/* $NetBSD: ip_state.c,v 1.40 2002/06/09 16:33:42 itojun Exp $ */ /* * Copyright (C) 1995-2002 by Darren Reed. * * See the IPFILTER.LICENCE file for details on licencing. */ #ifdef __sgi # include #endif #include #include #include #include #if defined(__NetBSD__) && (NetBSD >= 199905) && !defined(IPFILTER_LKM) && \ defined(_KERNEL) # include "opt_ipfilter_log.h" #endif #if defined(_KERNEL) && defined(__FreeBSD_version) && \ (__FreeBSD_version >= 400000) && !defined(KLD_MODULE) #include "opt_inet6.h" #endif #if !defined(_KERNEL) && !defined(KERNEL) && !defined(__KERNEL__) # include # include # include #else # ifdef linux # include # include # endif #endif #if (defined(KERNEL) || defined(_KERNEL)) && (__FreeBSD_version >= 220000) # include # include # if (__FreeBSD_version >= 300000) && !defined(IPFILTER_LKM) # include "opt_ipfilter.h" # endif #else # include #endif #include #ifndef linux # include #endif #include #if (defined(_KERNEL) || defined(KERNEL)) && !defined(linux) # include #endif #if !defined(__SVR4) && !defined(__svr4__) # ifndef linux # include # endif #else # include # include # ifdef _KERNEL # include # endif # include # include #endif #include #ifdef sun # include #endif #include #include #include #include #include #ifndef linux # include # include #endif #include #include #include "netinet/ip_compat.h" #include #include "netinet/ip_fil.h" #include "netinet/ip_nat.h" #include "netinet/ip_frag.h" #include "netinet/ip_state.h" #ifdef USE_INET6 #include #endif #if (__FreeBSD_version >= 300000) # include # if (defined(_KERNEL) || defined(KERNEL)) && !defined(IPFILTER_LKM) # include # include # endif #endif #if !defined(lint) #if defined(__NetBSD__) #include __KERNEL_RCSID(0, "$NetBSD: ip_state.c,v 1.40 2002/06/09 16:33:42 itojun Exp $"); #else static const char sccsid[] = "@(#)ip_state.c 1.8 6/5/96 (C) 1993-2000 Darren Reed"; static const char rcsid[] = "@(#)Id: ip_state.c,v 2.30.2.70 2002/04/27 16:06:15 darrenr Exp"; #endif #endif #ifndef MIN # define MIN(a,b) (((a)<(b))?(a):(b)) #endif #define TCP_CLOSE (TH_FIN|TH_RST) static ipstate_t **ips_table = NULL; static int ips_num = 0; static int ips_wild = 0; static ips_stat_t ips_stats; #if (SOLARIS || defined(__sgi)) && defined(_KERNEL) extern KRWLOCK_T ipf_state, ipf_mutex; extern kmutex_t ipf_rw; #endif #ifdef USE_INET6 static frentry_t *fr_checkicmp6matchingstate __P((ip6_t *, fr_info_t *)); #endif static int fr_matchsrcdst __P((ipstate_t *, union i6addr, union i6addr, fr_info_t *, tcphdr_t *)); static frentry_t *fr_checkicmpmatchingstate __P((ip_t *, fr_info_t *)); static int fr_matchicmpqueryreply __P((int, ipstate_t *, icmphdr_t *)); static int fr_state_flush __P((int)); static ips_stat_t *fr_statetstats __P((void)); static void fr_delstate __P((ipstate_t *)); static int fr_state_remove __P((caddr_t)); static void fr_ipsmove __P((ipstate_t **, ipstate_t *, u_int)); static int fr_tcpoptions __P((tcphdr_t *)); int fr_stputent __P((caddr_t)); int fr_stgetent __P((caddr_t)); void fr_stinsert __P((ipstate_t *)); #define FIVE_DAYS (2 * 5 * 86400) /* 5 days: half closed session */ #define TCP_MSL 240 /* 2 minutes */ u_long fr_tcpidletimeout = FIVE_DAYS, fr_tcpclosewait = 2 * TCP_MSL, fr_tcplastack = 2 * TCP_MSL, fr_tcptimeout = 2 * TCP_MSL, fr_tcpclosed = 120, fr_tcphalfclosed = 2 * 2 * 3600, /* 2 hours */ fr_udptimeout = 240, fr_udpacktimeout = 24, fr_icmptimeout = 120, fr_icmpacktimeout = 12; int fr_statemax = IPSTATE_MAX, fr_statesize = IPSTATE_SIZE; int fr_state_doflush = 0, fr_state_lock = 0; ipstate_t *ips_list = NULL; static int icmpreplytype4[ICMP_MAXTYPE + 1]; #ifdef USE_INET6 static int icmpreplytype6[ICMP6_MAXTYPE + 1]; #endif int fr_stateinit() { int i; KMALLOCS(ips_table, ipstate_t **, fr_statesize * sizeof(ipstate_t *)); if (ips_table != NULL) bzero((char *)ips_table, fr_statesize * sizeof(ipstate_t *)); else return -1; /* fill icmp reply type table */ for (i = 0; i <= ICMP_MAXTYPE; i++) icmpreplytype4[i] = -1; icmpreplytype4[ICMP_ECHO] = ICMP_ECHOREPLY; icmpreplytype4[ICMP_TSTAMP] = ICMP_TSTAMPREPLY; icmpreplytype4[ICMP_IREQ] = ICMP_IREQREPLY; icmpreplytype4[ICMP_MASKREQ] = ICMP_MASKREPLY; #ifdef USE_INET6 /* fill icmp reply type table */ for (i = 0; i <= ICMP6_MAXTYPE; i++) icmpreplytype6[i] = -1; icmpreplytype6[ICMP6_ECHO_REQUEST] = ICMP6_ECHO_REPLY; icmpreplytype6[ICMP6_MEMBERSHIP_QUERY] = ICMP6_MEMBERSHIP_REPORT; icmpreplytype6[ICMP6_NI_QUERY] = ICMP6_NI_REPLY; icmpreplytype6[ND_ROUTER_SOLICIT] = ND_ROUTER_ADVERT; icmpreplytype6[ND_NEIGHBOR_SOLICIT] = ND_NEIGHBOR_ADVERT; #endif return 0; } static ips_stat_t *fr_statetstats() { ips_stats.iss_active = ips_num; ips_stats.iss_table = ips_table; ips_stats.iss_list = ips_list; return &ips_stats; } /* * flush state tables. two actions currently defined: * which == 0 : flush all state table entries * which == 1 : flush TCP connections which have started to close but are * stuck for some reason. * which == 2 : flush TCP connections which have been idle for a long time, * starting at > 4 days idle and working back in successive half- * days to at most 12 hours old. */ static int fr_state_flush(which) int which; { ipstate_t *is, **isp; #if defined(_KERNEL) && !SOLARIS int s; #endif int delete, removed = 0, try; SPL_NET(s); for (isp = &ips_list; (is = *isp); ) { delete = 0; switch (which) { case 0 : delete = 1; break; case 1 : case 2 : if (is->is_p != IPPROTO_TCP) break; if ((is->is_state[0] != TCPS_ESTABLISHED) || (is->is_state[1] != TCPS_ESTABLISHED)) delete = 1; break; } if (delete) { if (is->is_p == IPPROTO_TCP) ips_stats.iss_fin++; else ips_stats.iss_expire++; #ifdef IPFILTER_LOG ipstate_log(is, ISL_FLUSH); #endif fr_delstate(is); removed++; } else isp = &is->is_next; } /* * Asked to remove inactive entries, try again if first attempt * failed. In this case, 86400 is half a day because the counter is * activated every half second. */ if ((which == 2) && (removed == 0)) { try = 86400; /* half a day */ for (; (try < FIVE_DAYS) && (removed == 0); try += 86400) { for (isp = &ips_list; (is = *isp); ) { delete = 0; if ((is->is_p == IPPROTO_TCP) && ((is->is_state[0] == TCPS_ESTABLISHED) || (is->is_state[1] == TCPS_ESTABLISHED)) && (is->is_age < try)) { ips_stats.iss_fin++; delete = 1; } else if ((is->is_p != IPPROTO_TCP) && (is->is_pkts > 1)) { ips_stats.iss_expire++; delete = 1; } if (delete) { #ifdef IPFILTER_LOG ipstate_log(is, ISL_FLUSH); #endif fr_delstate(is); removed++; } else isp = &is->is_next; } } } SPL_X(s); return removed; } static int fr_state_remove(data) caddr_t data; { ipstate_t *sp, st; int error; sp = &st; error = IRCOPYPTR(data, (caddr_t)&st, sizeof(st)); if (error) return EFAULT; WRITE_ENTER(&ipf_state); for (sp = ips_list; sp; sp = sp->is_next) if ((sp->is_p == st.is_p) && (sp->is_v == st.is_v) && !bcmp((char *)&sp->is_src, (char *)&st.is_src, sizeof(st.is_src)) && !bcmp((char *)&sp->is_dst, (char *)&st.is_dst, sizeof(st.is_dst)) && !bcmp((char *)&sp->is_ps, (char *)&st.is_ps, sizeof(st.is_ps))) { #ifdef IPFILTER_LOG ipstate_log(sp, ISL_REMOVE); #endif fr_delstate(sp); RWLOCK_EXIT(&ipf_state); return 0; } RWLOCK_EXIT(&ipf_state); return ESRCH; } int fr_state_ioctl(data, cmd, mode) caddr_t data; #if defined(__NetBSD__) || defined(__OpenBSD__) u_long cmd; #else int cmd; #endif int mode; { int arg, ret, error = 0; switch (cmd) { case SIOCDELST : error = fr_state_remove(data); break; case SIOCIPFFL : error = IRCOPY(data, (caddr_t)&arg, sizeof(arg)); if (error) break; if (arg == 0 || arg == 1) { WRITE_ENTER(&ipf_state); ret = fr_state_flush(arg); RWLOCK_EXIT(&ipf_state); error = IWCOPY((caddr_t)&ret, data, sizeof(ret)); } else error = EINVAL; break; #ifdef IPFILTER_LOG case SIOCIPFFB : if (!(mode & FWRITE)) error = EPERM; else { int tmp; tmp = ipflog_clear(IPL_LOGSTATE); IWCOPY((char *)&tmp, data, sizeof(tmp)); } break; #endif case SIOCGETFS : error = IWCOPYPTR((caddr_t)fr_statetstats(), data, sizeof(ips_stat_t)); break; case FIONREAD : #ifdef IPFILTER_LOG arg = (int)iplused[IPL_LOGSTATE]; error = IWCOPY((caddr_t)&arg, (caddr_t)data, sizeof(arg)); #endif break; case SIOCSTLCK : error = fr_lock(data, &fr_state_lock); break; case SIOCSTPUT : if (!fr_state_lock) { error = EACCES; break; } error = fr_stputent(data); break; case SIOCSTGET : if (!fr_state_lock) { error = EACCES; break; } error = fr_stgetent(data); break; default : error = EINVAL; break; } return error; } /* * Copy out state information from the kernel to a user space process. */ int fr_stgetent(data) caddr_t data; { register ipstate_t *is, *isn; ipstate_save_t ips; int error; error = IRCOPYPTR(data, (caddr_t)&ips, sizeof(ips)); if (error) return error; isn = ips.ips_next; if (!isn) { isn = ips_list; if (isn == NULL) { if (ips.ips_next == NULL) return ENOENT; return 0; } } else { /* * Make sure the pointer we're copying from exists in the * current list of entries. Security precaution to prevent * copying of random kernel data. */ for (is = ips_list; is; is = is->is_next) if (is == isn) break; if (!is) return ESRCH; } ips.ips_next = isn->is_next; bcopy((char *)isn, (char *)&ips.ips_is, sizeof(ips.ips_is)); if (isn->is_rule) bcopy((char *)isn->is_rule, (char *)&ips.ips_fr, sizeof(ips.ips_fr)); error = IWCOPYPTR((caddr_t)&ips, data, sizeof(ips)); if (error) error = EFAULT; return error; } int fr_stputent(data) caddr_t data; { register ipstate_t *is, *isn; ipstate_save_t ips; int error, out, i; frentry_t *fr; char *name; error = IRCOPYPTR(data, (caddr_t)&ips, sizeof(ips)); if (error) return error; KMALLOC(isn, ipstate_t *); if (isn == NULL) return ENOMEM; bcopy((char *)&ips.ips_is, (char *)isn, sizeof(*isn)); fr = isn->is_rule; if (fr != NULL) { if (isn->is_flags & FI_NEWFR) { KMALLOC(fr, frentry_t *); if (fr == NULL) { KFREE(isn); return ENOMEM; } bcopy((char *)&ips.ips_fr, (char *)fr, sizeof(*fr)); out = fr->fr_flags & FR_OUTQUE ? 1 : 0; isn->is_rule = fr; ips.ips_is.is_rule = fr; /* * Look up all the interface names in the rule. */ for (i = 0; i < 4; i++) { name = fr->fr_ifnames[i]; if ((name[1] == '\0') && ((name[0] == '-') || (name[0] == '*'))) { fr->fr_ifas[i] = NULL; } else if (*name != '\0') { fr->fr_ifas[i] = GETUNIT(name, fr->fr_v); if (fr->fr_ifas[i] == NULL) fr->fr_ifas[i] = (void *)-1; else { strncpy(isn->is_ifname[i], IFNAME(fr->fr_ifas[i]), IFNAMSIZ); } } isn->is_ifp[out] = fr->fr_ifas[i]; } /* * send a copy back to userland of what we ended up * to allow for verification. */ error = IWCOPYPTR((caddr_t)&ips, data, sizeof(ips)); if (error) { KFREE(isn); KFREE(fr); return EFAULT; } } else { for (is = ips_list; is; is = is->is_next) if (is->is_rule == fr) break; if (!is) { KFREE(isn); return ESRCH; } } } fr_stinsert(isn); return 0; } /* * Insert a state table entry manually. */ void fr_stinsert(is) register ipstate_t *is; { register u_int hv = is->is_hv; char *name; int i; MUTEX_INIT(&is->is_lock, "ipf state entry", NULL); /* * Look up all the interface names in the state entry. */ for (i = 0; i < 4; i++) { name = is->is_ifname[i]; if ((name[1] == '\0') && ((name[0] == '-') || (name[0] == '*'))) { is->is_ifp[0] = NULL; } else if (*name != '\0') { is->is_ifp[i] = GETUNIT(name, is->is_v); if (is->is_ifp[i] == NULL) is->is_ifp[i] = (void *)-1; } } /* * add into list table. */ if (ips_list) ips_list->is_pnext = &is->is_next; is->is_pnext = &ips_list; is->is_next = ips_list; ips_list = is; if (ips_table[hv]) ips_table[hv]->is_phnext = &is->is_hnext; else ips_stats.iss_inuse++; is->is_phnext = ips_table + hv; is->is_hnext = ips_table[hv]; ips_table[hv] = is; ips_num++; } /* * Create a new ipstate structure and hang it off the hash table. */ ipstate_t *fr_addstate(ip, fin, stsave, flags) ip_t *ip; fr_info_t *fin; ipstate_t **stsave; u_int flags; { register tcphdr_t *tcp = NULL; register ipstate_t *is; register u_int hv; struct icmp *ic; ipstate_t ips; u_int pass; void *ifp; int out; if (fr_state_lock || (fin->fin_off != 0) || (fin->fin_fl & FI_SHORT) || (fin->fin_misc & FM_BADSTATE)) return NULL; if (ips_num == fr_statemax) { ips_stats.iss_max++; fr_state_doflush = 1; return NULL; } out = fin->fin_out; is = &ips; bzero((char *)is, sizeof(*is)); ips.is_age = 1; /* * Copy and calculate... */ hv = (is->is_p = fin->fin_fi.fi_p); is->is_src = fin->fin_fi.fi_src; hv += is->is_saddr; is->is_dst = fin->fin_fi.fi_dst; hv += is->is_daddr; #ifdef USE_INET6 if (fin->fin_v == 6) { if ((is->is_p == IPPROTO_ICMPV6) && IN6_IS_ADDR_MULTICAST(&is->is_dst.in6)) { /* * So you can do keep state with neighbour discovery. */ flags |= FI_W_DADDR; hv -= is->is_daddr; } else { hv += is->is_dst.i6[1]; hv += is->is_dst.i6[2]; hv += is->is_dst.i6[3]; } hv += is->is_src.i6[1]; hv += is->is_src.i6[2]; hv += is->is_src.i6[3]; } #endif switch (is->is_p) { int off; #ifdef USE_INET6 case IPPROTO_ICMPV6 : ic = (struct icmp *)fin->fin_dp; if ((ic->icmp_type & ICMP6_INFOMSG_MASK) == 0) return NULL; switch (ic->icmp_type) { case ICMP6_ECHO_REQUEST : is->is_icmp.ics_type = ic->icmp_type; hv += (is->is_icmp.ics_id = ic->icmp_id); hv += (is->is_icmp.ics_seq = ic->icmp_seq); break; case ICMP6_MEMBERSHIP_QUERY : case ND_ROUTER_SOLICIT : case ND_NEIGHBOR_SOLICIT : case ICMP6_NI_QUERY : is->is_icmp.ics_type = ic->icmp_type; break; default : return NULL; } ATOMIC_INCL(ips_stats.iss_icmp); is->is_age = fr_icmptimeout; break; #endif case IPPROTO_ICMP : ic = (struct icmp *)fin->fin_dp; switch (ic->icmp_type) { case ICMP_ECHO : case ICMP_TSTAMP : case ICMP_IREQ : case ICMP_MASKREQ : is->is_icmp.ics_type = ic->icmp_type; hv += (is->is_icmp.ics_id = ic->icmp_id); hv += (is->is_icmp.ics_seq = ic->icmp_seq); break; default : return NULL; } ATOMIC_INCL(ips_stats.iss_icmp); is->is_age = fr_icmptimeout; break; case IPPROTO_TCP : tcp = (tcphdr_t *)fin->fin_dp; if (tcp->th_flags & TH_RST) return NULL; /* * The endian of the ports doesn't matter, but the ack and * sequence numbers do as we do mathematics on them later. */ is->is_sport = htons(fin->fin_data[0]); is->is_dport = htons(fin->fin_data[1]); if ((flags & (FI_W_DPORT|FI_W_SPORT)) == 0) { hv += is->is_sport; hv += is->is_dport; } if ((flags & FI_IGNOREPKT) == 0) { is->is_send = ntohl(tcp->th_seq) + fin->fin_dlen - (off = (tcp->th_off << 2)) + ((tcp->th_flags & TH_SYN) ? 1 : 0) + ((tcp->th_flags & TH_FIN) ? 1 : 0); is->is_maxsend = is->is_send; if ((tcp->th_flags & TH_SYN) && ((tcp->th_off << 2) >= (sizeof(*tcp) + 4))) { int wscale = fr_tcpoptions(tcp); if (wscale >= 0) is->is_swscale = wscale; } } is->is_maxdwin = 1; is->is_maxswin = ntohs(tcp->th_win); if (is->is_maxswin == 0) is->is_maxswin = 1; if ((tcp->th_flags & TH_OPENING) == TH_SYN) is->is_fsm = 1; /* * If we're creating state for a starting connection, start the * timer on it as we'll never see an error if it fails to * connect. */ ATOMIC_INCL(ips_stats.iss_tcp); break; case IPPROTO_UDP : tcp = (tcphdr_t *)fin->fin_dp; is->is_sport = htons(fin->fin_data[0]); is->is_dport = htons(fin->fin_data[1]); if ((flags & (FI_W_DPORT|FI_W_SPORT)) == 0) { hv += is->is_sport; hv += is->is_dport; } ATOMIC_INCL(ips_stats.iss_udp); is->is_age = fr_udptimeout; break; default : is->is_age = fr_udptimeout; break; } KMALLOC(is, ipstate_t *); if (is == NULL) { ATOMIC_INCL(ips_stats.iss_nomem); return NULL; } bcopy((char *)&ips, (char *)is, sizeof(*is)); hv %= fr_statesize; is->is_hv = hv; is->is_rule = fin->fin_fr; if (is->is_rule != NULL) { ATOMIC_INC32(is->is_rule->fr_ref); pass = is->is_rule->fr_flags; is->is_frage[0] = is->is_rule->fr_age[0]; is->is_frage[1] = is->is_rule->fr_age[1]; if (is->is_frage[0] != 0) is->is_age = is->is_frage[0]; is->is_ifp[(out << 1) + 1] = is->is_rule->fr_ifas[1]; is->is_ifp[(1 - out) << 1] = is->is_rule->fr_ifas[2]; is->is_ifp[((1 - out) << 1) + 1] = is->is_rule->fr_ifas[3]; if (((ifp = is->is_rule->fr_ifas[1]) != NULL) && (ifp != (void *)-1)) strncpy(is->is_ifname[(out << 1) + 1], IFNAME(ifp), IFNAMSIZ); if (((ifp = is->is_rule->fr_ifas[2]) != NULL) && (ifp != (void *)-1)) strncpy(is->is_ifname[(1 - out) << 1], IFNAME(ifp), IFNAMSIZ); if (((ifp = is->is_rule->fr_ifas[3]) != NULL) && (ifp != (void *)-1)) strncpy(is->is_ifname[((1 - out) << 1) + 1], IFNAME(ifp), IFNAMSIZ); } else pass = fr_flags; is->is_ifp[out << 1] = fin->fin_ifp; strncpy(is->is_ifname[out << 1], IFNAME(fin->fin_ifp), IFNAMSIZ); WRITE_ENTER(&ipf_state); is->is_pass = pass; if ((flags & FI_IGNOREPKT) == 0) { is->is_pkts = 1; is->is_bytes = fin->fin_dlen + fin->fin_hlen; } /* * We want to check everything that is a property of this packet, * but we don't (automatically) care about it's fragment status as * this may change. */ is->is_v = fin->fin_v; is->is_rulen = fin->fin_rule; is->is_opt = fin->fin_fi.fi_optmsk; is->is_optmsk = 0xffffffff; is->is_sec = fin->fin_fi.fi_secmsk; is->is_secmsk = 0xffff; is->is_auth = fin->fin_fi.fi_auth; is->is_authmsk = 0xffff; is->is_flags = fin->fin_fl & FI_CMP; is->is_flags |= FI_CMP << 4; is->is_flags |= flags & (FI_WILDP|FI_WILDA); if (flags & (FI_WILDP|FI_WILDA)) ips_wild++; if (pass & FR_LOGFIRST) is->is_pass &= ~(FR_LOGFIRST|FR_LOG); fr_stinsert(is); is->is_me = stsave; if (is->is_p == IPPROTO_TCP) { fr_tcp_age(&is->is_age, is->is_state, fin, 0, is->is_fsm); /* 0 = packet from the source */ } #ifdef IPFILTER_LOG ipstate_log(is, ISL_NEW); #endif RWLOCK_EXIT(&ipf_state); fin->fin_rev = IP6NEQ(is->is_dst, fin->fin_fi.fi_dst); if ((fin->fin_fl & FI_FRAG) && (pass & FR_KEEPFRAG)) ipfr_newfrag(ip, fin, pass ^ FR_KEEPSTATE); return is; } static int fr_tcpoptions(tcp) tcphdr_t *tcp; { u_char *opt, *last; int wscale; opt = (u_char *) (tcp + 1); last = ((u_char *)tcp) + (tcp->th_off << 2); /* If we don't find wscale here, we need to clear it */ wscale = -2; /* Termination condition picked such that opt[0 .. 2] exist */ while ((opt < last - 2) && (*opt != TCPOPT_EOL)) { switch (*opt) { case TCPOPT_NOP: opt++; continue; case TCPOPT_WSCALE: /* Proper length ? */ if (opt[1] == 3) { if (opt[2] > 14) wscale = 14; else wscale = opt[2]; } break; default: /* Unknown options must be two bytes+ */ if (opt[1] < 2) break; opt += opt[1]; continue; } break; } return wscale; } /* * check to see if a packet with TCP headers fits within the TCP window. * change timeout depending on whether new packet is a SYN-ACK returning for a * SYN or a RST or FIN which indicate time to close up shop. */ int fr_tcpstate(is, fin, ip, tcp) register ipstate_t *is; fr_info_t *fin; ip_t *ip; tcphdr_t *tcp; { register tcp_seq seq, ack, end; register int ackskew; tcpdata_t *fdata, *tdata; u_32_t win, maxwin; int ret = 0, off; int source; int wscale; /* * Find difference between last checked packet and this packet. */ source = IP6EQ(fin->fin_fi.fi_src, is->is_src); if (source && (ntohs(is->is_sport) != fin->fin_data[0])) source = 0; fdata = &is->is_tcp.ts_data[!source]; tdata = &is->is_tcp.ts_data[source]; off = tcp->th_off << 2; seq = ntohl(tcp->th_seq); ack = ntohl(tcp->th_ack); win = ntohs(tcp->th_win); end = seq + fin->fin_dlen - off + ((tcp->th_flags & TH_SYN) ? 