>
#include <sys/priv.h>
#include <sys/proc.h>
#include <sys/protosw.h>
#include <sys/signalvar.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/sx.h>
#include <sys/sysctl.h>
#include <sys/syslog.h>
#include <sys/systm.h>
#include <net/if.h>
#include <net/if_types.h>
#include <net/route.h>
#include <netinet/in.h>
#include <netinet/in_pcb.h>
#include <netinet/in_systm.h>
#include <netinet/in_var.h>
#include <netinet/ip.h>
#include <netinet/ip_icmp.h>
#include <netinet/ip6.h>
#include <netinet/icmp_var.h>
#include <netinet/icmp6.h>
#include <netinet/ip_var.h>
#include <netinet/udp.h>
#include <netinet/udp_var.h>
#include <netinet6/ip6protosw.h>
#include <netinet6/ip6_var.h>
#include <netinet6/in6_pcb.h>
#include <netinet6/udp6_var.h>
#include <netinet6/scope6_var.h>
#ifdef IPSEC
#include <netipsec/ipsec.h>
#include <netipsec/ipsec6.h>
#endif /* IPSEC */
#include <security/mac/mac_framework.h>
/*
* UDP protocol implementation.
* Per RFC 768, August, 1980.
*/
extern struct protosw inetsw[];
static void udp6_detach(struct socket *so);
static void
udp6_append(struct inpcb *inp, struct mbuf *n, int off,
struct sockaddr_in6 *fromsa)
{
struct socket *so;
struct mbuf *opts;
INP_LOCK_ASSERT(inp);
#ifdef IPSEC
/* Check AH/ESP integrity. */
if (ipsec6_in_reject(n, inp)) {
m_freem(n);
V_ipsec6stat.in_polvio++;
return;
}
#endif /* IPSEC */
#ifdef MAC
if (mac_inpcb_check_deliver(inp, n) != 0) {
m_freem(n);
return;
}
#endif
opts = NULL;
if (inp->inp_flags & INP_CONTROLOPTS ||
inp->inp_socket->so_options & SO_TIMESTAMP)
ip6_savecontrol(inp, n, &opts);
m_adj(n, off + sizeof(struct udphdr));
so = inp->inp_socket;
SOCKBUF_LOCK(&so->so_rcv);
if (sbappendaddr_locked(&so->so_rcv, (struct sockaddr *)fromsa, n,
opts) == 0) {
SOCKBUF_UNLOCK(&so->so_rcv);
m_freem(n);
if (opts)
m_freem(opts);
UDPSTAT_INC(udps_fullsock);
} else
sorwakeup_locked(so);
}
int
udp6_input(struct mbuf **mp, int *offp, int proto)
{
struct mbuf *m = *mp;
struct ifnet *ifp;
struct ip6_hdr *ip6;
struct udphdr *uh;
struct inpcb *inp;
struct udpcb *up;
int off = *offp;
int plen, ulen;
struct sockaddr_in6 fromsa;
ifp = m->m_pkthdr.rcvif;
ip6 = mtod(m, struct ip6_hdr *);
if (faithprefix_p != NULL && (*faithprefix_p)(&ip6->ip6_dst)) {
/* XXX send icmp6 host/port unreach? */
m_freem(m);
return (IPPROTO_DONE);
}
#ifndef PULLDOWN_TEST
IP6_EXTHDR_CHECK(m, off, sizeof(struct udphdr), IPPROTO_DONE);
ip6 = mtod(m, struct ip6_hdr *);
uh = (struct udphdr *)((caddr_t)ip6 + off);
#else
IP6_EXTHDR_GET(uh, struct udphdr *, m, off, sizeof(*uh));
if (!uh)
return (IPPROTO_DONE);
#endif
UDPSTAT_INC(udps_ipackets);
/*
* Destination port of 0 is illegal, based on RFC768.
*/
if (uh->uh_dport == 0)
goto badunlocked;
plen = ntohs(ip6->ip6_plen) - off + sizeof(*ip6);
ulen = ntohs((u_short)uh->uh_ulen);
if (plen != ulen) {
UDPSTAT_INC(udps_badlen);
goto badunlocked;
}
/*
* Checksum extended UDP header and data.
