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 1.105 retrieving revision 1.125 diff -u -p -r1.105 -r1.125 --- src/sys/netinet/raw_ip.c 2008/04/12 05:58:22 1.105 +++ src/sys/netinet/raw_ip.c 2014/06/22 08:10:18 1.125 @@ -1,4 +1,4 @@ -/* $NetBSD: raw_ip.c,v 1.105 2008/04/12 05:58:22 thorpej Exp $ */ +/* $NetBSD: raw_ip.c,v 1.125 2014/06/22 08:10:18 rtr Exp $ */ /* * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. @@ -60,10 +60,15 @@ * @(#)raw_ip.c 8.7 (Berkeley) 5/15/95 */ +/* + * Raw interface to IP protocol. + */ + #include -__KERNEL_RCSID(0, "$NetBSD: raw_ip.c,v 1.105 2008/04/12 05:58:22 thorpej Exp $"); +__KERNEL_RCSID(0, "$NetBSD: raw_ip.c,v 1.125 2014/06/22 08:10:18 rtr Exp $"); #include "opt_inet.h" +#include "opt_compat_netbsd.h" #include "opt_ipsec.h" #include "opt_mrouting.h" @@ -93,16 +98,15 @@ __KERNEL_RCSID(0, "$NetBSD: raw_ip.c,v 1 #include #include -#include - #ifdef IPSEC -#include -#endif /*IPSEC*/ - -#ifdef FAST_IPSEC #include -#include /* XXX ipsecstat namespace */ -#endif /* FAST_IPSEC*/ +#include +#include +#endif /* IPSEC */ + +#ifdef COMPAT_50 +#include +#endif struct inpcbtable rawcbtable; @@ -112,12 +116,17 @@ int rip_bind(struct inpcb *, struct mbu int rip_connect(struct inpcb *, struct mbuf *); void rip_disconnect(struct inpcb *); +static void sysctl_net_inet_raw_setup(struct sysctllog **); + /* * Nominal space allocated to a raw ip socket. */ #define RIPSNDQ 8192 #define RIPRCVQ 8192 +static u_long rip_sendspace = RIPSNDQ; +static u_long rip_recvspace = RIPRCVQ; + /* * Raw interface to IP protocol. */ @@ -129,6 +138,7 @@ void rip_init(void) { + sysctl_net_inet_raw_setup(NULL); in_pcbinit(&rawcbtable, 1, 1); } @@ -138,8 +148,11 @@ rip_sbappendaddr(struct inpcb *last, str { if (last->inp_flags & INP_NOHEADER) m_adj(n, hlen); - if (last->inp_flags & INP_CONTROLOPTS || - last->inp_socket->so_options & SO_TIMESTAMP) + if (last->inp_flags & INP_CONTROLOPTS +#ifdef SO_OTIMESTAMP + || last->inp_socket->so_options & SO_OTIMESTAMP +#endif + || 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 */ @@ -183,7 +196,7 @@ rip_input(struct mbuf *m, ...) ip->ip_len = ntohs(ip->ip_len) - hlen; NTOHS(ip->ip_off); - CIRCLEQ_FOREACH(inph, &rawcbtable.inpt_queue, inph_queue) { + TAILQ_FOREACH(inph, &rawcbtable.inpt_queue, inph_queue) { inp = (struct inpcb *)inph; if (inp->inp_af != AF_INET) continue; @@ -197,10 +210,11 @@ rip_input(struct mbuf *m, ...) continue; if (last == NULL) ; -#if defined(IPSEC) || defined(FAST_IPSEC) +#if defined(IPSEC) /* check AH/ESP integrity. */ - else if (ipsec4_in_reject_so(m, last->inp_socket)) { - ipsecstat.in_polvio++; + else if (ipsec_used && + ipsec4_in_reject_so(m, last->inp_socket)) { + IPSEC_STATINC(IPSEC_STAT_IN_POLVIO); /* do not inject data to pcb */ } #endif /*IPSEC*/ @@ -211,11 +225,12 @@ rip_input(struct mbuf *m, ...) } last = inp; } -#if defined(IPSEC) || defined(FAST_IPSEC) +#if defined(IPSEC) /* check AH/ESP integrity. */ - if (last != NULL && ipsec4_in_reject_so(m, last->inp_socket)) { + if (ipsec_used && last != NULL + && ipsec4_in_reject_so(m, last->inp_socket)) { m_freem(m); - ipsecstat.in_polvio++; + IPSEC_STATINC(IPSEC_STAT_IN_POLVIO); IP_STATDEC(IP_STAT_DELIVERED); /* do not inject data to pcb */ } else @@ -241,14 +256,12 @@ rip_pcbnotify(struct inpcbtable *table, struct in_addr faddr, struct in_addr laddr, int proto, int errno, void (*notify)(struct inpcb *, int)) { - struct inpcb *inp, *ninp; + struct inpcb_hdr *inph, *ninph; int nmatch; nmatch = 0; - for (inp = (struct inpcb *)CIRCLEQ_FIRST(&table->inpt_queue); - inp != (struct inpcb *)&table->inpt_queue; - inp = ninp) { - ninp = (struct inpcb *)inp->inp_queue.cqe_next; + TAILQ_FOREACH_SAFE(inph, &table->inpt_queue, inph_queue, ninph) { + struct inpcb *inp = (struct inpcb *)inph; if (inp->inp_af != AF_INET) continue; if (inp->inp_ip.ip_p && inp->inp_ip.ip_p != proto) @@ -370,48 +383,56 @@ rip_output(struct mbuf *m, ...) flags |= IP_RAWOUTPUT; IP_STATINC(IP_STAT_RAWOUT); } - return (ip_output(m, opts, &inp->inp_route, flags, inp->inp_moptions, - inp->inp_socket, &inp->inp_errormtu)); + + /* + * IP output. Note: if IP_RETURNMTU flag is set, the MTU size + * will be stored in inp_errormtu. + */ + return ip_output(m, opts, &inp->inp_route, flags, inp->inp_moptions, + inp->inp_socket); } /* * Raw IP socket option processing. */ int -rip_ctloutput(int op, struct socket *so, int level, int optname, - struct mbuf **m) +rip_ctloutput(int op, struct socket *so, struct sockopt *sopt) { struct inpcb *inp = sotoinpcb(so); int error = 0; + int optval; - if (level == SOL_SOCKET && optname == SO_NOHEADER) { + if (sopt->sopt_level == SOL_SOCKET && sopt->sopt_name == 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); + optval = (inp->inp_flags & INP_NOHEADER) ? 1 : 0; + error = sockopt_set(sopt, &optval, sizeof(optval)); + } else if (op == PRCO_SETOPT) { + error = sockopt_getint(sopt, &optval); + if (error) + goto out; + if (optval) { + inp->inp_flags &= ~INP_HDRINCL; + inp->inp_flags |= INP_NOHEADER; + } else + inp->inp_flags &= ~INP_NOHEADER; + } + goto out; + } else if (sopt->sopt_level != IPPROTO_IP) + return ip_ctloutput(op, so, sopt); switch (op) { case PRCO_SETOPT: - switch (optname) { + switch (sopt->sopt_name) { case IP_HDRINCL: - if (*m == NULL || (*m)->m_len != sizeof(int)) - error = EINVAL; - else if (*mtod(*m, int *)) + error = sockopt_getint(sopt, &optval); + if (error) + break; + if (optval) inp->inp_flags |= INP_HDRINCL; else inp->inp_flags &= ~INP_HDRINCL; - goto free_m; + break; #ifdef MROUTING case MRT_INIT: @@ -424,20 +445,21 @@ rip_ctloutput(int op, struct socket *so, case MRT_API_CONFIG: case MRT_ADD_BW_UPCALL: case MRT_DEL_BW_UPCALL: - error = ip_mrouter_set(so, optname, m); + error = ip_mrouter_set(so, sopt); break; #endif default: - error = ip_ctloutput(op, so, level, optname, m); + error = ip_ctloutput(op, so, sopt); break; } break; case PRCO_GETOPT: - switch (optname) { + switch (sopt->sopt_name) { case IP_HDRINCL: - *m = m_intopt(so, inp->inp_flags & INP_HDRINCL ? 1 : 0); + optval = inp->inp_flags & INP_HDRINCL; + error = sockopt_set(sopt, &optval, sizeof(optval)); break; #ifdef MROUTING @@ -445,20 +467,17 @@ rip_ctloutput(int op, struct socket *so, case MRT_ASSERT: case MRT_API_SUPPORT: case MRT_API_CONFIG: - error = ip_mrouter_get(so, optname, m); + error = ip_mrouter_get(so, sopt); break; #endif default: - error = ip_ctloutput(op, so, level, optname, m); + error = ip_ctloutput(op, so, sopt); break; } break; } - return error; -free_m: - if (op == PRCO_SETOPT && *m != NULL) - (void)m_free(*m); + out: return error; } @@ -469,10 +488,9 @@ rip_bind(struct inpcb *inp, struct mbuf if (nam->m_len != sizeof(*addr)) return (EINVAL); - if (TAILQ_FIRST(&ifnet) == 0) + if (!IFNET_FIRST()) return (EADDRNOTAVAIL); - if (addr->sin_family != AF_INET && - addr->sin_family != AF_IMPLINK) + if (addr->sin_family != AF_INET) return (EAFNOSUPPORT); if (!in_nullhost(addr->sin_addr) && ifa_ifwithaddr(sintosa(addr)) == 0) @@ -488,10 +506,9 @@ rip_connect(struct inpcb *inp, struct mb if (nam->m_len != sizeof(*addr)) return (EINVAL); - if (TAILQ_FIRST(&ifnet) == 0) + if (!IFNET_FIRST()) return (EADDRNOTAVAIL); - if (addr->sin_family != AF_INET && - addr->sin_family != AF_IMPLINK) + if (addr->sin_family != AF_INET) return (EAFNOSUPPORT); inp->inp_faddr = addr->sin_addr; return (0); @@ -504,80 +521,91 @@ rip_disconnect(struct inpcb *inp) inp->inp_faddr = zeroin_addr; } -u_long rip_sendspace = RIPSNDQ; -u_long rip_recvspace = RIPRCVQ; +static int +rip_attach(struct socket *so, int proto) +{ + struct inpcb *inp; + int error; -/*ARGSUSED*/ -int -rip_usrreq(struct socket *so, int req, - struct mbuf *m, struct mbuf *nam, struct mbuf *control, struct lwp *l) + KASSERT(sotoinpcb(so) == NULL); + sosetlock(so); + + if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { + error = soreserve(so, rip_sendspace, rip_recvspace); + if (error) { + return error; + } + } + + error = in_pcballoc(so, &rawcbtable); + if (error) { + return error; + } + inp = sotoinpcb(so); + inp->inp_ip.ip_p = proto; + KASSERT(solocked(so)); + + return 0; +} + +static void +rip_detach(struct socket *so) { struct inpcb *inp; - int s; - int error = 0; + + KASSERT(solocked(so)); + inp = sotoinpcb(so); + KASSERT(inp != NULL); + #ifdef MROUTING extern struct socket *ip_mrouter; + if (so == ip_mrouter) { + ip_mrouter_done(); + } #endif + in_pcbdetach(inp); +} - if (req == PRU_CONTROL) - return (in_control(so, (long)m, (void *)nam, - (struct ifnet *)control, l)); +static int +rip_ioctl(struct socket *so, struct mbuf *m, struct mbuf *nam, + struct mbuf *control, struct lwp *l) +{ + return in_control(so, (long)m, nam, (ifnet_t *)control, l); +} - s = splsoftnet(); +int +rip_usrreq(struct socket *so, int req, struct mbuf *m, struct mbuf *nam, + struct mbuf *control, struct lwp *l) +{ + struct inpcb *inp; + int s, error = 0; + + KASSERT(req != PRU_ATTACH); + KASSERT(req != PRU_DETACH); + KASSERT(req != PRU_CONTROL); + 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); + return 0; } + KASSERT(solocked(so)); inp = sotoinpcb(so); -#ifdef DIAGNOSTIC - if (req != PRU_SEND && req != PRU_SENDOOB && control) - panic("rip_usrreq: unexpected control mbuf"); -#endif - if (inp == 0 && req != PRU_ATTACH) { - error = EINVAL; - goto release; + + KASSERT(!control || (req == PRU_SEND || req == PRU_SENDOOB)); + if (inp == NULL) { + splx(s); + return EINVAL; } switch (req) { - case PRU_ATTACH: - if (inp != 0) { - error = EISCONN; - break; - } - - 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) - break; - } - error = in_pcballoc(so, &rawcbtable); - if (error) - break; - inp = sotoinpcb(so); - inp->inp_ip.ip_p = (long)nam; - break; - - case PRU_DETACH: -#ifdef MROUTING - if (so == ip_mrouter) - ip_mrouter_done(); -#endif - in_pcbdetach(inp); - break; - case PRU_BIND: error = rip_bind(inp, nam); break; @@ -676,22 +704,30 @@ rip_usrreq(struct socket *so, int req, default: panic("rip_usrreq"); } - -release: splx(s); - return (error); + + return error; } -SYSCTL_SETUP(sysctl_net_inet_raw_setup, "sysctl net.inet.raw subtree setup") +PR_WRAP_USRREQS(rip) +#define rip_attach rip_attach_wrapper +#define rip_detach rip_detach_wrapper +#define rip_ioctl rip_ioctl_wrapper +#define rip_usrreq rip_usrreq_wrapper + +const struct pr_usrreqs rip_usrreqs = { + .pr_attach = rip_attach, + .pr_detach = rip_detach, + .pr_ioctl = rip_ioctl, + .pr_generic = rip_usrreq, +}; + +static void +sysctl_net_inet_raw_setup(struct sysctllog **clog) { sysctl_createv(clog, 0, NULL, NULL, CTLFLAG_PERMANENT, - CTLTYPE_NODE, "net", NULL, - NULL, 0, NULL, 0, - CTL_NET, CTL_EOL); - sysctl_createv(clog, 0, NULL, NULL, - CTLFLAG_PERMANENT, CTLTYPE_NODE, "inet", NULL, NULL, 0, NULL, 0, CTL_NET, PF_INET, CTL_EOL);