diff options
Diffstat (limited to 'sys/contrib/ipfilter/netinet/ip_state.c')
| -rw-r--r-- | sys/contrib/ipfilter/netinet/ip_state.c | 4218 |
1 files changed, 4218 insertions, 0 deletions
diff --git a/sys/contrib/ipfilter/netinet/ip_state.c b/sys/contrib/ipfilter/netinet/ip_state.c new file mode 100644 index 0000000..6c8b158 --- /dev/null +++ b/sys/contrib/ipfilter/netinet/ip_state.c @@ -0,0 +1,4218 @@ +/* $FreeBSD$ */ + +/* + * Copyright (C) 1995-2003 by Darren Reed. + * + * See the IPFILTER.LICENCE file for details on licencing. + */ +#if defined(KERNEL) || defined(_KERNEL) +# undef KERNEL +# undef _KERNEL +# define KERNEL 1 +# define _KERNEL 1 +#endif +#include <sys/errno.h> +#include <sys/types.h> +#include <sys/param.h> +#include <sys/file.h> +#if defined(__NetBSD__) && (NetBSD >= 199905) && !defined(IPFILTER_LKM) && \ + defined(_KERNEL) +# if (__NetBSD_Version__ < 399001400) +# include "opt_ipfilter_log.h" +# else +# include "opt_ipfilter.h" +# endif +#endif +#if defined(_KERNEL) && defined(__FreeBSD_version) && \ + (__FreeBSD_version >= 400000) && !defined(KLD_MODULE) +#include "opt_inet6.h" +#endif +#if !defined(_KERNEL) && !defined(__KERNEL__) +# include <stdio.h> +# include <stdlib.h> +# include <string.h> +# define _KERNEL +# ifdef __OpenBSD__ +struct file; +# endif +# include <sys/uio.h> +# undef _KERNEL +#endif +#if defined(_KERNEL) && (__FreeBSD_version >= 220000) +# include <sys/filio.h> +# include <sys/fcntl.h> +# if (__FreeBSD_version >= 300000) && !defined(IPFILTER_LKM) +# include "opt_ipfilter.h" +# endif +#else +# include <sys/ioctl.h> +#endif +#include <sys/time.h> +#if !defined(linux) +# include <sys/protosw.h> +#endif +#include <sys/socket.h> +#if defined(_KERNEL) +# include <sys/systm.h> +# if !defined(__SVR4) && !defined(__svr4__) +# include <sys/mbuf.h> +# endif +#endif +#if defined(__SVR4) || defined(__svr4__) +# include <sys/filio.h> +# include <sys/byteorder.h> +# ifdef _KERNEL +# include <sys/dditypes.h> +# endif +# include <sys/stream.h> +# include <sys/kmem.h> +#endif + +#include <net/if.h> +#ifdef sun +# include <net/af.h> +#endif +#include <net/route.h> +#include <netinet/in.h> +#include <netinet/in_systm.h> +#include <netinet/ip.h> +#include <netinet/tcp.h> +#if !defined(linux) +# include <netinet/ip_var.h> +#endif +#if !defined(__hpux) && !defined(linux) +# include <netinet/tcp_fsm.h> +#endif +#include <netinet/udp.h> +#include <netinet/ip_icmp.h> +#include "netinet/ip_compat.h" +#include <netinet/tcpip.h> +#include "netinet/ip_fil.h" +#include "netinet/ip_nat.h" +#include "netinet/ip_frag.h" +#include "netinet/ip_state.h" +#include "netinet/ip_proxy.h" +#ifdef IPFILTER_SYNC +#include "netinet/ip_sync.h" +#endif +#ifdef IPFILTER_SCAN +#include "netinet/ip_scan.h" +#endif +#ifdef USE_INET6 +#include <netinet/icmp6.h> +#endif +#if (__FreeBSD_version >= 300000) +# include <sys/malloc.h> +# if defined(_KERNEL) && !defined(IPFILTER_LKM) +# include <sys/libkern.h> +# include <sys/systm.h> +# endif +#endif +/* END OF INCLUDES */ + + +#if !defined(lint) +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 2.186.2.80 2007/10/16 09:33:23 darrenr Exp $"; +#endif + +static ipstate_t **ips_table = NULL; +static u_long *ips_seed = NULL; +static int ips_num = 0; +static u_long ips_last_force_flush = 0; +ips_stat_t ips_stats; + +#ifdef USE_INET6 +static ipstate_t *fr_checkicmp6matchingstate __P((fr_info_t *)); +#endif +static ipstate_t *fr_matchsrcdst __P((fr_info_t *, ipstate_t *, i6addr_t *, + i6addr_t *, tcphdr_t *, u_32_t)); +static ipstate_t *fr_checkicmpmatchingstate __P((fr_info_t *)); +static int fr_state_flush __P((int, int)); +static int fr_state_flush_entry __P((void *)); +static ips_stat_t *fr_statetstats __P((void)); +static int fr_delstate __P((ipstate_t *, int)); +static int fr_state_remove __P((caddr_t)); +static void fr_ipsmove __P((ipstate_t *, u_int)); +static int fr_tcpstate __P((fr_info_t *, tcphdr_t *, ipstate_t *)); +static int fr_tcpoptions __P((fr_info_t *, tcphdr_t *, tcpdata_t *)); +static ipstate_t *fr_stclone __P((fr_info_t *, tcphdr_t *, ipstate_t *)); +static void fr_fixinisn __P((fr_info_t *, ipstate_t *)); +static void fr_fixoutisn __P((fr_info_t *, ipstate_t *)); +static void fr_checknewisn __P((fr_info_t *, ipstate_t *)); +static int fr_stateiter __P((ipftoken_t *, ipfgeniter_t *)); +static int fr_stgettable __P((char *)); + +int fr_stputent __P((caddr_t)); +int fr_stgetent __P((caddr_t)); + +#define ONE_DAY IPF_TTLVAL(1 * 86400) /* 1 day */ +#define FIVE_DAYS (5 * ONE_DAY) +#define DOUBLE_HASH(x) (((x) + ips_seed[(x) % fr_statesize]) % fr_statesize) + +u_long fr_tcpidletimeout = FIVE_DAYS, + fr_tcpclosewait = IPF_TTLVAL(2 * TCP_MSL), + fr_tcplastack = IPF_TTLVAL(30), + fr_tcptimeout = IPF_TTLVAL(2 * TCP_MSL), + fr_tcptimewait = IPF_TTLVAL(2 * TCP_MSL), + fr_tcpclosed = IPF_TTLVAL(30), + fr_tcphalfclosed = IPF_TTLVAL(2 * 3600), /* 2 hours */ + fr_udptimeout = IPF_TTLVAL(120), + fr_udpacktimeout = IPF_TTLVAL(12), + fr_icmptimeout = IPF_TTLVAL(60), + fr_icmpacktimeout = IPF_TTLVAL(6), + fr_iptimeout = IPF_TTLVAL(60); +int fr_statemax = IPSTATE_MAX, + fr_statesize = IPSTATE_SIZE; +int fr_state_doflush = 0, + fr_state_lock = 0, + fr_state_maxbucket = 0, + fr_state_maxbucket_reset = 1, + fr_state_init = 0; +ipftq_t ips_tqtqb[IPF_TCP_NSTATES], + ips_udptq, + ips_udpacktq, + ips_iptq, + ips_icmptq, + ips_icmpacktq, + ips_deletetq, + *ips_utqe = NULL; +#ifdef IPFILTER_LOG +int ipstate_logging = 1; +#else +int ipstate_logging = 0; +#endif +ipstate_t *ips_list = NULL; + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_stateinit */ +/* Returns: int - 0 == success, -1 == failure */ +/* Parameters: Nil */ +/* */ +/* Initialise all the global variables used within the state code. */ +/* This action also includes initiailising locks. */ +/* ------------------------------------------------------------------------ */ +int fr_stateinit() +{ +#if defined(NEED_LOCAL_RAND) || !defined(_KERNEL) + struct timeval tv; +#endif + int i; + + KMALLOCS(ips_table, ipstate_t **, fr_statesize * sizeof(ipstate_t *)); + if (ips_table == NULL) + return -1; + bzero((char *)ips_table, fr_statesize * sizeof(ipstate_t *)); + + KMALLOCS(ips_seed, u_long *, fr_statesize * sizeof(*ips_seed)); + if (ips_seed == NULL) + return -2; +#if defined(NEED_LOCAL_RAND) || !defined(_KERNEL) + tv.tv_sec = 0; + GETKTIME(&tv); +#endif + for (i = 0; i < fr_statesize; i++) { + /* + * XXX - ips_seed[X] should be a random number of sorts. + */ +#if !defined(NEED_LOCAL_RAND) && defined(_KERNEL) + ips_seed[i] = arc4random(); +#else + ips_seed[i] = ((u_long)ips_seed + i) * fr_statesize; + ips_seed[i] += tv.tv_sec; + ips_seed[i] *= (u_long)ips_seed; + ips_seed[i] ^= 0x5a5aa5a5; + ips_seed[i] *= fr_statemax; +#endif + } +#if defined(NEED_LOCAL_RAND) && defined(_KERNEL) + ipf_rand_push(ips_seed, fr_statesize * sizeof(*ips_seed)); +#endif + + /* 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 + + KMALLOCS(ips_stats.iss_bucketlen, u_long *, + fr_statesize * sizeof(u_long)); + if (ips_stats.iss_bucketlen == NULL) + return -1; + bzero((char *)ips_stats.iss_bucketlen, fr_statesize * sizeof(u_long)); + + if (fr_state_maxbucket == 0) { + for (i = fr_statesize; i > 0; i >>= 1) + fr_state_maxbucket++; + fr_state_maxbucket *= 2; + } + + ips_stats.iss_tcptab = ips_tqtqb; + fr_sttab_init(ips_tqtqb); + ips_tqtqb[IPF_TCP_NSTATES - 1].ifq_next = &ips_udptq; + ips_udptq.ifq_ttl = (u_long)fr_udptimeout; + ips_udptq.ifq_ref = 1; + ips_udptq.ifq_head = NULL; + ips_udptq.ifq_tail = &ips_udptq.ifq_head; + MUTEX_INIT(&ips_udptq.ifq_lock, "ipftq udp tab"); + ips_udptq.ifq_next = &ips_udpacktq; + ips_udpacktq.ifq_ttl = (u_long)fr_udpacktimeout; + ips_udpacktq.ifq_ref = 1; + ips_udpacktq.ifq_head = NULL; + ips_udpacktq.ifq_tail = &ips_udpacktq.ifq_head; + MUTEX_INIT(&ips_udpacktq.ifq_lock, "ipftq udpack tab"); + ips_udpacktq.ifq_next = &ips_icmptq; + ips_icmptq.ifq_ttl = (u_long)fr_icmptimeout; + ips_icmptq.ifq_ref = 1; + ips_icmptq.ifq_head = NULL; + ips_icmptq.ifq_tail = &ips_icmptq.ifq_head; + MUTEX_INIT(&ips_icmptq.ifq_lock, "ipftq icmp tab"); + ips_icmptq.ifq_next = &ips_icmpacktq; + ips_icmpacktq.ifq_ttl = (u_long)fr_icmpacktimeout; + ips_icmpacktq.ifq_ref = 1; + ips_icmpacktq.ifq_head = NULL; + ips_icmpacktq.ifq_tail = &ips_icmpacktq.ifq_head; + MUTEX_INIT(&ips_icmpacktq.ifq_lock, "ipftq icmpack tab"); + ips_icmpacktq.ifq_next = &ips_iptq; + ips_iptq.ifq_ttl = (u_long)fr_iptimeout; + ips_iptq.ifq_ref = 1; + ips_iptq.ifq_head = NULL; + ips_iptq.ifq_tail = &ips_iptq.ifq_head; + MUTEX_INIT(&ips_iptq.ifq_lock, "ipftq ip tab"); + ips_iptq.ifq_next = &ips_deletetq; + ips_deletetq.ifq_ttl = (u_long)1; + ips_deletetq.ifq_ref = 1; + ips_deletetq.ifq_head = NULL; + ips_deletetq.ifq_tail = &ips_deletetq.ifq_head; + MUTEX_INIT(&ips_deletetq.ifq_lock, "state delete queue"); + ips_deletetq.ifq_next = NULL; + + RWLOCK_INIT(&ipf_state, "ipf IP state rwlock"); + MUTEX_INIT(&ipf_stinsert, "ipf state insert mutex"); + fr_state_init = 1; + + ips_last_force_flush = fr_ticks; + return 0; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_stateunload */ +/* Returns: Nil */ +/* Parameters: Nil */ +/* */ +/* Release and destroy any resources acquired or initialised so that */ +/* IPFilter can be unloaded or re-initialised. */ +/* ------------------------------------------------------------------------ */ +void fr_stateunload() +{ + ipftq_t *ifq, *ifqnext; + ipstate_t *is; + + while ((is = ips_list) != NULL) + fr_delstate(is, ISL_UNLOAD); + + /* + * Proxy timeout queues are not cleaned here because although they + * exist on the state list, appr_unload is called after fr_stateunload + * and the proxies actually are responsible for them being created. + * Should the proxy timeouts have their own list? There's no real + * justification as this is the only complicationA + */ + for (ifq = ips_utqe; ifq != NULL; ifq = ifqnext) { + ifqnext = ifq->ifq_next; + if (((ifq->ifq_flags & IFQF_PROXY) == 0) && + (fr_deletetimeoutqueue(ifq) == 0)) + fr_freetimeoutqueue(ifq); + } + + ips_stats.iss_inuse = 0; + ips_num = 0; + + if (fr_state_init == 1) { + fr_sttab_destroy(ips_tqtqb); + MUTEX_DESTROY(&ips_udptq.ifq_lock); + MUTEX_DESTROY(&ips_icmptq.ifq_lock); + MUTEX_DESTROY(&ips_udpacktq.ifq_lock); + MUTEX_DESTROY(&ips_icmpacktq.ifq_lock); + MUTEX_DESTROY(&ips_iptq.ifq_lock); + MUTEX_DESTROY(&ips_deletetq.ifq_lock); + } + + if (ips_table != NULL) { + KFREES(ips_table, fr_statesize * sizeof(*ips_table)); + ips_table = NULL; + } + + if (ips_seed != NULL) { + KFREES(ips_seed, fr_statesize * sizeof(*ips_seed)); + ips_seed = NULL; + } + + if (ips_stats.iss_bucketlen != NULL) { + KFREES(ips_stats.iss_bucketlen, fr_statesize * sizeof(u_long)); + ips_stats.iss_bucketlen = NULL; + } + + if (fr_state_maxbucket_reset == 1) + fr_state_maxbucket = 0; + + if (fr_state_init == 1) { + fr_state_init = 0; + RW_DESTROY(&ipf_state); + MUTEX_DESTROY(&ipf_stinsert); + } +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_statetstats */ +/* Returns: ips_state_t* - pointer to state stats structure */ +/* Parameters: Nil */ +/* */ +/* Put all the current numbers and pointers into a single struct and return */ +/* a pointer to it. */ +/* ------------------------------------------------------------------------ */ +static ips_stat_t *fr_statetstats() +{ + ips_stats.iss_active = ips_num; + ips_stats.iss_statesize = fr_statesize; + ips_stats.iss_statemax = fr_statemax; + ips_stats.iss_table = ips_table; + ips_stats.iss_list = ips_list; + ips_stats.iss_ticks = fr_ticks; + return &ips_stats; +} + +/* ------------------------------------------------------------------------ */ +/* Function: fr_state_remove */ +/* Returns: int - 0 == success, != 0 == failure */ +/* Parameters: data(I) - pointer to state structure to delete from table */ +/* */ +/* Search for a state structure that matches the one passed, according to */ +/* the IP addresses and other protocol specific information. */ +/* ------------------------------------------------------------------------ */ +static int fr_state_remove(data) +caddr_t data; +{ + ipstate_t *sp, st; + int error; + + sp = &st; + error = fr_inobj(data, &st, IPFOBJ_IPSTATE); + 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((caddr_t)&sp->is_src, (caddr_t)&st.is_src, + sizeof(st.is_src)) && + !bcmp((caddr_t)&sp->is_dst, (caddr_t)&st.is_src, + sizeof(st.is_dst)) && + !bcmp((caddr_t)&sp->is_ps, (caddr_t)&st.is_ps, + sizeof(st.is_ps))) { + fr_delstate(sp, ISL_REMOVE); + RWLOCK_EXIT(&ipf_state); + return 