*/
if (uh->uh_sum == 0) {
UDPSTAT_INC(udps_nosum);
goto badunlocked;
}
if (in6_cksum(m, IPPROTO_UDP, off, ulen) != 0) {
UDPSTAT_INC(udps_badsum);
goto badunlocked;
}
/*
* Construct sockaddr format source address.
*/
init_sin6(&fromsa, m);
fromsa.sin6_port = uh->uh_sport;
if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
struct inpcb *last;
struct ip6_moptions *imo;
INP_INFO_RLOCK(&V_udbinfo);
/*
* In the event that laddr should be set to the link-local
* address (this happens in RIPng), the multicast address
* specified in the received packet will not match laddr. To
* handle this situation, matching is relaxed if the
* receiving interface is the same as one specified in the
* socket and if the destination multicast address matches
* one of the multicast groups specified in the socket.
*/
/*
* KAME note: traditionally we dropped udpiphdr from mbuf
* here. We need udphdr for IPsec processing so we do that
* later.
*/
last = NULL;
LIST_FOREACH(inp, &V_udb, inp_list) {
if ((inp->inp_vflag & INP_IPV6) == 0)
continue;
if (inp->inp_lport != uh->uh_dport)
continue;
if (inp->inp_fport != 0 &&
inp->inp_fport != uh->uh_sport)
continue;
if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr)) {
if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_laddr,
&ip6->ip6_dst))
continue;
}
if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) {
if (!IN6_ARE_ADDR_EQUAL(&inp->in6p_faddr,
&ip6->ip6_src) ||
inp->inp_fport != uh->uh_sport)
continue;
}
/*
* XXXRW: Because we weren't holding either the inpcb
* or the hash lock when we checked for a match
* before, we should probably recheck now that the
* inpcb lock is (supposed to be) held.
*/
/*
* Handle socket delivery policy for any-source
* and source-specific multicast. [RFC3678]
*/
imo = inp->in6p_moptions;
if (imo && IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
struct sockaddr_in6 mcaddr;
int blocked;
INP_RLOCK(inp);
bzero(&mcaddr, sizeof(struct sockaddr_in6));
mcaddr.sin6_len = sizeof(struct sockaddr_in6);
mcaddr.sin6_family = AF_INET6;
mcaddr.sin6_addr = ip6->ip6_dst;
blocked = im6o_mc_filter(imo, ifp,
(struct sockaddr *)&mcaddr,
(struct sockaddr *)&fromsa);
if (blocked != MCAST_PASS) {
if (blocked == MCAST_NOTGMEMBER)
IP6STAT_INC(ip6s_notmember);
if (blocked == MCAST_NOTSMEMBER ||
blocked == MCAST_MUTED)
UDPSTAT_INC(udps_filtermcast);
INP_RUNLOCK(inp); /* XXX */
continue;
}
INP_RUNLOCK(inp);
}
if (last != NULL) {
struct mbuf *n;
if ((n = m_copy(m, 0, M_COPYALL)) != NULL) {
INP_RLOCK(last);
up = intoudpcb(last);
if (up->u_tun_func == NULL) {
udp6_append(last, n, off, &fromsa);
} else {
/*
* Engage the tunneling
* protocol we will have to
* leave the info_lock up,
* since we are hunting
* through multiple UDP's.
*
*/
(*up->u_tun_func)(n, off, last);
}
INP_RUNLOCK(last);
}
}
last = inp;
/*
* Don't look for additional matches if this one does
* not have either the SO_REUSEPORT or SO_REUSEADDR
* socket options set. This heuristic avoids
* searching through all pcbs in the common case of a
* non-shared port. It assumes that an application
* will never clear these options after setting them.
*/
if ((last->inp_socket->so_options &
(SO_REUSEPORT|SO_REUSEADDR)) == 0)
break;
}
if (last == NULL) {
/*
* No matching pcb found; discard datagram. (No need
* to send an ICMP Port Unreachable for a broadcast
* or multicast datgram.)
