Please note that diffs are not public domain; they are subject to the copyright notices on the relevant files. =================================================================== RCS file: /ftp/cvs/cvsroot/src/sys/netinet/raw_ip.c,v rcsdiff: /ftp/cvs/cvsroot/src/sys/netinet/raw_ip.c,v: warning: Unknown phrases like `commitid ...;' are present. retrieving revision 1.88.10.3 retrieving revision 1.103.6.1 diff -u -p -r1.88.10.3 -r1.103.6.1 --- src/sys/netinet/raw_ip.c 2006/05/06 23:32:11 1.88.10.3 +++ src/sys/netinet/raw_ip.c 2008/06/02 13:24:24 1.103.6.1 @@ -1,4 +1,4 @@ -/* $NetBSD: raw_ip.c,v 1.88.10.3 2006/05/06 23:32:11 christos Exp $ */ +/* $NetBSD: raw_ip.c,v 1.103.6.1 2008/06/02 13:24:24 mjf Exp $ */ /* * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. @@ -61,7 +61,7 @@ */ #include -__KERNEL_RCSID(0, "$NetBSD: raw_ip.c,v 1.88.10.3 2006/05/06 23:32:11 christos Exp $"); +__KERNEL_RCSID(0, "$NetBSD: raw_ip.c,v 1.103.6.1 2008/06/02 13:24:24 mjf Exp $"); #include "opt_inet.h" #include "opt_ipsec.h" @@ -86,6 +86,7 @@ __KERNEL_RCSID(0, "$NetBSD: raw_ip.c,v 1 #include #include #include +#include #include #include #include @@ -96,12 +97,14 @@ __KERNEL_RCSID(0, "$NetBSD: raw_ip.c,v 1 #ifdef IPSEC #include -#endif /*IPSEC*/ +#include +#endif /* IPSEC */ #ifdef FAST_IPSEC #include -#include /* XXX ipsecstat namespace */ -#endif /* FAST_IPSEC*/ +#include +#include +#endif /* FAST_IPSEC */ struct inpcbtable rawcbtable; @@ -131,6 +134,24 @@ rip_init(void) in_pcbinit(&rawcbtable, 1, 1); } +static void +rip_sbappendaddr(struct inpcb *last, struct ip *ip, const struct sockaddr *sa, + int hlen, struct mbuf *opts, struct mbuf *n) +{ + if (last->inp_flags & INP_NOHEADER) + m_adj(n, hlen); + if (last->inp_flags & INP_CONTROLOPTS || + last->inp_socket->so_options & SO_TIMESTAMP) + ip_savecontrol(last, &opts, ip, n); + if (sbappendaddr(&last->inp_socket->so_rcv, sa, n, opts) == 0) { + /* should notify about lost packet */ + m_freem(n); + if (opts) + m_freem(opts); + } else + sorwakeup(last->inp_socket); +} + /* * Setup generic address and protocol structures * for raw_input routine, then pass them along with @@ -139,12 +160,12 @@ rip_init(void) void rip_input(struct mbuf *m, ...) { - int proto; + int hlen, proto; struct ip *ip = mtod(m, struct ip *); struct inpcb_hdr *inph; struct inpcb *inp; - struct inpcb *last = 0; - struct mbuf *opts = 0; + struct inpcb *last = NULL; + struct mbuf *n, *opts = NULL; struct sockaddr_in ripsrc; va_list ap; @@ -153,18 +174,15 @@ rip_input(struct mbuf *m, ...) proto = va_arg(ap, int); va_end(ap); - ripsrc.sin_family = AF_INET; - ripsrc.sin_len = sizeof(struct sockaddr_in); - ripsrc.sin_addr = ip->ip_src; - ripsrc.sin_port = 0; - bzero((caddr_t)ripsrc.sin_zero, sizeof(ripsrc.sin_zero)); + sockaddr_in_init(&ripsrc, &ip->ip_src, 0); /* * XXX Compatibility: programs using raw IP expect ip_len * XXX to have the header length subtracted, and in host order. * XXX ip_off is also expected to be host order. */ - ip->ip_len = ntohs(ip->ip_len) - (ip->ip_hl << 2); + hlen = ip->ip_hl << 2; + ip->ip_len = ntohs(ip->ip_len) - hlen; NTOHS(ip->ip_off); CIRCLEQ_FOREACH(inph, &rawcbtable.inpt_queue, inph_queue) { @@ -179,62 +197,44 @@ rip_input(struct mbuf *m, ...) if (!in_nullhost(inp->inp_faddr) && !in_hosteq(inp->inp_faddr, ip->ip_src)) continue; - if (last) { - struct mbuf *n; - + if (last == NULL) + ; #if defined(IPSEC) || defined(FAST_IPSEC) - /* check AH/ESP integrity. */ - if (ipsec4_in_reject_so(m, last->inp_socket)) { - ipsecstat.in_polvio++; - /* do not inject data to pcb */ - } else + /* check AH/ESP integrity. */ + else if (ipsec4_in_reject_so(m, last->inp_socket)) { + IPSEC_STATINC(IPSEC_STAT_IN_POLVIO); + /* do not inject data to pcb */ + } #endif /*IPSEC*/ - if ((n = m_copy(m, 0, (int)M_COPYALL)) != NULL) { - if (last->inp_flags & INP_CONTROLOPTS || - last->inp_socket->so_options & SO_TIMESTAMP) - ip_savecontrol(last, &opts, ip, n); - if (sbappendaddr(&last->inp_socket->so_rcv, - sintosa(&ripsrc), n, opts) == 0) { - /* should notify about lost packet */ - m_freem(n); - if (opts) - m_freem(opts); - } else - sorwakeup(last->inp_socket); - opts = NULL; - } + else if ((n = m_copypacket(m, M_DONTWAIT)) != NULL) { + rip_sbappendaddr(last, ip, sintosa(&ripsrc), hlen, opts, + n); + opts = NULL; } last = inp; } #if defined(IPSEC) || defined(FAST_IPSEC) /* check AH/ESP integrity. */ - if (last && ipsec4_in_reject_so(m, last->inp_socket)) { + if (last != NULL && ipsec4_in_reject_so(m, last->inp_socket)) { m_freem(m); - ipsecstat.in_polvio++; - ipstat.ips_delivered--; + IPSEC_STATINC(IPSEC_STAT_IN_POLVIO); + IP_STATDEC(IP_STAT_DELIVERED); /* do not inject data to pcb */ } else #endif /*IPSEC*/ - if (last) { - if (last->inp_flags & INP_CONTROLOPTS || - last->inp_socket->so_options & SO_TIMESTAMP) - ip_savecontrol(last, &opts, ip, m); - if (sbappendaddr(&last->inp_socket->so_rcv, - sintosa(&ripsrc), m, opts) == 0) { - m_freem(m); - if (opts) - m_freem(opts); - } else - sorwakeup(last->inp_socket); - } else { - if (inetsw[ip_protox[ip->ip_p]].pr_input == rip_input) { - icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PROTOCOL, - 0, 0); - ipstat.ips_noproto++; - ipstat.ips_delivered--; - } else - m_freem(m); - } + if (last != NULL) + rip_sbappendaddr(last, ip, sintosa(&ripsrc), hlen, opts, m); + else if (inetsw[ip_protox[ip->ip_p]].pr_input == rip_input) { + uint64_t *ips; + + icmp_error(m, ICMP_UNREACH, ICMP_UNREACH_PROTOCOL, + 0, 0); + ips = IP_STAT_GETREF(); + ips[IP_STAT_NOPROTO]++; + ips[IP_STAT_DELIVERED]--; + IP_STAT_PUTREF(); + } else + m_freem(m); return; } @@ -266,7 +266,7 @@ rip_pcbnotify(struct inpcbtable *table, } void * -rip_ctlinput(int cmd, struct sockaddr *sa, void *v) +rip_ctlinput(int cmd, const struct sockaddr *sa, void *v) { struct ip *ip = v; void (*notify)(struct inpcb *, int) = in_rtchange; @@ -285,12 +285,12 @@ rip_ctlinput(int cmd, struct sockaddr *s else if (errno == 0) return NULL; if (ip) { - rip_pcbnotify(&rawcbtable, satosin(sa)->sin_addr, + rip_pcbnotify(&rawcbtable, satocsin(sa)->sin_addr, ip->ip_src, ip->ip_p, errno, notify); /* XXX mapped address case */ } else - in_pcbnotifyall(&rawcbtable, satosin(sa)->sin_addr, errno, + in_pcbnotifyall(&rawcbtable, satocsin(sa)->sin_addr, errno, notify); return NULL; } @@ -365,12 +365,12 @@ rip_output(struct mbuf *m, ...) } HTONS(ip->ip_len); HTONS(ip->ip_off); - if (ip->ip_id == 0) - ip->ip_id = ip_newid(); + if (ip->ip_id != 0 || m->m_pkthdr.len < IP_MINFRAGSIZE) + flags |= IP_NOIPNEWID; opts = NULL; /* XXX prevent ip_output from overwriting header fields */ flags |= IP_RAWOUTPUT; - ipstat.ips_rawout++; + IP_STATINC(IP_STAT_RAWOUT); } return (ip_output(m, opts, &inp->inp_route, flags, inp->inp_moptions, inp->inp_socket, &inp->inp_errormtu)); @@ -386,26 +386,34 @@ rip_ctloutput(int op, struct socket *so, struct inpcb *inp = sotoinpcb(so); int error = 0; - if (level != IPPROTO_IP) { - error = ENOPROTOOPT; - if (op == PRCO_SETOPT && *m != 0) - (void) m_free(*m); - } else switch (op) { + if (level == SOL_SOCKET && optname == SO_NOHEADER) { + if (op == PRCO_GETOPT) { + *m = m_intopt(so, + (inp->inp_flags & INP_NOHEADER) ? 1 : 0); + return 0; + } else if (*m == NULL || (*m)->m_len != sizeof(int)) + error = EINVAL; + else if (*mtod(*m, int *)) { + inp->inp_flags &= ~INP_HDRINCL; + inp->inp_flags |= INP_NOHEADER; + } else + inp->inp_flags &= ~INP_NOHEADER; + goto free_m; + } else if (level != IPPROTO_IP) + return ip_ctloutput(op, so, level, optname, m); + + switch (op) { case PRCO_SETOPT: switch (optname) { case IP_HDRINCL: - if (*m == 0 || (*m)->m_len < sizeof (int)) + if (*m == NULL || (*m)->m_len != sizeof(int)) error = EINVAL; - else { - if (*mtod(*m, int *)) - inp->inp_flags |= INP_HDRINCL; - else - inp->inp_flags &= ~INP_HDRINCL; - } - if (*m != 0) - (void) m_free(*m); - break; + else if (*mtod(*m, int *)) + inp->inp_flags |= INP_HDRINCL; + else + inp->inp_flags &= ~INP_HDRINCL; + goto free_m; #ifdef MROUTING case MRT_INIT: @@ -431,10 +439,7 @@ rip_ctloutput(int op, struct socket *so, case PRCO_GETOPT: switch (optname) { case IP_HDRINCL: - *m = m_get(M_WAIT, MT_SOOPTS); - MCLAIM((*m), so->so_mowner); - (*m)->m_len = sizeof (int); - *mtod(*m, int *) = inp->inp_flags & INP_HDRINCL ? 1 : 0; + *m = m_intopt(so, inp->inp_flags & INP_HDRINCL ? 1 : 0); break; #ifdef MROUTING @@ -452,7 +457,11 @@ rip_ctloutput(int op, struct socket *so, } break; } - return (error); + return error; +free_m: + if (op == PRCO_SETOPT && *m != NULL) + (void)m_free(*m); + return error; } int @@ -506,26 +515,28 @@ rip_usrreq(struct socket *so, int req, struct mbuf *m, struct mbuf *nam, struct mbuf *control, struct lwp *l) { struct inpcb *inp; - struct proc *p; int s; int error = 0; #ifdef MROUTING extern struct socket *ip_mrouter; #endif - p = l ? l->l_proc : NULL; if (req == PRU_CONTROL) - return (in_control(so, (long)m, (caddr_t)nam, - (struct ifnet *)control, p)); + return (in_control(so, (long)m, (void *)nam, + (struct ifnet *)control, l)); + + s = splsoftnet(); if (req == PRU_PURGEIF) { + mutex_enter(softnet_lock); in_pcbpurgeif0(&rawcbtable, (struct ifnet *)control); in_purgeif((struct ifnet *)control); in_pcbpurgeif(&rawcbtable, (struct ifnet *)control); + mutex_exit(softnet_lock); + splx(s); return (0); } - s = splsoftnet(); inp = sotoinpcb(so); #ifdef DIAGNOSTIC if (req != PRU_SEND && req != PRU_SENDOOB && control) @@ -539,15 +550,19 @@ rip_usrreq(struct socket *so, int req, switch (req) { case PRU_ATTACH: + sosetlock(so); if (inp != 0) { error = EISCONN; break; } - if (p == 0 || (error = kauth_authorize_generic(p->p_cred, - KAUTH_GENERIC_ISSUSER, &p->p_acflag))) { + + if (l == NULL) { error = EACCES; break; } + + /* XXX: raw socket permissions are checked in socreate() */ + if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { error = soreserve(so, rip_sendspace, rip_recvspace); if (error)