0; + } + RWLOCK_EXIT(&ipf_state); + return ESRCH; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_state_ioctl */ +/* Returns: int - 0 == success, != 0 == failure */ +/* Parameters: data(I) - pointer to ioctl data */ +/* cmd(I) - ioctl command integer */ +/* mode(I) - file mode bits used with open */ +/* */ +/* Processes an ioctl call made to operate on the IP Filter state device. */ +/* ------------------------------------------------------------------------ */ +int fr_state_ioctl(data, cmd, mode, uid, ctx) +caddr_t data; +ioctlcmd_t cmd; +int mode, uid; +void *ctx; +{ + int arg, ret, error = 0; + SPL_INT(s); + + switch (cmd) + { + /* + * Delete an entry from the state table. + */ + case SIOCDELST : + error = fr_state_remove(data); + break; + + /* + * Flush the state table + */ + case SIOCIPFFL : + error = BCOPYIN(data, (char *)&arg, sizeof(arg)); + if (error != 0) { + error = EFAULT; + } else { + WRITE_ENTER(&ipf_state); + ret = fr_state_flush(arg, 4); + RWLOCK_EXIT(&ipf_state); + error = BCOPYOUT((char *)&ret, data, sizeof(ret)); + if (error != 0) + error = EFAULT; + } + break; + +#ifdef USE_INET6 + case SIOCIPFL6 : + error = BCOPYIN(data, (char *)&arg, sizeof(arg)); + if (error != 0) { + error = EFAULT; + } else { + WRITE_ENTER(&ipf_state); + ret = fr_state_flush(arg, 6); + RWLOCK_EXIT(&ipf_state); + error = BCOPYOUT((char *)&ret, data, sizeof(ret)); + if (error != 0) + error = EFAULT; + } + break; +#endif +#ifdef IPFILTER_LOG + /* + * Flush the state log. + */ + case SIOCIPFFB : + if (!(mode & FWRITE)) + error = EPERM; + else { + int tmp; + + tmp = ipflog_clear(IPL_LOGSTATE); + error = BCOPYOUT((char *)&tmp, data, sizeof(tmp)); + if (error != 0) + error = EFAULT; + } + break; + + /* + * Turn logging of state information on/off. + */ + case SIOCSETLG : + if (!(mode & FWRITE)) + error = EPERM; + else { + error = BCOPYIN((char *)data, (char *)&ipstate_logging, + sizeof(ipstate_logging)); + if (error != 0) + error = EFAULT; + } + break; + + /* + * Return the current state of logging. + */ + case SIOCGETLG : + error = BCOPYOUT((char *)&ipstate_logging, (char *)data, + sizeof(ipstate_logging)); + if (error != 0) + error = EFAULT; + break; + + /* + * Return the number of bytes currently waiting to be read. + */ + case FIONREAD : + arg = iplused[IPL_LOGSTATE]; /* returned in an int */ + error = BCOPYOUT((char *)&arg, data, sizeof(arg)); + if (error != 0) + error = EFAULT; + break; +#endif + + /* + * Get the current state statistics. + */ + case SIOCGETFS : + error = fr_outobj(data, fr_statetstats(), IPFOBJ_STATESTAT); + break; + + /* + * Lock/Unlock the state table. (Locking prevents any changes, which + * means no packets match). + */ + case SIOCSTLCK : + if (!(mode & FWRITE)) { + error = EPERM; + } else { + error = fr_lock(data, &fr_state_lock); + } + break; + + /* + * Add an entry to the current state table. + */ + case SIOCSTPUT : + if (!fr_state_lock || !(mode &FWRITE)) { + error = EACCES; + break; + } + error = fr_stputent(data); + break; + + /* + * Get a state table entry. + */ + case SIOCSTGET : + if (!fr_state_lock) { + error = EACCES; + break; + } + error = fr_stgetent(data); + break; + + /* + * Return a copy of the hash table bucket lengths + */ + case SIOCSTAT1 : + error = BCOPYOUT(ips_stats.iss_bucketlen, data, + fr_statesize * sizeof(u_long)); + if (error != 0) + error = EFAULT; + break; + + case SIOCGENITER : + { + ipftoken_t *token; + ipfgeniter_t iter; + + error = fr_inobj(data, &iter, IPFOBJ_GENITER); + if (error != 0) + break; + + SPL_SCHED(s); + token = ipf_findtoken(IPFGENITER_STATE, uid, ctx); + if (token != NULL) + error = fr_stateiter(token, &iter); + else + error = ESRCH; + RWLOCK_EXIT(&ipf_tokens); + SPL_X(s); + break; + } + + case SIOCGTABL : + error = fr_stgettable(data); + break; + + case SIOCIPFDELTOK : + error = BCOPYIN(data, (char *)&arg, sizeof(arg)); + if (error != 0) { + error = EFAULT; + } else { + SPL_SCHED(s); + error = ipf_deltoken(arg, uid, ctx); + SPL_X(s); + } + break; + + case SIOCGTQTAB : + error = fr_outobj(data, ips_tqtqb, IPFOBJ_STATETQTAB); + break; + + default : + error = EINVAL; + break; + } + return error; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_stgetent */ +/* Returns: int - 0 == success, != 0 == failure */ +/* Parameters: data(I) - pointer to state structure to retrieve from table */ +/* */ +/* Copy out state information from the kernel to a user space process. If */ +/* there is a filter rule associated with the state entry, copy that out */ +/* as well. The entry to copy out is taken from the value of "ips_next" in */ +/* the struct passed in and if not null and not found in the list of current*/ +/* state entries, the retrieval fails. */ +/* ------------------------------------------------------------------------ */ +int fr_stgetent(data) +caddr_t data; +{ + ipstate_t *is, *isn; + ipstate_save_t ips; + int error; + + error = fr_inobj(data, &ips, IPFOBJ_STATESAVE); + if (error != 0) + return error; + + READ_ENTER(&ipf_state); + isn = ips.ips_next; + if (isn == NULL) { + isn = ips_list; + if (isn == NULL) { + RWLOCK_EXIT(&ipf_state); + 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 == NULL) { + RWLOCK_EXIT(&ipf_state); + return ESRCH; + } + } + ips.ips_next = isn->is_next; + bcopy((char *)isn, (char *)&ips.ips_is, sizeof(ips.ips_is)); + ips.ips_rule = isn->is_rule; + if (isn->is_rule != NULL) + bcopy((char *)isn->is_rule, (char *)&ips.ips_fr, + sizeof(ips.ips_fr)); + RWLOCK_EXIT(&ipf_state); + error = fr_outobj(data, &ips, IPFOBJ_STATESAVE); + return error; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_stputent */ +/* Returns: int - 0 == success, != 0 == failure */ +/* Parameters: data(I) - pointer to state information struct */ +/* */ +/* This function implements the SIOCSTPUT ioctl: insert a state entry into */ +/* the state table. If the state info. includes a pointer to a filter rule */ +/* then also add in an orphaned rule (will not show up in any "ipfstat -io" */ +/* output. */ +/* ------------------------------------------------------------------------ */ +int fr_stputent(data) +caddr_t data; +{ + ipstate_t *is, *isn; + ipstate_save_t ips; + int error, out, i; + frentry_t *fr; + char *name; + + error = fr_inobj(data, &ips, IPFOBJ_STATESAVE); + if (error) + return EFAULT; + + KMALLOC(isn, ipstate_t *); + if (isn == NULL) + return ENOMEM; + + bcopy((char *)&ips.ips_is, (char *)isn, sizeof(*isn)); + bzero((char *)isn, offsetof(struct ipstate, is_pkts)); + isn->is_sti.tqe_pnext = NULL; + isn->is_sti.tqe_next = NULL; + isn->is_sti.tqe_ifq = NULL; + isn->is_sti.tqe_parent = isn; + isn->is_ifp[0] = NULL; + isn->is_ifp[1] = NULL; + isn->is_ifp[2] = NULL; + isn->is_ifp[3] = NULL; + isn->is_sync = NULL; + fr = ips.ips_rule; + + if (fr == NULL) { + READ_ENTER(&ipf_state); + fr_stinsert(isn, 0); + MUTEX_EXIT(&isn->is_lock); + RWLOCK_EXIT(&ipf_state); + return 0; + } + + if (isn->is_flags & SI_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; + MUTEX_NUKE(&fr->fr_lock); + MUTEX_INIT(&fr->fr_lock, "state filter rule lock"); + + /* + * Look up all the interface names in the rule. + */ + for (i = 0; i < 4; i++) { + name = fr->fr_ifnames[i]; + fr->fr_ifas[i] = fr_resolvenic(name, fr->fr_v); + name = isn->is_ifname[i]; + isn->is_ifp[i] = fr_resolvenic(name, isn->is_v); + } + + fr->fr_ref = 0; + fr->fr_dsize = 0; + fr->fr_data = NULL; + fr->fr_type = FR_T_NONE; + + fr_resolvedest(&fr->fr_tifs[0], fr->fr_v); + fr_resolvedest(&fr->fr_tifs[1], fr->fr_v); + fr_resolvedest(&fr->fr_dif, fr->fr_v); + + /* + * send a copy back to userland of what we ended up + * to allow for verification. + */ + error = fr_outobj(data, &ips, IPFOBJ_STATESAVE); + if (error) { + KFREE(isn); + MUTEX_DESTROY(&fr->fr_lock); + KFREE(fr); + return EFAULT; + } + READ_ENTER(&ipf_state); + fr_stinsert(isn, 0); + MUTEX_EXIT(&isn->is_lock); + RWLOCK_EXIT(&ipf_state); + + } else { + READ_ENTER(&ipf_state); + for (is = ips_list; is; is = is->is_next) + if (is->is_rule == fr) { + fr_stinsert(isn, 0); + MUTEX_EXIT(&isn->is_lock); + break; + } + + if (is == NULL) { + KFREE(isn); + isn = NULL; + } + RWLOCK_EXIT(&ipf_state); + + return (isn == NULL) ? ESRCH : 0; + } + + return 0; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_stinsert */ +/* Returns: Nil */ +/* Parameters: is(I) - pointer to state structure */ +/* rev(I) - flag indicating forward/reverse direction of packet */ +/* */ +/* Inserts a state structure into the hash table (for lookups) and the list */ +/* of state entries (for enumeration). Resolves all of the interface names */ +/* to pointers and adjusts running stats for the hash table as appropriate. */ +/* */ +/* Locking: it is assumed that some kind of lock on ipf_state is held. */ +/* Exits with is_lock initialised and held. */ +/* ------------------------------------------------------------------------ */ +void fr_stinsert(is, rev) +ipstate_t *is; +int rev; +{ + frentry_t *fr; + u_int hv; + int i; + + MUTEX_INIT(&is->is_lock, "ipf state entry"); + + fr = is->is_rule; + if (fr != NULL) { + MUTEX_ENTER(&fr->fr_lock); + fr->fr_ref++; + fr->fr_statecnt++; + MUTEX_EXIT(&fr->fr_lock); + } + + /* + * Look up all the interface names in the state entry. + */ + for (i = 0; i < 4; i++) { + if (is->is_ifp[i] != NULL) + continue; + is->is_ifp[i] = fr_resolvenic(is->is_ifname[i], is->is_v); + } + + /* + * If we could trust is_hv, then the modulous would not be needed, but + * when running with IPFILTER_SYNC, this stops bad values. + */ + hv = is->is_hv % fr_statesize; + is->is_hv = hv; + + /* + * We need to get both of these locks...the first because it is + * possible that once the insert is complete another packet might + * come along, match the entry and want to update it. + */ + MUTEX_ENTER(&is->is_lock); + MUTEX_ENTER(&ipf_stinsert); + + /* + * add into list table. + */ + if (ips_list != NULL) + ips_list->is_pnext = &is->is_next; + is->is_pnext = &ips_list; + is->is_next = ips_list; + ips_list = is; + + if (ips_table[hv] != NULL) + 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_stats.iss_bucketlen[hv]++; + ips_num++; + MUTEX_EXIT(&ipf_stinsert); + + fr_setstatequeue(is, rev); +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_addstate */ +/* Returns: ipstate_t* - NULL == failure, else pointer to new state */ +/* Parameters: fin(I) - pointer to packet information */ +/* stsave(O) - pointer to place to save pointer to created */ +/* state structure. */ +/* flags(I) - flags to use when creating the structure */ +/* */ +/* Creates a new IP state structure from the packet information collected. */ +/* Inserts it into the state table and appends to the bottom of the active */ +/* list. If the capacity of the table has reached the maximum allowed then */ +/* the call will fail and a flush is scheduled for the next timeout call. */ +/* */ +/* NOTE: The use of stsave to point to nat_state will result in memory */ +/* corruption. It should only be used to point to objects that will */ +/* either outlive this (not expired) or will deref the ip_state_t */ +/* when they are deleted. */ +/* ------------------------------------------------------------------------ */ +ipstate_t *fr_addstate(fin, stsave, flags) +fr_info_t *fin; +ipstate_t **stsave; +u_int flags; +{ + ipstate_t *is, ips; + struct icmp *ic; + u_int pass, hv; + frentry_t *fr; + tcphdr_t *tcp; + grehdr_t *gre; + int out; + + if (fr_state_lock || + (fin->fin_flx & (FI_SHORT|FI_STATE|FI_FRAGBODY|FI_BAD))) + return NULL; + + if ((fin->fin_flx & FI_OOW) && !(fin->fin_tcpf & TH_SYN)) + return NULL; + + /* + * If a "keep state" rule has reached the maximum number of references + * to it, then schedule an automatic flush in case we can clear out + * some "dead old wood". Note that because the lock isn't held on + * fr it is possible that we could overflow. The cost of overflowing + * is being ignored here as the number by which it can overflow is + * a product of the number of simultaneous threads that could be + * executing in here, so a limit of 100 won't result in 200, but could + * result in 101 or 102. + */ + fr = fin->fin_fr; + if (fr != NULL) { + if ((ips_num >= fr_statemax) && (fr->fr_statemax == 0)) { + ATOMIC_INCL(ips_stats.iss_max); + fr_state_doflush = 1; + return NULL; + } + if ((fr->fr_statemax != 0) && + (fr->fr_statecnt >= fr->fr_statemax)) { + ATOMIC_INCL(ips_stats.iss_maxref); + return NULL; + } + } + + pass = (fr == NULL) ? 