*/
UDPSTAT_INC(udps_noport);
UDPSTAT_INC(udps_noportmcast);
goto badheadlocked;
}
INP_RLOCK(last);
INP_INFO_RUNLOCK(&V_udbinfo);
up = intoudpcb(last);
if (up->u_tun_func == NULL) {
udp6_append(last, m, off, &fromsa);
} else {
/*
* Engage the tunneling protocol.
*/
(*up->u_tun_func)(m, off, last);
}
INP_RUNLOCK(last);
return (IPPROTO_DONE);
}
/*
* Locate pcb for datagram.
*/
inp = in6_pcblookup(&V_udbinfo, &ip6->ip6_src, uh->uh_sport,
&ip6->ip6_dst, uh->uh_dport, INPLOOKUP_WILDCARD |
INPLOOKUP_RLOCKPCB, m->m_pkthdr.rcvif);
if (inp == NULL) {
if (udp_log_in_vain) {
char ip6bufs[INET6_ADDRSTRLEN];
char ip6bufd[INET6_ADDRSTRLEN];
log(LOG_INFO,
"Connection attempt to UDP [%s]:%d from [%s]:%d\n",
ip6_sprintf(ip6bufd, &ip6->ip6_dst),
ntohs(uh->uh_dport),
ip6_sprintf(ip6bufs, &ip6->ip6_src),
ntohs(uh->uh_sport));
}
UDPSTAT_INC(udps_noport);
if (m->m_flags & M_MCAST) {
printf("UDP6: M_MCAST is set in a unicast packet.\n");
UDPSTAT_INC(udps_noportmcast);
goto badunlocked;
}
if (V_udp_blackhole)
goto badunlocked;
if (badport_bandlim(BANDLIM_ICMP6_UNREACH) < 0)
goto badunlocked;
icmp6_error(m, ICMP6_DST_UNREACH, ICMP6_DST_UNREACH_NOPORT, 0);
return (IPPROTO_DONE);
}
INP_RLOCK_ASSERT(inp);
up = intoudpcb(inp);
if (up->u_tun_func == NULL) {
udp6_append(inp, m, off, &fromsa);
} else {
/*
* Engage the tunneling protocol.
*/
(*up->u_tun_func)(m, off, inp);
}
INP_RUNLOCK(inp);
return (IPPROTO_DONE);
badheadlocked:
INP_INFO_RUNLOCK(&V_udbinfo);
badunlocked:
if (m)
m_freem(m);
return (IPPROTO_DONE);
}
void
udp6_ctlinput(int cmd, struct sockaddr *sa, void *d)
{
struct udphdr uh;
struct ip6_hdr *ip6;
struct mbuf *m;
int off = 0;
struct ip6ctlparam *ip6cp = NULL;
const struct sockaddr_in6 *sa6_src = NULL;
void *cmdarg;
struct inpcb *(*notify)(struct inpcb *, int) = udp_notify;
struct udp_portonly {
u_int16_t uh_sport;
u_int16_t uh_dport;
} *uhp;
if (sa->sa_family != AF_INET6 ||
sa->sa_len != sizeof(struct sockaddr_in6))
return;
if ((unsigned)cmd >= PRC_NCMDS)
return;
if (PRC_IS_REDIRECT(cmd))
notify = in6_rtchange, d = NULL;
else if (cmd == PRC_HOSTDEAD)
d = NULL;
else if (inet6ctlerrmap[cmd] == 0)
return;
/* if the parameter is from icmp6, decode it. */
if (d != NULL) {
ip6cp = (struct ip6ctlparam *)d;
m = ip6cp->ip6c_m;
ip6 = ip6cp->ip6c_ip6;
off = ip6cp->ip6c_off;
cmdarg = ip6cp->ip6c_cmdarg;
sa6_src = ip6cp->ip6c_src;
} else {
m = NULL;
ip6 = NULL;
cmdarg = NULL;
sa6_src = &sa6_any;
}
if (ip6) {
/*
* XXX: We assume that when IPV6 is non NULL,
* M and OFF are valid.