1 : 0) + ((tcp->th_flags & TH_FIN) ? 1 : 0); if ((tcp->th_flags & TH_SYN) && (off >= sizeof(*tcp) + 4)) wscale = fr_tcpoptions(tcp); else wscale = -1; MUTEX_ENTER(&is->is_lock); if (wscale >= 0) fdata->td_wscale = wscale; else if (wscale == -2) fdata->td_wscale = tdata->td_wscale = 0; if ((fdata->td_end == 0) && (!is->is_fsm || ((tcp->th_flags & TH_OPENING) == TH_OPENING))) { /* * Must be a (outgoing) SYN-ACK in reply to a SYN. */ fdata->td_end = end; fdata->td_maxwin = 1; fdata->td_maxend = end + 1; } if (!(tcp->th_flags & TH_ACK)) { /* Pretend an ack was sent */ ack = tdata->td_end; } else if (((tcp->th_flags & (TH_ACK|TH_RST)) == (TH_ACK|TH_RST)) && (ack == 0)) { /* gross hack to get around certain broken tcp stacks */ ack = tdata->td_end; } if (seq == end) seq = end = fdata->td_end; win <<= fdata->td_wscale; maxwin = tdata->td_maxwin; ackskew = tdata->td_end - ack; #define SEQ_GE(a,b) ((int)((a) - (b)) >= 0) #define SEQ_GT(a,b) ((int)((a) - (b)) > 0) if ((SEQ_GE(fdata->td_maxend, end)) && (SEQ_GE(seq, fdata->td_end - maxwin)) && /* XXX what about big packets */ #define MAXACKWINDOW 66000 (ackskew >= -MAXACKWINDOW) && (ackskew <= MAXACKWINDOW)) { /* if ackskew < 0 then this should be due to fragented * packets. There is no way to know the length of the * total packet in advance. * We do know the total length from the fragment cache though. * Note however that there might be more sessions with * exactly the same source and destination paramters in the * state cache (and source and destination is the only stuff * that is saved in the fragment cache). Note further that * some TCP connections in the state cache are hashed with * sport and dport as well which makes it not worthwhile to * look for them. * Thus, when ackskew is negative but still seems to belong * to this session, we bump up the destinations end value. */ /* * Nearing end of connection, start timeout. */ /* source ? 0 : 1 -> !source */ if (fr_tcp_age(&is->is_age, is->is_state, fin, !source, (int)is->is_fsm) == 0) { if (ackskew < 0) tdata->td_end = ack; /* update max window seen */ if (fdata->td_maxwin < win) fdata->td_maxwin = win; if (SEQ_GT(end, fdata->td_end)) fdata->td_end = end; if (SEQ_GE(ack + win, tdata->td_maxend)) { tdata->td_maxend = ack + win; if (win == 0) tdata->td_maxend++; } ATOMIC_INCL(ips_stats.iss_hits); ret = 1; } } MUTEX_EXIT(&is->is_lock); if ((ret == 0) && ((tcp->th_flags & TH_OPENING) != TH_SYN)) fin->fin_misc |= FM_BADSTATE; return ret; } /* * Match a state table entry against an IP packet. */ static int fr_matchsrcdst(is, src, dst, fin, tcp) ipstate_t *is; union i6addr src, dst; fr_info_t *fin; tcphdr_t *tcp; { int ret = 0, rev, out, flags, idx; u_short sp, dp; void *ifp; rev = IP6NEQ(is->is_dst, dst); ifp = fin->fin_ifp; out = fin->fin_out; flags = is->is_flags & (FI_WILDA|FI_WILDP); sp = 0; dp = 0; if (tcp != NULL) { flags = is->is_flags; sp = tcp->th_sport; dp = tcp->th_dport; if (!rev) { if (!(flags & FI_W_SPORT) && (sp != is->is_sport)) rev = 1; else if (!(flags & FI_W_DPORT) && (dp != is->is_dport)) rev = 1; } } idx = (out << 1) + rev; if ((is->is_ifp[idx] == NULL && (*is->is_ifname[idx] == '\0' || *is->is_ifname[idx] == '*')) || is->is_ifp[idx] == ifp) ret = 1; if (ret == 0) return 0; ret = 0; if (rev == 0) { if ((IP6EQ(is->is_dst, dst) || (flags & FI_W_DADDR)) && (IP6EQ(is->is_src, src) || (flags & FI_W_SADDR)) && (!tcp || ((sp == is->is_sport || flags & FI_W_SPORT) && (dp == is->is_dport || flags & FI_W_DPORT)))) { ret = 1; } } else { if ((IP6EQ(is->is_dst, src) || (flags & FI_W_DADDR)) && (IP6EQ(is->is_src, dst) || (flags & FI_W_SADDR)) && (!tcp || ((sp == is->is_dport || flags & FI_W_DPORT) && (dp == is->is_sport || flags & FI_W_SPORT)))) { ret = 1; } } if (ret == 0) return 0; /* * Whether or not this should be here, is questionable, but the aim * is to get this out of the main line. */ if (tcp == NULL) flags = is->is_flags & (FI_CMP|(FI_CMP<<4)); if (((fin->fin_fl & (flags >> 4)) != (flags & FI_CMP)) || (fin->fin_fi.fi_optmsk != is->is_opt) || (fin->fin_fi.fi_secmsk != is->is_sec) || (fin->fin_fi.fi_auth != is->is_auth)) return 0; flags = is->is_flags & (FI_WILDA|FI_WILDP); if ((flags & (FI_W_SADDR|FI_W_DADDR))) { if ((flags & FI_W_SADDR) != 0) { if (rev == 0) { is->is_src = fin->fin_fi.fi_src; } else { is->is_src = fin->fin_fi.fi_dst; } } else if ((flags & FI_W_DPORT) != 0) { if (rev == 0) { is->is_dst = fin->fin_fi.fi_dst; } else { is->is_dst = fin->fin_fi.fi_src; } } is->is_flags &= ~(FI_W_SADDR|FI_W_DADDR); if ((is->is_flags & (FI_WILDA|FI_WILDP)) == 0) ips_wild--; } if ((flags & (FI_W_SPORT|FI_W_DPORT))) { if ((flags & FI_W_SPORT) != 0) { if (rev == 0) { is->is_sport = sp; is->is_send = htonl(tcp->th_seq); } else { is->is_sport = dp; is->is_send = htonl(tcp->th_ack); } is->is_maxsend = is->is_send + 1; } else if ((flags & FI_W_DPORT) != 0) { if (rev == 0) { is->is_dport = dp; is->is_dend = htonl(tcp->th_ack); } else { is->is_dport = sp; is->is_dend = htonl(tcp->th_seq); } is->is_maxdend = is->is_dend + 1; } is->is_flags &= ~(FI_W_SPORT|FI_W_DPORT); ips_wild--; } ret = -1; if (is->is_ifp[idx] == NULL && (*is->is_ifname[idx] == '\0' || *is->is_ifname[idx] == '*')) ret = idx; if (ret >= 0) { is->is_ifp[ret] = ifp; strncpy(is->is_ifname[ret], IFNAME(ifp), sizeof(is->is_ifname[ret])); } fin->fin_rev = rev; return 1; } static int fr_matchicmpqueryreply(v, is, icmp) int v; ipstate_t *is; icmphdr_t *icmp; { if (v == 4) { /* * If we matched its type on the way in, then when going out * it will still be the same type. */ if (((icmp->icmp_type == is->is_type) || (icmpreplytype4[is->is_type] == icmp->icmp_type))) { if (icmp->icmp_type != ICMP_ECHOREPLY) return 1; if ((icmp->icmp_id == is->is_icmp.ics_id) && (icmp->icmp_seq == is->is_icmp.ics_seq)) return 1; } } #ifdef USE_INET6 else if (is->is_v == 6) { if (((icmp->icmp_type == is->is_type) || (icmpreplytype6[is->is_type] == icmp->icmp_type))) { if (icmp->icmp_type != ICMP6_ECHO_REPLY) return 1; if ((icmp->icmp_id == is->is_icmp.ics_id) && (icmp->icmp_seq == is->is_icmp.ics_seq)) return 1; } } #endif return 0; } static frentry_t *fr_checkicmpmatchingstate(ip, fin) ip_t *ip; fr_info_t *fin; { register ipstate_t *is, **isp; register u_short sport, dport; register u_char pr; u_short savelen, ohlen; union i6addr dst, src; struct icmp *ic; icmphdr_t *icmp; fr_info_t ofin; int type, len; tcphdr_t *tcp; frentry_t *fr; ip_t *oip; u_int hv; /* * Does it at least have the return (basic) IP header ? * Only a basic IP header (no options) should be with * an ICMP error header. */ if (((ip->ip_v != 4) || (ip->ip_hl != 5)) || (fin->fin_plen < ICMPERR_MINPKTLEN)) return NULL; ic = (struct icmp *)fin->fin_dp; type = ic->icmp_type; /* * If it's not an error type, then return */ if ((type != ICMP_UNREACH) && (type != ICMP_SOURCEQUENCH) && (type != ICMP_REDIRECT) && (type != ICMP_TIMXCEED) && (type != ICMP_PARAMPROB)) return NULL; oip = (ip_t *)((char *)ic + ICMPERR_ICMPHLEN); ohlen = oip->ip_hl << 2; if (fin->fin_plen < ICMPERR_MAXPKTLEN + ohlen - sizeof(*oip)) return NULL; /* * Sanity checks. */ len = fin->fin_dlen - ICMPERR_ICMPHLEN; if ((len <= 0) || (ohlen > len)) return NULL; /* * Is the buffer big enough for all of it ? It's the size of the IP * header claimed in the encapsulated part which is of concern. It * may be too big to be in this buffer but not so big that it's * outside the ICMP packet, leading to TCP deref's causing problems. * This is possible because we don't know how big oip_hl is when we * do the pullup early in fr_check() and thus can't gaurantee it is * all here now. */ #ifdef _KERNEL { mb_t *m; # if SOLARIS m = fin->fin_qfm; if ((char *)oip + len > (char *)m->b_wptr) return NULL; # else m = *(mb_t **)fin->fin_mp; if ((char *)oip + len > (char *)ip + m->m_len) return NULL; # endif } #endif /* * in the IPv4 case we must zero the i6addr union otherwise * the IP6EQ and IP6NEQ macros produce the wrong results because * of the 'junk' in the unused part of the union */ bzero((char *)&src, sizeof(src)); bzero((char *)&dst, sizeof(dst)); bzero((char *)&ofin, sizeof(ofin)); ofin.fin_ifp = fin->fin_ifp; ofin.fin_out = !fin->fin_out; ofin.fin_v = 4; fr = NULL; switch (oip->ip_p) { case IPPROTO_ICMP : icmp = (icmphdr_t *)((char *)oip + ohlen); /* * a ICMP error can only be generated as a result of an * ICMP query, not as the response on an ICMP error * * XXX theoretically ICMP_ECHOREP and the other reply's are * ICMP query's as well, but adding them here seems strange XXX */ if ((icmp->icmp_type != ICMP_ECHO) && (icmp->icmp_type != ICMP_TSTAMP) && (icmp->icmp_type != ICMP_IREQ) && (icmp->icmp_type != ICMP_MASKREQ)) return NULL; /* * perform a lookup of the ICMP packet in the state table */ hv = (pr = oip->ip_p); src.in4 = oip->ip_src; hv += src.in4.s_addr; dst.in4 = oip->ip_dst; hv += dst.in4.s_addr; hv += icmp->icmp_id; hv += icmp->icmp_seq; hv %= fr_statesize; savelen = oip->ip_len; oip->ip_len = len; fr_makefrip(ohlen, oip, &ofin); oip->ip_len = savelen; READ_ENTER(&ipf_state); for (isp = &ips_table[hv]; (is = *isp); isp = &is->is_hnext) if ((is->is_p == pr) && (is->is_v == 4) && fr_matchsrcdst(is, src, dst, &ofin, NULL) && fr_matchicmpqueryreply(is->is_v, is, icmp)) { ips_stats.iss_hits++; is->is_pkts++; is->is_bytes += ip->ip_len; fr = is->is_rule; break; } RWLOCK_EXIT(&ipf_state); return fr; case IPPROTO_TCP : case IPPROTO_UDP : if (fin->fin_plen < ICMPERR_MAXPKTLEN) return NULL; break; default : return NULL; } tcp = (tcphdr_t *)((char *)oip + ohlen); dport = tcp->th_dport; sport = tcp->th_sport; hv = (pr = oip->ip_p); src.in4 = oip->ip_src; hv += src.in4.s_addr; dst.in4 = oip->ip_dst; hv += dst.in4.s_addr; hv += dport; hv += sport; hv %= fr_statesize; /* * we make an fin entry to be able to feed it to * matchsrcdst note that not all fields are encessary * but this is the cleanest way. Note further we fill * in fin_mp such that if someone uses it we'll get * a kernel panic. fr_matchsrcdst does not use this. * * watch out here, as ip is in host order and oip in network * order. Any change we make must be undone afterwards. */ savelen = oip->ip_len; oip->ip_len = len; fr_makefrip(ohlen, oip, &ofin); oip->ip_len = savelen; READ_ENTER(&ipf_state); for (isp = &ips_table[hv]; (is = *isp); isp = &is->is_hnext) { /* * Only allow this icmp though if the * encapsulated packet was allowed through the * other way around. Note that the minimal amount * of info present does not allow for checking against * tcp internals such as seq and ack numbers. */ if ((is->is_p == pr) && (is->is_v == 4) && fr_matchsrcdst(is, src, dst, &ofin, tcp)) { fr = is->is_rule; ips_stats.iss_hits++; is->is_pkts++; is->is_bytes += fin->fin_plen; /* * we deliberately do not touch the timeouts * for the accompanying state table entry. * It remains to be seen if that is correct. XXX */ break; } } RWLOCK_EXIT(&ipf_state); return fr; } /* * Move a state hash table entry from its old location at is->is_hv to * its new location, indexed by hv % fr_statesize. */ static void fr_ipsmove(isp, is, hv) ipstate_t **isp, *is; u_int hv; { u_int hvm; hvm = is->is_hv; /* * Remove the hash from the old location... */ if (is->is_hnext) is->is_hnext->is_phnext = isp; *isp = is->is_hnext; if (ips_table[hvm] == NULL) ips_stats.iss_inuse--; /* * ...and put the hash in the new one. */ hvm = hv % fr_statesize; is->is_hv = hvm; isp = &ips_table[hvm]; if (*isp) (*isp)->is_phnext = &is->is_hnext; else ips_stats.iss_inuse++; is->is_phnext = isp; is->is_hnext = *isp; *isp = is; } /* * Check if a packet has a registered state. */ frentry_t *fr_checkstate(ip, fin) ip_t *ip; fr_info_t *fin; { union i6addr dst, src; register ipstate_t *is, **isp; register u_char pr; u_int hv, hvm, hlen, tryagain, pass, v; struct icmp *ic; frentry_t *fr; tcphdr_t *tcp; int rev; if (fr_state_lock || (fin->fin_off != 0) || (fin->fin_fl & FI_SHORT)) return NULL; is = NULL; hlen = fin->fin_hlen; tcp = (tcphdr_t *)((char *)ip + hlen); ic = (struct icmp *)tcp; hv = (pr = fin->fin_fi.fi_p); src = fin->fin_fi.fi_src; dst = fin->fin_fi.fi_dst; hv += src.in4.s_addr; hv += dst.in4.s_addr; /* * Search the hash table for matching packet header info. * At the bottom of this switch statement, the following is expected: * is == NULL, no lock on ipf_state is held. * is != NULL, a lock on ipf_state is held. */ v = fin->fin_fi.fi_v; #ifdef USE_INET6 if (v == 6) { hv += fin->fin_fi.fi_src.i6[1]; hv += fin->fin_fi.fi_src.i6[2]; hv += fin->fin_fi.fi_src.i6[3]; if ((fin->fin_p == IPPROTO_ICMPV6) && IN6_IS_ADDR_MULTICAST(&fin->fin_fi.fi_dst.in6)) { hv -= dst.in4.s_addr; } else { hv += fin->fin_fi.fi_dst.i6[1]; hv += fin->fin_fi.fi_dst.i6[2]; hv += fin->fin_fi.fi_dst.i6[3]; } } #endif switch (fin->fin_p) { #ifdef USE_INET6 case IPPROTO_ICMPV6 : tcp = NULL; tryagain = 0; if (v == 6) { if ((ic->icmp_type == ICMP6_ECHO_REQUEST) || (ic->icmp_type == ICMP6_ECHO_REPLY)) { hv += ic->icmp_id; hv += ic->icmp_seq; } } READ_ENTER(&ipf_state); icmp6again: hvm = hv % fr_statesize; for (isp = &ips_table[hvm]; (is = *isp); isp = &is->is_hnext) if ((is->is_p == pr) && (is->is_v == v) && fr_matchsrcdst(is, src, dst, fin, NULL) && fr_matchicmpqueryreply(v, is, ic)) { rev = fin->fin_rev; if (is->is_frage[rev] != 0) is->is_age = is->is_frage[rev]; else if (fin->fin_rev) is->is_age = fr_icmpacktimeout; else is->is_age = fr_icmptimeout; break; } if (is != NULL) { if (tryagain && !(is->is_flags & FI_W_DADDR)) { hv += fin->fin_fi.fi_src.i6[0]; hv += fin->fin_fi.fi_src.i6[1]; hv += fin->fin_fi.fi_src.i6[2]; hv += fin->fin_fi.fi_src.i6[3]; fr_ipsmove(isp, is, hv); MUTEX_DOWNGRADE(&ipf_state); } break; } RWLOCK_EXIT(&ipf_state); /* * No matching icmp state entry. Perhaps this is a * response to another state entry. */ if ((ips_wild != 0) && (v == 6) && (tryagain == 0) && !IN6_IS_ADDR_MULTICAST(&fin->fin_fi.fi_src.in6)) { hv -= fin->fin_fi.fi_src.i6[0]; hv -= fin->fin_fi.fi_src.i6[1]; hv -= fin->fin_fi.fi_src.i6[2]; hv -= fin->fin_fi.fi_src.i6[3]; tryagain = 1; WRITE_ENTER(&ipf_state); goto icmp6again; } fr = fr_checkicmp6matchingstate((ip6_t *)ip, fin); if (fr) return fr; break; #endif case IPPROTO_ICMP : tcp = NULL; if (v == 4) { hv += ic->icmp_id; hv += ic->icmp_seq; } hvm = hv % fr_statesize; READ_ENTER(&ipf_state); for (isp = &ips_table[hvm]; (is = *isp); isp = &is->is_hnext) if ((is->is_p == pr) && (is->is_v == v) && fr_matchsrcdst(is, src, dst, fin, NULL) && fr_matchicmpqueryreply(v, is, ic)) { rev = fin->fin_rev; if (is->is_frage[rev] != 0) is->is_age = is->is_frage[rev]; else if (fin->fin_rev) is->is_age = fr_icmpacktimeout; else is->is_age = fr_icmptimeout; break; } if (is != NULL) break; RWLOCK_EXIT(&ipf_state); /* * No matching icmp state entry. Perhaps this is a * response to another state entry. */ fr = fr_checkicmpmatchingstate(ip, fin); if (fr) return fr; break; case IPPROTO_TCP : /* * Just plain ignore RST flag set with either FIN or SYN. */ if ((tcp->th_flags & TH_RST) && ((tcp->th_flags & (TH_FIN|TH_SYN|TH_RST)) != TH_RST)) break; case IPPROTO_UDP : { register u_short dport, sport; dport = tcp->th_dport; sport = tcp->th_sport; tryagain = 0; hv += dport; hv += sport; READ_ENTER(&ipf_state); retry_tcpudp: hvm = hv % fr_statesize; for (isp = &ips_table[hvm]; (is = *isp); isp = &is->is_hnext) if ((is->is_p == pr) && (is->is_v == v) && fr_matchsrcdst(is, src, dst, fin, tcp)) { rev = fin->fin_rev; if ((pr == IPPROTO_TCP)) { if (!fr_tcpstate(is, fin, ip, tcp)) is = NULL; } else if ((pr == IPPROTO_UDP)) { if (is->is_frage[rev] != 0) is->is_age = is->is_frage[rev]; else if (fin->fin_rev) is->is_age = fr_udpacktimeout; else is->is_age = fr_udptimeout; } break; } if (is != NULL) { if (tryagain && !(is->is_flags & (FI_WILDP|FI_WILDA))) { hv += dport; hv += sport; fr_ipsmove(isp, is, hv); MUTEX_DOWNGRADE(&ipf_state); } break; } RWLOCK_EXIT(&ipf_state); if (!tryagain && ips_wild) { hv -= dport; hv -= sport; tryagain = 1; WRITE_ENTER(&ipf_state); goto retry_tcpudp; } break; } default : tcp = NULL; hv %= fr_statesize; READ_ENTER(&ipf_state); for (isp = &ips_table[hv]; (is = *isp); isp = &is->is_hnext) { if ((is->is_p == pr) && (is->is_v == v) && fr_matchsrcdst(is, src, dst, fin, NULL)) { rev = fin->fin_rev; if (is->is_frage[rev] != 0) is->is_age = is->is_frage[rev]; else is->is_age = fr_udptimeout; break; } } if (is == NULL) { RWLOCK_EXIT(&ipf_state); } break; } if (is == NULL) { ATOMIC_INCL(ips_stats.iss_miss); return NULL; } MUTEX_ENTER(&is->is_lock); is->is_bytes += fin->fin_plen; ips_stats.iss_hits++; is->is_pkts++; MUTEX_EXIT(&is->is_lock); fr = is->is_rule; fin->fin_rule = is->is_rulen; if (fr != NULL) { fin->fin_group = fr->fr_group; fin->fin_icode = fr->fr_icode; } fin->fin_fr = fr; pass = is->is_pass; RWLOCK_EXIT(&ipf_state); if ((fin->fin_fl & FI_FRAG) && (pass & FR_KEEPFRAG)) ipfr_newfrag(ip, fin, pass ^ FR_KEEPSTATE); #ifndef _KERNEL if ((tcp != NULL) && (tcp->th_flags & TCP_CLOSE)) fr_delstate(is); #endif return fr; } /* * Sync. state entries. If interfaces come or go or just change position, * this is needed. */ void ip_statesync(ifp) void *ifp; { register ipstate_t *is; int i; WRITE_ENTER(&ipf_state); for (is = ips_list; is; is = is->is_next) { for (i = 0; i < 4; i++) { if (is->is_ifp[i] == ifp) { is->is_ifpin = GETUNIT(is->is_ifname[i], is->is_v); if (!is->is_ifp[i]) is->is_ifp[i] = (void *)-1; } } } RWLOCK_EXIT(&ipf_state); } /* * Must always be called with fr_ipfstate held as a write lock. */ static void fr_delstate(is) ipstate_t *is; { frentry_t *fr; if (is->is_flags & (FI_WILDP|FI_WILDA)) ips_wild--; if (is->is_next) is->is_next->is_pnext = is->is_pnext; *is->is_pnext = is->is_next; if (is->is_hnext) is->is_hnext->is_phnext = is->is_phnext; *is->is_phnext = is->is_hnext; if (ips_table[is->is_hv] == NULL) ips_stats.iss_inuse--; if (is->is_me) *is->is_me = NULL; fr = is->is_rule; if (fr != NULL) { fr->fr_ref--; if (fr->fr_ref == 0) { KFREE(fr); } } #ifdef _KERNEL MUTEX_DESTROY(&is->is_lock); #endif KFREE(is); ips_num--; } /* * Free memory in use by all state info. kept. */ void fr_stateunload() { register ipstate_t *is; WRITE_ENTER(&ipf_state); while ((is = ips_list)) fr_delstate(is); ips_stats.iss_inuse = 0; ips_num = 0; RWLOCK_EXIT(&ipf_state); if (ips_table) KFREES(ips_table, fr_statesize * sizeof(ipstate_t *)); ips_table = NULL; } /* * Slowly expire held state for thingslike UDP and ICMP. Timeouts are set * in expectation of this being called twice per second. */ void fr_timeoutstate() { register ipstate_t *is, **isp; #if defined(_KERNEL) && !SOLARIS int s; #endif SPL_NET(s); WRITE_ENTER(&ipf_state); for (isp = &ips_list; (is = *isp); ) if (is->is_age && !