0 : fr->fr_flags; + + ic = NULL; + tcp = NULL; + out = fin->fin_out; + is = &ips; + bzero((char *)is, sizeof(*is)); + is->is_die = 1 + fr_ticks; + + /* + * 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) { + /* + * For ICMPv6, we check to see if the destination address is + * a multicast address. If it is, do not include it in the + * calculation of the hash because the correct reply will come + * back from a real address, not a multicast address. + */ + if ((is->is_p == IPPROTO_ICMPV6) && + IN6_IS_ADDR_MULTICAST(&is->is_dst.in6)) { + /* + * So you can do keep state with neighbour discovery. + * + * Here we could use the address from the neighbour + * solicit message to put in the state structure and + * we could use that without a wildcard flag too... + */ + flags |= SI_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 + if ((fin->fin_v == 4) && + (fin->fin_flx & (FI_MULTICAST|FI_BROADCAST|FI_MBCAST))) { + if (fin->fin_out == 0) { + flags |= SI_W_DADDR|SI_CLONE; + hv -= is->is_daddr; + } else { + flags |= SI_W_SADDR|SI_CLONE; + hv -= is->is_saddr; + } + } + + switch (is->is_p) + { +#ifdef USE_INET6 + case IPPROTO_ICMPV6 : + ic = fin->fin_dp; + + switch (ic->icmp_type) + { + case ICMP6_ECHO_REQUEST : + is->is_icmp.ici_type = ic->icmp_type; + hv += (is->is_icmp.ici_id = ic->icmp_id); + break; + case ICMP6_MEMBERSHIP_QUERY : + case ND_ROUTER_SOLICIT : + case ND_NEIGHBOR_SOLICIT : + case ICMP6_NI_QUERY : + is->is_icmp.ici_type = ic->icmp_type; + break; + default : + return NULL; + } + ATOMIC_INCL(ips_stats.iss_icmp); + break; +#endif + case IPPROTO_ICMP : + ic = fin->fin_dp; + + switch (ic->icmp_type) + { + case ICMP_ECHO : + case ICMP_TSTAMP : + case ICMP_IREQ : + case ICMP_MASKREQ : + is->is_icmp.ici_type = ic->icmp_type; + hv += (is->is_icmp.ici_id = ic->icmp_id); + break; + default : + return NULL; + } + ATOMIC_INCL(ips_stats.iss_icmp); + break; + + case IPPROTO_GRE : + gre = fin->fin_dp; + + is->is_gre.gs_flags = gre->gr_flags; + is->is_gre.gs_ptype = gre->gr_ptype; + if (GRE_REV(is->is_gre.gs_flags) == 1) { + is->is_call[0] = fin->fin_data[0]; + is->is_call[1] = fin->fin_data[1]; + } + break; + + case IPPROTO_TCP : + tcp = 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 & (SI_W_DPORT|SI_W_SPORT)) == 0) { + hv += is->is_sport; + hv += is->is_dport; + } + + /* + * If this is a real packet then initialise fields in the + * state information structure from the TCP header information. + */ + + is->is_maxdwin = 1; + is->is_maxswin = ntohs(tcp->th_win); + if (is->is_maxswin == 0) + is->is_maxswin = 1; + + if ((fin->fin_flx & FI_IGNORE) == 0) { + is->is_send = ntohl(tcp->th_seq) + fin->fin_dlen - + (TCP_OFF(tcp) << 2) + + ((tcp->th_flags & TH_SYN) ? 1 : 0) + + ((tcp->th_flags & TH_FIN) ? 1 : 0); + is->is_maxsend = is->is_send; + + /* + * Window scale option is only present in + * SYN/SYN-ACK packet. + */ + if ((tcp->th_flags & ~(TH_FIN|TH_ACK|TH_ECNALL)) == + TH_SYN && + (TCP_OFF(tcp) > (sizeof(tcphdr_t) >> 2))) { + if (fr_tcpoptions(fin, tcp, + &is->is_tcp.ts_data[0]) == -1) { + fin->fin_flx |= FI_BAD; + } + } + + if ((fin->fin_out != 0) && (pass & FR_NEWISN) != 0) { + fr_checknewisn(fin, is); + fr_fixoutisn(fin, is); + } + + if ((tcp->th_flags & TH_OPENING) == TH_SYN) + flags |= IS_TCPFSM; + else { + is->is_maxdwin = is->is_maxswin * 2; + is->is_dend = ntohl(tcp->th_ack); + is->is_maxdend = ntohl(tcp->th_ack); + is->is_maxdwin *= 2; + } + } + + /* + * 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 = fin->fin_dp; + + is->is_sport = htons(fin->fin_data[0]); + is->is_dport = htons(fin->fin_data[1]); + if ((flags & (SI_W_DPORT|SI_W_SPORT)) == 0) { + hv += tcp->th_dport; + hv += tcp->th_sport; + } + ATOMIC_INCL(ips_stats.iss_udp); + break; + + default : + break; + } + hv = DOUBLE_HASH(hv); + is->is_hv = hv; + is->is_rule = fr; + is->is_flags = flags & IS_INHERITED; + + /* + * Look for identical state. + */ + for (is = ips_table[is->is_hv % fr_statesize]; is != NULL; + is = is->is_hnext) { + if (bcmp(&ips.is_src, &is->is_src, + offsetof(struct ipstate, is_ps) - + offsetof(struct ipstate, is_src)) == 0) + break; + } + if (is != NULL) + return NULL; + + if (ips_stats.iss_bucketlen[hv] >= fr_state_maxbucket) { + ATOMIC_INCL(ips_stats.iss_bucketfull); + return NULL; + } + KMALLOC(is, ipstate_t *); + if (is == NULL) { + ATOMIC_INCL(ips_stats.iss_nomem); + return NULL; + } + bcopy((char *)&ips, (char *)is, sizeof(*is)); + /* + * Do not do the modulous here, it is done in fr_stinsert(). + */ + if (fr != NULL) { + (void) strncpy(is->is_group, fr->fr_group, FR_GROUPLEN); + if (fr->fr_age[0] != 0) { + is->is_tqehead[0] = fr_addtimeoutqueue(&ips_utqe, + fr->fr_age[0]); + is->is_sti.tqe_flags |= TQE_RULEBASED; + } + if (fr->fr_age[1] != 0) { + is->is_tqehead[1] = fr_addtimeoutqueue(&ips_utqe, + fr->fr_age[1]); + is->is_sti.tqe_flags |= TQE_RULEBASED; + } + + is->is_tag = fr->fr_logtag; + + /* + * The name '-' is special for network interfaces and causes + * a NULL name to be present, always, allowing packets to + * match it, regardless of their interface. + */ + if ((fin->fin_ifp == NULL) || + (fr->fr_ifnames[out << 1][0] == '-' && + fr->fr_ifnames[out << 1][1] == '\0')) { + is->is_ifp[out << 1] = fr->fr_ifas[0]; + strncpy(is->is_ifname[out << 1], fr->fr_ifnames[0], + sizeof(fr->fr_ifnames[0])); + } else { + is->is_ifp[out << 1] = fin->fin_ifp; + COPYIFNAME(is->is_v, fin->fin_ifp, + is->is_ifname[out << 1]); + } + + is->is_ifp[(out << 1) + 1] = fr->fr_ifas[1]; + strncpy(is->is_ifname[(out << 1) + 1], fr->fr_ifnames[1], + sizeof(fr->fr_ifnames[1])); + + is->is_ifp[(1 - out) << 1] = fr->fr_ifas[2]; + strncpy(is->is_ifname[((1 - out) << 1)], fr->fr_ifnames[2], + sizeof(fr->fr_ifnames[2])); + + is->is_ifp[((1 - out) << 1) + 1] = fr->fr_ifas[3]; + strncpy(is->is_ifname[((1 - out) << 1) + 1], fr->fr_ifnames[3], + sizeof(fr->fr_ifnames[3])); + } else { + pass = fr_flags; + is->is_tag = FR_NOLOGTAG; + + if (fin->fin_ifp != NULL) { + is->is_ifp[out << 1] = fin->fin_ifp; + COPYIFNAME(is->is_v, fin->fin_ifp, + is->is_ifname[out << 1]); + } + } + + /* + * It may seem strange to set is_ref to 2, but fr_check() will call + * fr_statederef() after calling fr_addstate() and the idea is to + * have it exist at the end of fr_check() with is_ref == 1. + */ + is->is_ref = 2; + is->is_pass = pass; + is->is_pkts[0] = 0, is->is_bytes[0] = 0; + is->is_pkts[1] = 0, is->is_bytes[1] = 0; + is->is_pkts[2] = 0, is->is_bytes[2] = 0; + is->is_pkts[3] = 0, is->is_bytes[3] = 0; + if ((fin->fin_flx & FI_IGNORE) == 0) { + is->is_pkts[out] = 1; + is->is_bytes[out] = fin->fin_plen; + is->is_flx[out][0] = fin->fin_flx & FI_CMP; + is->is_flx[out][0] &= ~FI_OOW; + } + + if (pass & FR_STSTRICT) + is->is_flags |= IS_STRICT; + + if (pass & FR_STATESYNC) + is->is_flags |= IS_STATESYNC; + + /* + * 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_opt[0] = fin->fin_optmsk; + is->is_optmsk[0] = 0xffffffff; + is->is_optmsk[1] = 0xffffffff; + if (is->is_v == 6) { + is->is_opt[0] &= ~0x8; + is->is_optmsk[0] &= ~0x8; + is->is_optmsk[1] &= ~0x8; + } + is->is_me = stsave; + is->is_sec = fin->fin_secmsk; + is->is_secmsk = 0xffff; + is->is_auth = fin->fin_auth; + is->is_authmsk = 0xffff; + if (flags & (SI_WILDP|SI_WILDA)) { + ATOMIC_INCL(ips_stats.iss_wild); + } + is->is_rulen = fin->fin_rule; + + + if (pass & FR_LOGFIRST) + is->is_pass &= ~(FR_LOGFIRST|FR_LOG); + + READ_ENTER(&ipf_state); + + fr_stinsert(is, fin->fin_rev); + + if (fin->fin_p == IPPROTO_TCP) { + /* + * 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. + */ + (void) fr_tcp_age(&is->is_sti, fin, ips_tqtqb, is->is_flags); + MUTEX_EXIT(&is->is_lock); +#ifdef IPFILTER_SCAN + if ((is->is_flags & SI_CLONE) == 0) + (void) ipsc_attachis(is); +#endif + } else { + MUTEX_EXIT(&is->is_lock); + } +#ifdef IPFILTER_SYNC + if ((is->is_flags & IS_STATESYNC) && ((is->is_flags & SI_CLONE) == 0)) + is->is_sync = ipfsync_new(SMC_STATE, fin, is); +#endif + if (ipstate_logging) + ipstate_log(is, ISL_NEW); + + RWLOCK_EXIT(&ipf_state); + fin->fin_state = is; + fin->fin_rev = IP6_NEQ(&is->is_dst, &fin->fin_daddr); + fin->fin_flx |= FI_STATE; + if (fin->fin_flx & FI_FRAG) + (void) fr_newfrag(fin, pass ^ FR_KEEPSTATE); + + return is; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_tcpoptions */ +/* Returns: int - 1 == packet matches state entry, 0 == it does not, */ +/* -1 == packet has bad TCP options data */ +/* Parameters: fin(I) - pointer to packet information */ +/* tcp(I) - pointer to TCP packet header */ +/* td(I) - pointer to TCP data held as part of the state */ +/* */ +/* Look after the TCP header for any options and deal with those that are */ +/* present. Record details about those that we recogise. */ +/* ------------------------------------------------------------------------ */ +static int fr_tcpoptions(fin, tcp, td) +fr_info_t *fin; +tcphdr_t *tcp; +tcpdata_t *td; +{ + int off, mlen, ol, i, len, retval; + char buf[64], *s, opt; + mb_t *m = NULL; + + len = (TCP_OFF(tcp) << 2); + if (fin->fin_dlen < len) + return 0; + len -= sizeof(*tcp); + + off = fin->fin_plen - fin->fin_dlen + sizeof(*tcp) + fin->fin_ipoff; + + m = fin->fin_m; + mlen = MSGDSIZE(m) - off; + if (len > mlen) { + len = mlen; + retval = 0; + } else { + retval = 1; + } + + COPYDATA(m, off, len, buf); + + for (s = buf; len > 0; ) { + opt = *s; + if (opt == TCPOPT_EOL) + break; + else if (opt == TCPOPT_NOP) + ol = 1; + else { + if (len < 2) + break; + ol = (int)*(s + 1); + if (ol < 2 || ol > len) + break; + + /* + * Extract the TCP options we are interested in out of + * the header and store them in the the tcpdata struct. + */ + switch (opt) + { + case TCPOPT_WINDOW : + if (ol == TCPOLEN_WINDOW) { + i = (int)*(s + 2); + if (i > TCP_WSCALE_MAX) + i = TCP_WSCALE_MAX; + else if (i < 0) + i = 0; + td->td_winscale = i; + td->td_winflags |= TCP_WSCALE_SEEN| + TCP_WSCALE_FIRST; + } else + retval = -1; + break; + case TCPOPT_MAXSEG : + /* + * So, if we wanted to set the TCP MAXSEG, + * it should be done here... + */ + if (ol == TCPOLEN_MAXSEG) { + i = (int)*(s + 2); + i <<= 8; + i += (int)*(s + 3); + td->td_maxseg = i; + } else + retval = -1; + break; + case TCPOPT_SACK_PERMITTED : + if (ol == TCPOLEN_SACK_PERMITTED) + td->td_winflags |= TCP_SACK_PERMIT; + else + retval = -1; + break; + } + } + len -= ol; + s += ol; + } + return retval; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_tcpstate */ +/* Returns: int - 1 == packet matches state entry, 0 == it does not */ +/* Parameters: fin(I) - pointer to packet information */ +/* tcp(I) - pointer to TCP packet header */ +/* is(I) - pointer to master state structure */ +/* */ +/* 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. */ +/* ------------------------------------------------------------------------ */ +static int fr_tcpstate(fin, tcp, is) +fr_info_t *fin; +tcphdr_t *tcp; +ipstate_t *is; +{ + int source, ret = 0, flags; + tcpdata_t *fdata, *tdata; + + source = !fin->fin_rev; + if (((is->is_flags & IS_TCPFSM) != 0) && (source == 1) && + (ntohs(is->is_sport) != fin->fin_data[0])) + source = 0; + fdata = &is->is_tcp.ts_data[!source]; + tdata = &is->is_tcp.ts_data[source]; + + MUTEX_ENTER(&is->is_lock); + + /* + * If a SYN packet is received for a connection that is on the way out + * but hasn't yet departed then advance this session along the way. + */ + if ((tcp->th_flags & TH_OPENING) == TH_SYN) { + if ((is->is_state[0] > IPF_TCPS_ESTABLISHED) && + (is->is_state[1] > IPF_TCPS_ESTABLISHED)) { + is->is_state[!source] = IPF_TCPS_CLOSED; + fr_movequeue(&is->is_sti, is->is_sti.tqe_ifq, + &ips_deletetq); + MUTEX_EXIT(&is->is_lock); + return 0; + } + } + + ret = fr_tcpinwindow(fin, fdata, tdata, tcp, is->is_flags); + if (ret > 0) { +#ifdef IPFILTER_SCAN + if (is->is_flags & (IS_SC_CLIENT|IS_SC_SERVER)) { + ipsc_packet(fin, is); + if (FR_ISBLOCK(is->is_pass)) { + MUTEX_EXIT(&is->is_lock); + return 1; + } + } +#endif + + /* + * Nearing end of connection, start timeout. + */ + ret = fr_tcp_age(&is->is_sti, fin, ips_tqtqb, is->is_flags); + if (ret == 0) { + MUTEX_EXIT(&is->is_lock); + return 0; + } + + /* + * set s0's as appropriate. Use syn-ack packet as it + * contains both pieces of required information. + */ + /* + * Window