*/
/* Check if we can safely examine src and dst ports. */
if (m->m_pkthdr.len < off + sizeof(*uhp))
return;
bzero(&uh, sizeof(uh));
m_copydata(m, off, sizeof(*uhp), (caddr_t)&uh);
(void) in6_pcbnotify(&V_udbinfo, sa, uh.uh_dport,
(struct sockaddr *)ip6cp->ip6c_src, uh.uh_sport, cmd,
cmdarg, notify);
} else
(void) in6_pcbnotify(&V_udbinfo, sa, 0,
(const struct sockaddr *)sa6_src, 0, cmd, cmdarg, notify);
}
static int
udp6_getcred(SYSCTL_HANDLER_ARGS)
{
struct xucred xuc;
struct sockaddr_in6 addrs[2];
struct inpcb *inp;
int error;
error = priv_check(req->td, PRIV_NETINET_GETCRED);
if (error)
return (error);
if (req->newlen != sizeof(addrs))
return (EINVAL);
if (req->oldlen != sizeof(struct xucred))
return (EINVAL);
error = SYSCTL_IN(req, addrs, sizeof(addrs));
if (error)
return (error);
if ((error = sa6_embedscope(&addrs[0], V_ip6_use_defzone)) != 0 ||
(error = sa6_embedscope(&addrs[1], V_ip6_use_defzone)) != 0) {
return (error);
}
inp = in6_pcblookup(&V_udbinfo, &addrs[1].sin6_addr,
addrs[1].sin6_port, &addrs[0].sin6_addr, addrs[0].sin6_port,
INPLOOKUP_WILDCARD | INPLOOKUP_RLOCKPCB, NULL);
if (inp != NULL) {
INP_RLOCK_ASSERT(inp);
if (inp->inp_socket == NULL)
error = ENOENT;
if (error == 0)
error = cr_canseesocket(req->td->td_ucred,
inp->inp_socket);
if (error == 0)
cru2x(inp->inp_cred, &xuc);
INP_RUNLOCK(inp);
} else
error = ENOENT;
if (error == 0)
error = SYSCTL_OUT(req, &xuc, sizeof(struct xucred));
return (error);
}
SYSCTL_PROC(_net_inet6_udp6, OID_AUTO, getcred, CTLTYPE_OPAQUE|CTLFLAG_RW, 0,
0, udp6_getcred, "S,xucred", "Get the xucred of a UDP6 connection");
static int
udp6_output(struct inpcb *inp, struct mbuf *m, struct sockaddr *addr6,
struct mbuf *control, struct thread *td)
{
u_int32_t ulen = m->m_pkthdr.len;
u_int32_t plen = sizeof(struct udphdr) + ulen;
struct ip6_hdr *ip6;
struct udphdr *udp6;
struct in6_addr *laddr, *faddr, in6a;
struct sockaddr_in6 *sin6 = NULL;
struct ifnet *oifp = NULL;
int scope_ambiguous = 0;
u_short fport;
int error = 0;
struct ip6_pktopts *optp, opt;
int af = AF_INET6, hlen = sizeof(struct ip6_hdr);
int flags;
struct sockaddr_in6 tmp;
INP_WLOCK_ASSERT(inp);
INP_HASH_WLOCK_ASSERT(inp->inp_pcbinfo);
if (addr6) {
/* addr6 has been validated in udp6_send(). */
sin6 = (struct sockaddr_in6 *)addr6;
/* protect *sin6 from overwrites */
tmp = *sin6;
sin6 = &tmp;
/*
* Application should provide a proper zone ID or the use of
* default zone IDs should be enabled. Unfortunately, some
* applications do not behave as it should, so we need a
* workaround. Even if an appropriate ID is not determined,
* we'll see if we can determine the outgoing interface. If we
* can, determine the zone ID based on the interface below.