--is->is_age) { if (is->is_p == IPPROTO_TCP) ips_stats.iss_fin++; else ips_stats.iss_expire++; #ifdef IPFILTER_LOG ipstate_log(is, ISL_EXPIRE); #endif fr_delstate(is); } else isp = &is->is_next; if (fr_state_doflush) { (void) fr_state_flush(2); fr_state_doflush = 0; } RWLOCK_EXIT(&ipf_state); SPL_X(s); } /* * Original idea freom Pradeep Krishnan for use primarily with NAT code. * (pkrishna@netcom.com) * * Rewritten by Arjan de Vet , 2000-07-29: * * - (try to) base state transitions on real evidence only, * i.e. packets that are sent and have been received by ipfilter; * diagram 18.12 of TCP/IP volume 1 by W. Richard Stevens was used. * * - deal with half-closed connections correctly; * * - store the state of the source in state[0] such that ipfstat * displays the state as source/dest instead of dest/source; the calls * to fr_tcp_age have been changed accordingly. * * Parameters: * * state[0] = state of source (host that initiated connection) * state[1] = state of dest (host that accepted the connection) * * dir == 0 : a packet from source to dest * dir == 1 : a packet from dest to source * */ int fr_tcp_age(age, state, fin, dir, fsm) u_long *age; u_char *state; fr_info_t *fin; int dir, fsm; { tcphdr_t *tcp = (tcphdr_t *)fin->fin_dp; u_char flags = tcp->th_flags; int dlen, ostate; u_long newage; ostate = state[1 - dir]; dlen = fin->fin_plen - fin->fin_hlen - (tcp->th_off << 2); if (flags & TH_RST) { if (!(tcp->th_flags & TH_PUSH) && !dlen) { *age = fr_tcpclosed; state[dir] = TCPS_CLOSED; } else { *age = fr_tcpclosewait; state[dir] = TCPS_CLOSE_WAIT; } return 0; } newage = 0; switch(state[dir]) { case TCPS_CLOSED: /* 0 */ if ((flags & TH_OPENING) == TH_OPENING) { /* * 'dir' received an S and sends SA in response, * CLOSED -> SYN_RECEIVED */ state[dir] = TCPS_SYN_RECEIVED; newage = fr_tcptimeout; } else if ((flags & TH_OPENING) == TH_SYN) { /* 'dir' sent S, CLOSED -> SYN_SENT */ state[dir] = TCPS_SYN_SENT; newage = fr_tcptimeout; } /* * The next piece of code makes it possible to get * already established connections into the state table * after a restart or reload of the filter rules; this * does not work when a strict 'flags S keep state' is * used for tcp connections of course */ if (!fsm && (flags & (TH_FIN|TH_SYN|TH_RST|TH_ACK)) == TH_ACK) { /* we saw an A, guess 'dir' is in ESTABLISHED mode */ if (state[1 - dir] == TCPS_CLOSED || state[1 - dir] == TCPS_ESTABLISHED) { state[dir] = TCPS_ESTABLISHED; newage = fr_tcpidletimeout; } } /* * TODO: besides regular ACK packets we can have other * packets as well; it is yet to be determined how we * should initialize the states in those cases */ break; case TCPS_LISTEN: /* 1 */ /* NOT USED */ break; case TCPS_SYN_SENT: /* 2 */ if ((flags & ~(TH_ECN|TH_CWR)) == TH_SYN) { /* * A retransmitted SYN packet. We do not reset the * timeout here to fr_tcptimeout because a connection * connect timeout does not renew after every packet * that is sent. We need to set newage to something * to indicate the packet has passed the check for its * flags being valid in the TCP FSM. */ newage = *age; } else if ((flags & (TH_SYN|TH_FIN|TH_ACK)) == TH_ACK) { /* * We see an A from 'dir' which is in SYN_SENT * state: 'dir' sent an A in response to an SA * which it received, SYN_SENT -> ESTABLISHED */ state[dir] = TCPS_ESTABLISHED; newage = fr_tcpidletimeout; } else if (flags & TH_FIN) { /* * We see an F from 'dir' which is in SYN_SENT * state and wants to close its side of the * connection; SYN_SENT -> FIN_WAIT_1 */ state[dir] = TCPS_FIN_WAIT_1; newage = fr_tcpidletimeout; /* or fr_tcptimeout? */ } else if ((flags & TH_OPENING) == TH_OPENING) { /* * We see an SA from 'dir' which is already in * SYN_SENT state, this means we have a * simultaneous open; SYN_SENT -> SYN_RECEIVED */ state[dir] = TCPS_SYN_RECEIVED; newage = fr_tcptimeout; } break; case TCPS_SYN_RECEIVED: /* 3 */ if ((flags & (TH_SYN|TH_FIN|TH_ACK)) == TH_ACK) { /* * We see an A from 'dir' which was in SYN_RECEIVED * state so it must now be in established state, * SYN_RECEIVED -> ESTABLISHED */ state[dir] = TCPS_ESTABLISHED; newage = fr_tcpidletimeout; } else if ((flags & ~(TH_ECN|TH_CWR)) == TH_OPENING) { /* * We see an SA from 'dir' which is already in * SYN_RECEIVED state. */ newage = fr_tcptimeout; } else if (flags & TH_FIN) { /* * We see an F from 'dir' which is in SYN_RECEIVED * state and wants to close its side of the connection; * SYN_RECEIVED -> FIN_WAIT_1 */ state[dir] = TCPS_FIN_WAIT_1; newage = fr_tcpidletimeout; } break; case TCPS_ESTABLISHED: /* 4 */ if (flags & TH_FIN) { /* * 'dir' closed its side of the connection; this * gives us a half-closed connection; * ESTABLISHED -> FIN_WAIT_1 */ state[dir] = TCPS_FIN_WAIT_1; newage = fr_tcphalfclosed; } else if (flags & TH_ACK) { /* an ACK, should we exclude other flags here? */ if (ostate == TCPS_FIN_WAIT_1) { /* * We know the other side did an active close, * so we are ACKing the recvd FIN packet (does * the window matching code guarantee this?) * and go into CLOSE_WAIT state; this gives us * a half-closed connection */ state[dir] = TCPS_CLOSE_WAIT; newage = fr_tcphalfclosed; } else if (ostate < TCPS_CLOSE_WAIT) /* * Still a fully established connection, * reset timeout */ newage = fr_tcpidletimeout; } break; case TCPS_CLOSE_WAIT: /* 5 */ if (flags & TH_FIN) { /* * Application closed and 'dir' sent a FIN, we're now * going into LAST_ACK state */ newage = fr_tcplastack; state[dir] = TCPS_LAST_ACK; } else { /* * We remain in CLOSE_WAIT because the other side has * closed already and we did not close our side yet; * reset timeout */ newage = fr_tcphalfclosed; } break; case TCPS_FIN_WAIT_1: /* 6 */ if ((flags & TH_ACK) && ostate > TCPS_CLOSE_WAIT) { /* * If the other side is not active anymore it has sent * us a FIN packet