scale option is only present in SYN/SYN-ACK packet. + * Compare with ~TH_FIN to mask out T/TCP setups. + */ + flags = tcp->th_flags & ~(TH_FIN|TH_ECNALL); + if (flags == (TH_SYN|TH_ACK)) { + is->is_s0[source] = ntohl(tcp->th_ack); + is->is_s0[!source] = ntohl(tcp->th_seq) + 1; + if ((TCP_OFF(tcp) > (sizeof(tcphdr_t) >> 2))) { + if (fr_tcpoptions(fin, tcp, fdata) == -1) + fin->fin_flx |= FI_BAD; + } + if ((fin->fin_out != 0) && (is->is_pass & FR_NEWISN)) + fr_checknewisn(fin, is); + } else if (flags == TH_SYN) { + is->is_s0[source] = ntohl(tcp->th_seq) + 1; + if ((TCP_OFF(tcp) > (sizeof(tcphdr_t) >> 2))) { + if (fr_tcpoptions(fin, tcp, fdata) == -1) + fin->fin_flx |= FI_BAD; + } + + if ((fin->fin_out != 0) && (is->is_pass & FR_NEWISN)) + fr_checknewisn(fin, is); + + } + ret = 1; + } else { + fin->fin_flx |= FI_OOW; + } + MUTEX_EXIT(&is->is_lock); + return ret; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_checknewisn */ +/* Returns: Nil */ +/* Parameters: fin(I) - pointer to packet information */ +/* is(I) - pointer to master state structure */ +/* */ +/* Check to see if this TCP connection is expecting and needs a new */ +/* sequence number for a particular direction of the connection. */ +/* */ +/* NOTE: This does not actually change the sequence numbers, only gets new */ +/* one ready. */ +/* ------------------------------------------------------------------------ */ +static void fr_checknewisn(fin, is) +fr_info_t *fin; +ipstate_t *is; +{ + u_32_t sumd, old, new; + tcphdr_t *tcp; + int i; + + i = fin->fin_rev; + tcp = fin->fin_dp; + + if (((i == 0) && !(is->is_flags & IS_ISNSYN)) || + ((i == 1) && !(is->is_flags & IS_ISNACK))) { + old = ntohl(tcp->th_seq); + new = fr_newisn(fin); + is->is_isninc[i] = new - old; + CALC_SUMD(old, new, sumd); + is->is_sumd[i] = (sumd & 0xffff) + (sumd >> 16); + + is->is_flags |= ((i == 0) ? IS_ISNSYN : IS_ISNACK); + } +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_tcpinwindow */ +/* Returns: int - 1 == packet inside TCP "window", 0 == not inside, */ +/* 2 == packet seq number matches next expected */ +/* Parameters: fin(I) - pointer to packet information */ +/* fdata(I) - pointer to tcp state informatio (forward) */ +/* tdata(I) - pointer to tcp state informatio (reverse) */ +/* tcp(I) - pointer to TCP packet header */ +/* */ +/* Given a packet has matched addresses and ports, check to see if it is */ +/* within the TCP data window. In a show of generosity, allow packets that */ +/* are within the window space behind the current sequence # as well. */ +/* ------------------------------------------------------------------------ */ +int fr_tcpinwindow(fin, fdata, tdata, tcp, flags) +fr_info_t *fin; +tcpdata_t *fdata, *tdata; +tcphdr_t *tcp; +int flags; +{ + tcp_seq seq, ack, end; + int ackskew, tcpflags; + u_32_t win, maxwin; + int dsize, inseq; + + /* + * Find difference between last checked packet and this packet. + */ + tcpflags = tcp->th_flags; + seq = ntohl(tcp->th_seq); + ack = ntohl(tcp->th_ack); + if (tcpflags & TH_SYN) + win = ntohs(tcp->th_win); + else + win = ntohs(tcp->th_win) << fdata->td_winscale; + + /* + * A window of 0 produces undesirable behaviour from this function. + */ + if (win == 0) + win = 1; + + dsize = fin->fin_dlen - (TCP_OFF(tcp) << 2) + + ((tcpflags & TH_SYN) ? 1 : 0) + ((tcpflags & TH_FIN) ? 1 : 0); + + /* + * if window scaling is present, the scaling is only allowed + * for windows not in the first SYN packet. In that packet the + * window is 65535 to specify the largest window possible + * for receivers not implementing the window scale option. + * Currently, we do not assume TTCP here. That means that + * if we see a second packet from a host (after the initial + * SYN), we can assume that the receiver of the SYN did + * already send back the SYN/ACK (and thus that we know if + * the receiver also does window scaling) + */ + if (!(tcpflags & TH_SYN) && (fdata->td_winflags & TCP_WSCALE_FIRST)) { + fdata->td_winflags &= ~TCP_WSCALE_FIRST; + fdata->td_maxwin = win; + } + + end = seq + dsize; + + if ((fdata->td_end == 0) && + (!(flags & IS_TCPFSM) || + ((tcpflags & TH_OPENING) == TH_OPENING))) { + /* + * Must be a (outgoing) SYN-ACK in reply to a SYN. + */ + fdata->td_end = end - 1; + fdata->td_maxwin = 1; + fdata->td_maxend = end + win; + } + + if (!(tcpflags & TH_ACK)) { /* Pretend an ack was sent */ + ack = tdata->td_end; + } else if (((tcpflags & (TH_ACK|TH_RST)) == (TH_ACK|TH_RST)) && + (ack == 0)) { + /* gross hack to get around certain broken tcp stacks */ + ack = tdata->td_end; + } + + maxwin = tdata->td_maxwin; + ackskew = tdata->td_end - ack; + + /* + * Strict sequencing only allows in-order delivery. + */ + if (seq != fdata->td_end) { + if ((flags & IS_STRICT) != 0) { + return 0; + } + } + + inseq = 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 << fdata->td_winscale))) { + inseq = 1; + /* + * Microsoft Windows will send the next packet to the right of the + * window if SACK is in use. + */ + } else if ((seq == fdata->td_maxend) && (ackskew == 0) && + (fdata->td_winflags & TCP_SACK_PERMIT) && + (tdata->td_winflags & TCP_SACK_PERMIT)) { + inseq = 1; + /* + * Sometimes a TCP RST will be generated with only the ACK field + * set to non-zero. + */ + } else if ((seq == 0) && (tcpflags == (TH_RST|TH_ACK)) && + (ackskew >= -1) && (ackskew <= 1)) { + inseq = 1; + } else if (!(flags & IS_TCPFSM)) { + int i; + + i = (fin->fin_rev << 1) + fin->fin_out; + +#if 0 + if (is_pkts[i]0 == 0) { + /* + * Picking up a connection in the middle, the "next" + * packet seen from a direction that is new should be + * accepted, even if it appears out of sequence. + */ + inseq = 1; + } else +#endif + if (!