*/
if (sin6->sin6_scope_id == 0 && !V_ip6_use_defzone)
scope_ambiguous = 1;
if ((error = sa6_embedscope(sin6, V_ip6_use_defzone)) != 0)
return (error);
}
if (control) {
if ((error = ip6_setpktopts(control, &opt,
inp->in6p_outputopts, td->td_ucred, IPPROTO_UDP)) != 0)
goto release;
optp = &opt;
} else
optp = inp->in6p_outputopts;
if (sin6) {
faddr = &sin6->sin6_addr;
/*
* IPv4 version of udp_output calls in_pcbconnect in this case,
* which needs splnet and affects performance.
* Since we saw no essential reason for calling in_pcbconnect,
* we get rid of such kind of logic, and call in6_selectsrc
* and in6_pcbsetport in order to fill in the local address
* and the local port.
*/
if (sin6->sin6_port == 0) {
error = EADDRNOTAVAIL;
goto release;
}
if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) {
/* how about ::ffff:0.0.0.0 case? */
error = EISCONN;
goto release;
}
fport = sin6->sin6_port; /* allow 0 port */
if (IN6_IS_ADDR_V4MAPPED(faddr)) {
if ((inp->inp_flags & IN6P_IPV6_V6ONLY)) {
/*
* I believe we should explicitly discard the
* packet when mapped addresses are disabled,
* rather than send the packet as an IPv6 one.
* If we chose the latter approach, the packet
* might be sent out on the wire based on the
* default route, the situation which we'd
* probably want to avoid.
* (20010421 jinmei@kame.net)
*/
error = EINVAL;
goto release;
}
if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_laddr) &&
!IN6_IS_ADDR_V4MAPPED(&inp->in6p_laddr)) {
/*
* when remote addr is an IPv4-mapped address,
* local addr should not be an IPv6 address,
* since you cannot determine how to map IPv6
* source address to IPv4.
*/
error = EINVAL;
goto release;
}
af = AF_INET;
}
if (!IN6_IS_ADDR_V4MAPPED(faddr)) {
error = in6_selectsrc(sin6, optp, inp, NULL,
td->td_ucred, &oifp, &in6a);
if (error)
goto release;
if (oifp && scope_ambiguous &&
(error = in6_setscope(&sin6->sin6_addr,
oifp, NULL))) {
goto release;
}
laddr = &in6a;
} else
laddr = &inp->in6p_laddr; /* XXX */
if (laddr == NULL) {
if (error == 0)
error = EADDRNOTAVAIL;
goto release;
}
if (inp->inp_lport == 0 &&
(error = in6_pcbsetport(laddr, inp, td->td_ucred)) != 0) {
/* Undo an address bind that may have occurred. */
inp->in6p_laddr = in6addr_any;
goto release;
}
} else {
if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) {
error = ENOTCONN;
goto release;
}
if (IN6_IS_ADDR_V4MAPPED(&inp->in6p_faddr)) {
if ((inp->inp_flags & IN6P_IPV6_V6ONLY)) {
/*
* XXX: this case would happen when the
* application sets the V6ONLY flag after
* connecting the foreign address.
* Such applications should be fixed,
* so we bark here.
*/
log(LOG_INFO, "udp6_output: IPV6_V6ONLY "
"option was set for a connected socket\n");
error = EINVAL;
goto release;
} else
af = AF_INET;
}
laddr = &inp->in6p_laddr;
faddr = &inp->in6p_faddr;
fport = inp->inp_fport;
}
if (af == AF_INET)
hlen = sizeof(struct ip);
/*
* Calculate data length and get a mbuf
* for UDP and IP6 headers.
*/
M_PREPEND(m, hlen + sizeof(struct udphdr), M_DONTWAIT);
if (m == 0) {
error = ENOBUFS;
goto release;
}
/*
* Stuff checksum and output datagram.