that we are ack'ing now with an ACK; * this means both sides have now closed the connection * and we go into TIME_WAIT */ /* * XXX: how do we know we really are ACKing the FIN * packet here? does the window code guarantee that? */ state[dir] = TCPS_TIME_WAIT; newage = fr_tcptimeout; } else /* * We closed our side of the connection already but the * other side is still active (ESTABLISHED/CLOSE_WAIT); * continue with this half-closed connection */ newage = fr_tcphalfclosed; break; case TCPS_CLOSING: /* 7 */ /* NOT USED */ break; case TCPS_LAST_ACK: /* 8 */ if (flags & TH_ACK) { if ((flags & TH_PUSH) || dlen) /* * There is still data to be delivered, reset * timeout */ newage = fr_tcplastack; else newage = *age; } /* * We cannot detect when we go out of LAST_ACK state to CLOSED * because that is based on the reception of ACK packets; * ipfilter can only detect that a packet has been sent by a * host */ break; case TCPS_FIN_WAIT_2: /* 9 */ /* NOT USED */ break; case TCPS_TIME_WAIT: /* 10 */ newage = fr_tcptimeout; /* default 4 mins */ /* we're in 2MSL timeout now */ break; } if (newage != 0) { *age = newage; return 0; } return -1; } #ifdef IPFILTER_LOG void ipstate_log(is, type) struct ipstate *is; u_int type; { struct ipslog ipsl; void *items[1]; size_t sizes[1]; int types[1]; ipsl.isl_type = type; ipsl.isl_pkts = is->is_pkts; ipsl.isl_bytes = is->is_bytes; ipsl.isl_src = is->is_src; ipsl.isl_dst = is->is_dst; ipsl.isl_p = is->is_p; ipsl.isl_v = is->is_v; ipsl.isl_flags = is->is_flags; if (ipsl.isl_p == IPPROTO_TCP || ipsl.isl_p == IPPROTO_UDP) { ipsl.isl_sport = is->is_sport; ipsl.isl_dport = is->is_dport; if (ipsl.isl_p == IPPROTO_TCP) { ipsl.isl_state[0] = is->is_state[0]; ipsl.isl_state[1] = is->is_state[1]; } } else if (ipsl.isl_p == IPPROTO_ICMP) { ipsl.isl_itype = is->is_icmp.ics_type; } else if (ipsl.isl_p == IPPROTO_ICMPV6) { ipsl.isl_itype = is->is_icmp.ics_type; } else { ipsl.isl_ps.isl_filler[0] = 0; ipsl.isl_ps.isl_filler[1] = 0; } items[0] = &ipsl; sizes[0] = sizeof(ipsl); types[0] = 0; (void) ipllog(IPL_LOGSTATE, NULL, items, sizes, types, 1); } #endif #ifdef USE_INET6 frentry_t *fr_checkicmp6matchingstate(ip, fin) ip6_t *ip; fr_info_t *fin; { register ipstate_t *is, **isp; register u_short sport, dport; register u_char pr; struct icmp6_hdr *ic, *oic; union i6addr dst, src; u_short savelen; fr_info_t ofin; tcphdr_t *tcp; frentry_t *fr; ip6_t *oip; int type; u_int hv; /* * Does it at least have the return (basic) IP header ? * Only a basic IP header (no options) should be with * an ICMP error header. */ if ((fin->fin_v != 6) || (fin->fin_plen < ICMP6ERR_MINPKTLEN)) return NULL; ic = (struct icmp6_hdr *)fin->fin_dp; type = ic->icmp6_type; /* * If it's not an error type, then return */ if ((type != ICMP6_DST_UNREACH) && (type != ICMP6_PACKET_TOO_BIG) && (type != ICMP6_TIME_EXCEEDED) && (type != ICMP6_PARAM_PROB)) return NULL; oip = (ip6_t *)((char *)ic + ICMPERR_ICMPHLEN); if (fin->fin_plen < sizeof(*oip)) return NULL; if ((oip->ip6_nxt != IPPROTO_TCP) && (oip->ip6_nxt != IPPROTO_UDP) && (oip->ip6_nxt != IPPROTO_ICMPV6)) return NULL; bzero((char *)&ofin, sizeof(ofin)); ofin.fin_out = !fin->fin_out; ofin.fin_ifp = fin->fin_ifp; ofin.fin_v = 6; if (oip->ip6_nxt == IPPROTO_ICMPV6) { oic = (struct icmp6_hdr *)(oip + 1); /* * a ICMP error can only be generated as a result of an * ICMP query, not as the response on an ICMP error * * XXX theoretically ICMP_ECHOREP and the other reply's are * ICMP query's as well, but adding them here seems strange XXX */ if (!(oic->icmp6_type & ICMP6_INFOMSG_MASK)) return NULL; /* * perform a lookup of the ICMP packet in the state table */ hv = (pr = oip->ip6_nxt); src.in6 = oip->ip6_src; hv += src.in4.s_addr; dst.in6 = oip->ip6_dst; hv += dst.in4.s_addr; hv += oic->icmp6_id; hv += oic->icmp6_seq; hv %= fr_statesize; oip->ip6_plen = ntohs(oip->ip6_plen); fr_makefrip(sizeof(*oip), (ip_t *)oip, &ofin); oip->ip6_plen = htons(oip->ip6_plen); READ_ENTER(&ipf_state); for (isp = &ips_table[hv]; (is = *isp); isp = &is->is_hnext) if ((is->is_p == pr) && (oic->icmp6_id == is->is_icmp.ics_id) && (oic->icmp6_seq == is->is_icmp.ics_seq) && fr_matchsrcdst(is, src, dst, &ofin, NULL)) { /* * in the state table ICMP query's are stored * with the type of the corresponding ICMP * response. Correct here */ if (((is->is_type == ICMP6_ECHO_REPLY) && (oic->icmp6_type == ICMP6_ECHO_REQUEST)) || (is->is_type - 1 == oic->icmp6_type )) { ips_stats.iss_hits++; is->is_pkts++; is->is_bytes += fin->fin_plen; return is->is_rule; } } RWLOCK_EXIT(&ipf_state); return NULL; } tcp = (tcphdr_t *)(oip + 1); dport = tcp->th_dport; sport = tcp->th_sport; hv = (pr = oip->ip6_nxt); src.in6 = oip->ip6_src; hv += src.in4.s_addr; hv += src.i6[1]; hv += src.i6[2]; hv += src.i6[3]; dst.in6 = oip->ip6_dst; hv += dst.in4.s_addr; hv += dst.i6[1]; hv += dst.i6[2]; hv += dst.i6[3]; hv += dport; hv += sport; hv %= fr_statesize; /* * we make an fin entry to be able to feed it to * matchsrcdst note that not all fields are encessary * but this is the cleanest way. Note further we fill * in fin_mp such that if someone uses it we'll get * a kernel panic. fr_matchsrcdst does not use this. * * watch out here, as ip is in host order and oip in network * order. Any change we make must be undone afterwards. */ savelen = oip->ip6_plen; oip->ip6_plen = ip->ip6_plen - sizeof(*ip) - ICMPERR_ICMPHLEN; fr_makefrip(sizeof(*oip), (ip_t *)oip, &ofin); oip->ip6_plen = savelen; READ_ENTER(&ipf_state); for (isp = &ips_table[hv]; (is = *isp); isp = &is->is_hnext) { /* * Only allow this icmp though if the * encapsulated packet was allowed through the * other way around. Note that the minimal amount * of info present does not allow for checking against * tcp internals such as seq and ack numbers. */ if ((is->is_p == pr) && (is->is_v == 6) && fr_matchsrcdst(is, src, dst, &ofin, tcp)) { fr = is->is_rule; ips_stats.iss_hits++; is->is_pkts++; is->is_bytes += fin->fin_plen; /* * we deliberately do not touch the timeouts * for the accompanying state table entry. * It remains to be seen if that is correct. XXX */ RWLOCK_EXIT(&ipf_state); return fr; } } RWLOCK_EXIT(&ipf_state); return NULL; } #endif