*/
udp6 = (struct udphdr *)(mtod(m, caddr_t) + hlen);
udp6->uh_sport = inp->inp_lport; /* lport is always set in the PCB */
udp6->uh_dport = fport;
if (plen <= 0xffff)
udp6->uh_ulen = htons((u_short)plen);
else
udp6->uh_ulen = 0;
udp6->uh_sum = 0;
switch (af) {
case AF_INET6:
ip6 = mtod(m, struct ip6_hdr *);
ip6->ip6_flow = inp->inp_flow & IPV6_FLOWINFO_MASK;
ip6->ip6_vfc &= ~IPV6_VERSION_MASK;
ip6->ip6_vfc |= IPV6_VERSION;
#if 0 /* ip6_plen will be filled in ip6_output. */
ip6->ip6_plen = htons((u_short)plen);
#endif
ip6->ip6_nxt = IPPROTO_UDP;
ip6->ip6_hlim = in6_selecthlim(inp, NULL);
ip6->ip6_src = *laddr;
ip6->ip6_dst = *faddr;
if ((udp6->uh_sum = in6_cksum(m, IPPROTO_UDP,
sizeof(struct ip6_hdr), plen)) == 0) {
udp6->uh_sum = 0xffff;
}
flags = 0;
UDPSTAT_INC(udps_opackets);
error = ip6_output(m, optp, NULL, flags, inp->in6p_moptions,
NULL, inp);
break;
case AF_INET:
error = EAFNOSUPPORT;
goto release;
}
goto releaseopt;
release:
m_freem(m);
releaseopt:
if (control) {
ip6_clearpktopts(&opt, -1);
m_freem(control);
}
return (error);
}
static void
udp6_abort(struct socket *so)
{
struct inpcb *inp;
inp = sotoinpcb(so);
KASSERT(inp != NULL, ("udp6_abort: inp == NULL"));
#ifdef INET
if (inp->inp_vflag & INP_IPV4) {
struct pr_usrreqs *pru;
pru = inetsw[ip_protox[IPPROTO_UDP]].pr_usrreqs;
(*pru->pru_abort)(so);
return;
}
#endif
INP_WLOCK(inp);
if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) {
INP_HASH_WLOCK(&V_udbinfo);
in6_pcbdisconnect(inp);
inp->in6p_laddr = in6addr_any;
INP_HASH_WUNLOCK(&V_udbinfo);
soisdisconnected(so);
}
INP_WUNLOCK(inp);
}
static int
udp6_attach(struct socket *so, int proto, struct thread *td)
{
struct inpcb *inp;
int error;
inp = sotoinpcb(so);
KASSERT(inp == NULL, ("udp6_attach: inp != NULL"));
if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
error = soreserve(so, udp_sendspace, udp_recvspace);
if (error)
return (error);
}
INP_INFO_WLOCK(&V_udbinfo);
error = in_pcballoc(so, &V_udbinfo);
if (error) {
INP_INFO_WUNLOCK(&V_udbinfo);
return (error);
}
inp = (struct inpcb *)so->so_pcb;
inp->inp_vflag |= INP_IPV6;
if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0)
inp->inp_vflag |= INP_IPV4;
inp->in6p_hops = -1; /* use kernel default */
inp->in6p_cksum = -1; /* just to be sure */
/*
* XXX: ugly!!
* IPv4 TTL initialization is necessary for an IPv6 socket as well,
* because the socket may be bound to an IPv6 wildcard address,
* which may match an IPv4-mapped IPv6 address.
*/
inp->inp_ip_ttl = V_ip_defttl;
error = udp_newudpcb(inp);
if (error) {
in_pcbdetach(inp);
in_pcbfree(inp);
INP_INFO_WUNLOCK(&V_udbinfo);
return (error);
}
INP_WUNLOCK(inp);
INP_INFO_WUNLOCK(&V_udbinfo);
return (0);
}
static int
udp6_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
{
struct inpcb *inp;
int error;
inp = sotoinpcb(so);
KASSERT(inp != NULL, ("udp6_bind: inp == NULL"));
INP_WLOCK(inp);
INP_HASH_WLOCK(&V_udbinfo);
inp->inp_vflag &= ~INP_IPV4;
inp->inp_vflag |= INP_IPV6;
if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0) {
struct sockaddr_in6 *sin6_p;
sin6_p = (struct sockaddr_in6 *)nam;
if (IN6_IS_ADDR_UNSPECIFIED(&sin6_p->sin6_addr))
inp->inp_vflag |= INP_IPV4;
#ifdef INET
else if (IN6_IS_ADDR_V4MAPPED(&sin6_p->sin6_addr)) {
struct sockaddr_in sin;
in6_sin6_2_sin(&sin, sin6_p);
inp->inp_vflag |= INP_IPV4;
inp->inp_vflag &= ~INP_IPV6;
error = in_pcbbind(inp, (struct sockaddr *)&sin,
td->td_ucred);
goto out;
}
#endif
}
error = in6_pcbbind(inp, nam, td->td_ucred);
#ifdef INET
out:
#endif
INP_HASH_WUNLOCK(&V_udbinfo);
INP_WUNLOCK(inp);
return (error);
}
static void
udp6_close(struct socket *so)
{
struct inpcb *inp;
inp = sotoinpcb(so);
KASSERT(inp != NULL, ("udp6_close: inp == NULL"));
#ifdef INET
if (inp->inp_vflag & INP_IPV4) {
struct pr_usrreqs *pru;
pru = inetsw[ip_protox[IPPROTO_UDP]].pr_usrreqs;
(*pru->pru_disconnect)(so);
return;
}
#endif
INP_WLOCK(inp);
if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) {
INP_HASH_WLOCK(&V_udbinfo);
in6_pcbdisconnect(inp);
inp->in6p_laddr = in6addr_any;
INP_HASH_WUNLOCK(&V_udbinfo);
soisdisconnected(so);
}
INP_WUNLOCK(inp);
}
static int
udp6_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
{
struct inpcb *inp;
struct sockaddr_in6 *sin6;
int error;
inp = sotoinpcb(so);
sin6 = (struct sockaddr_in6 *)nam;
KASSERT(inp != NULL, ("udp6_connect: inp == NULL"));
/*
* XXXRW: Need to clarify locking of v4/v6 flags.
*/
INP_WLOCK(inp);
#ifdef INET
if (IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)) {
struct sockaddr_in sin;
if ((inp->inp_flags & IN6P_IPV6_V6ONLY) != 0) {
error = EINVAL;
goto out;
}
if (inp->inp_faddr.s_addr != INADDR_ANY) {
error = EISCONN;
goto out;
}
in6_sin6_2_sin(&sin, sin6);
inp->inp_vflag |= INP_IPV4;
inp->inp_vflag &= ~INP_IPV6;
error = prison_remote_ip4(td->td_ucred, &sin.sin_addr);
if (error != 0)
goto out;
INP_HASH_WLOCK(&V_udbinfo);
error = in_pcbconnect(inp, (struct sockaddr *)&sin,
td->td_ucred);
INP_HASH_WUNLOCK(&V_udbinfo);
if (error == 0)
soisconnected(so);
goto out;
}
#endif
if (!IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) {
error = EISCONN;
goto out;
}
inp->inp_vflag &= ~INP_IPV4;
inp->inp_vflag |= INP_IPV6;
error = prison_remote_ip6(td->td_ucred, &sin6->sin6_addr);
if (error != 0)
goto out;
INP_HASH_WLOCK(&V_udbinfo);
error = in6_pcbconnect(inp, nam, td->td_ucred);
INP_HASH_WUNLOCK(&V_udbinfo);
if (error == 0)
soisconnected(so);
out:
INP_WUNLOCK(inp);
return (error);
}
static void
udp6_detach(struct socket *so)
{
struct inpcb *inp;
struct udpcb *up;
inp = sotoinpcb(so);
KASSERT(inp != NULL, ("udp6_detach: inp == NULL"));
INP_INFO_WLOCK(&V_udbinfo);
INP_WLOCK(inp);
up = intoudpcb(inp);
KASSERT(up != NULL, ("%s: up == NULL", __func__));
in_pcbdetach(inp);
in_pcbfree(inp);
INP_INFO_WUNLOCK(&V_udbinfo);
udp_discardcb(up);
}
static int
udp6_disconnect(struct socket *so)
{
struct inpcb *inp;
int error;
inp = sotoinpcb(so);
KASSERT(inp != NULL, ("udp6_disconnect: inp == NULL"));
#ifdef INET
if (inp->inp_vflag & INP_IPV4) {
struct pr_usrreqs *pru;
pru = inetsw[ip_protox[IPPROTO_UDP]].pr_usrreqs;
(void)(*pru->pru_disconnect)(so);
return (0);
}
#endif
INP_WLOCK(inp);
if (IN6_IS_ADDR_UNSPECIFIED(&inp->in6p_faddr)) {
error = ENOTCONN;
goto out;
}
INP_HASH_WLOCK(&V_udbinfo);
in6_pcbdisconnect(inp);
inp->in6p_laddr = in6addr_any;
INP_HASH_WUNLOCK(&V_udbinfo);
SOCK_LOCK(so);
so->so_state &= ~SS_ISCONNECTED; /* XXX */
SOCK_UNLOCK(so);
out:
INP_WUNLOCK(inp);
return (0);
}
static int
udp6_send(struct socket *so, int flags, struct mbuf *m,
struct sockaddr *addr, struct mbuf *control, struct thread *td)
{
struct inpcb *inp;
int error = 0;
inp = sotoinpcb(so);
KASSERT(inp != NULL, ("udp6_send: inp == NULL"));
INP_WLOCK(inp);
if (addr) {
if (addr->sa_len != sizeof(struct sockaddr_in6)) {
error = EINVAL;
goto bad;
}
if (addr->sa_family != AF_INET6) {
error = EAFNOSUPPORT;
goto bad;
}
}
#ifdef INET
if ((inp->inp_flags & IN6P_IPV6_V6ONLY) == 0) {
int hasv4addr;
struct sockaddr_in6 *sin6 = 0;
if (addr == 0)
hasv4addr = (inp->inp_vflag & INP_IPV4);
else {
sin6 = (struct sockaddr_in6 *)addr;
hasv4addr = IN6_IS_ADDR_V4MAPPED(&sin6->sin6_addr)
? 1 : 0;
}
if (hasv4addr) {
struct pr_usrreqs *pru;
/*
* XXXRW: We release UDP-layer locks before calling
* udp_send() in order to avoid recursion. However,
* this does mean there is a short window where inp's
* fields are unstable. Could this lead to a
* potential race in which the factors causing us to
* select the UDPv4 output routine are invalidated?
*/
INP_WUNLOCK(inp);
if (sin6)
in6_sin6_2_sin_in_sock(addr);
pru = inetsw[ip_protox[IPPROTO_UDP]].pr_usrreqs;
/* addr will just be freed in sendit(). */
return ((*pru->pru_send)(so, flags, m, addr, control,
td));
}
}
#endif
#ifdef MAC
mac_inpcb_create_mbuf(inp, m);
#endif
INP_HASH_WLOCK(&V_udbinfo);
error = udp6_output(inp, m, addr, control, td);
INP_HASH_WUNLOCK(&V_udbinfo);
#ifdef INET
#endif
INP_WUNLOCK(inp);
return (error);
bad:
INP_WUNLOCK(inp);
m_freem(m);
return (error);
}
struct pr_usrreqs udp6_usrreqs = {
.pru_abort = udp6_abort,
.pru_attach = udp6_attach,
.pru_bind = udp6_bind,
.pru_connect = udp6_connect,
.pru_control = in6_control,
.pru_detach = udp6_detach,
.pru_disconnect = udp6_disconnect,
.pru_peeraddr = in6_mapped_peeraddr,
.pru_send = udp6_send,
.pru_shutdown = udp_shutdown,
.pru_sockaddr = in6_mapped_sockaddr,
.pru_soreceive = soreceive_dgram,
.pru_sosend = sosend_dgram,
.pru_sosetlabel = in_pcbsosetlabel,
.pru_close = udp6_close
};
Reference in New Issue View Git Blame Copy Permalink