(fdata->td_winflags & + (TCP_WSCALE_SEEN|TCP_WSCALE_FIRST))) { + /* + * No TCPFSM and no window scaling, so make some + * extra guesses. + */ + if ((seq == fdata->td_maxend) && (ackskew == 0)) + inseq = 1; + else if (SEQ_GE(seq + maxwin, fdata->td_end - maxwin)) + inseq = 1; + } + } + + /* TRACE(inseq, fdata, tdata, seq, end, ack, ackskew, win, maxwin) */ + + if (inseq) { + /* if ackskew < 0 then this should be due to fragmented + * 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 parameters 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. + */ + 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; + return 1; + } + return 0; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_stclone */ +/* Returns: ipstate_t* - NULL == cloning failed, */ +/* else pointer to new state structure */ +/* Parameters: fin(I) - pointer to packet information */ +/* tcp(I) - pointer to TCP/UDP header */ +/* is(I) - pointer to master state structure */ +/* */ +/* Create a "duplcate" state table entry from the master. */ +/* ------------------------------------------------------------------------ */ +static ipstate_t *fr_stclone(fin, tcp, is) +fr_info_t *fin; +tcphdr_t *tcp; +ipstate_t *is; +{ + ipstate_t *clone; + u_32_t send; + + if (ips_num == fr_statemax) { + ATOMIC_INCL(ips_stats.iss_max); + fr_state_doflush = 1; + return NULL; + } + KMALLOC(clone, ipstate_t *); + if (clone == NULL) + return NULL; + bcopy((char *)is, (char *)clone, sizeof(*clone)); + + MUTEX_NUKE(&clone->is_lock); + + clone->is_die = ONE_DAY + fr_ticks; + clone->is_state[0] = 0; + clone->is_state[1] = 0; + send = ntohl(tcp->th_seq) + fin->fin_dlen - (TCP_OFF(tcp) << 2) + + ((tcp->th_flags & TH_SYN) ? 1 : 0) + + ((tcp->th_flags & TH_FIN) ? 1 : 0); + + if (fin->fin_rev == 1) { + clone->is_dend = send; + clone->is_maxdend = send; + clone->is_send = 0; + clone->is_maxswin = 1; + clone->is_maxdwin = ntohs(tcp->th_win); + if (clone->is_maxdwin == 0) + clone->is_maxdwin = 1; + } else { + clone->is_send = send; + clone->is_maxsend = send; + clone->is_dend = 0; + clone->is_maxdwin = 1; + clone->is_maxswin = ntohs(tcp->th_win); + if (clone->is_maxswin == 0) + clone->is_maxswin = 1; + } + + clone->is_flags &= ~SI_CLONE; + clone->is_flags |= SI_CLONED; + fr_stinsert(clone, fin->fin_rev); + clone->is_ref = 2; + if (clone->is_p == IPPROTO_TCP) { + (void) fr_tcp_age(&clone->is_sti, fin, ips_tqtqb, + clone->is_flags); + } + MUTEX_EXIT(&clone->is_lock); +#ifdef IPFILTER_SCAN + (void) ipsc_attachis(is); +#endif +#ifdef IPFILTER_SYNC + if (is->is_flags & IS_STATESYNC) + clone->is_sync = ipfsync_new(SMC_STATE, fin, clone); +#endif + return clone; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_matchsrcdst */ +/* Returns: Nil */ +/* Parameters: fin(I) - pointer to packet information */ +/* is(I) - pointer to state structure */ +/* src(I) - pointer to source address */ +/* dst(I) - pointer to destination address */ +/* tcp(I) - pointer to TCP/UDP header */ +/* */ +/* Match a state table entry against an IP packet. The logic below is that */ +/* ret gets set to one if the match succeeds, else remains 0. If it is */ +/* still 0 after the test. no match. */ +/* ------------------------------------------------------------------------ */ +static ipstate_t *fr_matchsrcdst(fin, is, src, dst, tcp, cmask) +fr_info_t *fin; +ipstate_t *is; +i6addr_t *src, *dst; +tcphdr_t *tcp; +u_32_t cmask; +{ + int ret = 0, rev, out, flags, flx = 0, idx; + u_short sp, dp; + u_32_t cflx; + void *ifp; + + rev = IP6_NEQ(&is->is_dst, dst); + ifp = fin->fin_ifp; + out = fin->fin_out; + flags = is->is_flags; + sp = 0; + dp = 0; + + if (tcp != NULL) { + sp = htons(fin->fin_sport); + dp = ntohs(fin->fin_dport); + } + if (!rev) { + if (tcp != NULL) { + if (!(flags & SI_W_SPORT) && (sp != is->is_sport)) + rev = 1; + else if (!(flags & SI_W_DPORT) && (dp != is->is_dport)) + rev = 1; + } + } + + idx = (out << 1) + rev; + + /* + * If the interface for this 'direction' is set, make sure it matches. + * An interface name that is not set matches any, as does a name of *. + */ + if ((is->is_ifp[idx] == ifp) || (is->is_ifp[idx] == NULL && + (*is->is_ifname[idx] == '\0' || *is->is_ifname[idx] == '-' || + *is->is_ifname[idx] == '*'))) + ret = 1; + + if (ret == 0) + return NULL; + ret = 0; + + /* + * Match addresses and ports. + */ + if (rev == 0) { + if ((IP6_EQ(&is->is_dst, dst) || (flags & SI_W_DADDR)) && + (IP6_EQ(&is->is_src, src) || (flags & SI_W_SADDR))) { + if (tcp) { + if ((sp == is->is_sport || flags & SI_W_SPORT)&& + (dp == is->is_dport || flags & SI_W_DPORT)) + ret = 1; + } else { + ret = 1; + } + } + } else { + if ((IP6_EQ(&is->is_dst, src) || (flags & SI_W_DADDR)) && + (IP6_EQ(&is->is_src, dst) || (flags & SI_W_SADDR))) { + if (tcp) { + if ((dp == is->is_sport || flags & SI_W_SPORT)&& + (sp == is->is_dport || flags & SI_W_DPORT)) + ret = 1; + } else { + ret = 1; + } + } + } + + if (ret == 0) + return NULL; + + /* + * 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 & ~(SI_WILDP|SI_NEWFR|SI_CLONE|SI_CLONED); + + /* + * Only one of the source or destination address can be flaged as a + * wildcard. Fill in the missing address, if set. + * For IPv6, if the address being copied in is multicast, then + * don't reset the wild flag - multicast causes it to be set in the + * first place! + */ + if ((flags & (SI_W_SADDR|SI_W_DADDR))) { + fr_ip_t *fi = &fin->fin_fi; + + if ((flags & SI_W_SADDR) != 0) { + if (rev == 0) { +#ifdef USE_INET6 + if (is->is_v == 6 && + IN6_IS_ADDR_MULTICAST(&fi->fi_src.in6)) + /*EMPTY*/; + else +#endif + { + is->is_src = fi->fi_src; + is->is_flags &= ~SI_W_SADDR; + } + } else { +#ifdef USE_INET6 + if (is->is_v == 6 && + IN6_IS_ADDR_MULTICAST(&fi->fi_dst.in6)) + /*EMPTY*/; + else +#endif + { + is->is_src = fi->fi_dst; + is->is_flags &= ~SI_W_SADDR; + } + } + } else if ((flags & SI_W_DADDR) != 0) { + if (rev == 0) { +#ifdef USE_INET6 + if (is->is_v == 6 && + IN6_IS_ADDR_MULTICAST(&fi->fi_dst.in6)) + /*EMPTY*/; + else +#endif + { + is->is_dst = fi->fi_dst; + is->is_flags &= ~SI_W_DADDR; + } + } else { +#ifdef USE_INET6 + if (is->is_v == 6 && + IN6_IS_ADDR_MULTICAST(&fi->fi_src.in6)) + /*EMPTY*/; + else +#endif + { + is->is_dst = fi->fi_src; + is->is_flags &= ~SI_W_DADDR; + } + } + } + if ((is->is_flags & (SI_WILDA|SI_WILDP)) == 0) { + ATOMIC_DECL(ips_stats.iss_wild); + } + } + + flx = fin->fin_flx & cmask; + cflx = is->is_flx[out][rev]; + + /* + * Match up any flags set from IP options. + */ + if ((cflx && (flx != (cflx & cmask))) || + ((fin->fin_optmsk & is->is_optmsk[rev]) != is->is_opt[rev]) || + ((fin->fin_secmsk & is->is_secmsk) != is->is_sec) || + ((fin->fin_auth & is->is_authmsk) != is->is_auth)) + return NULL; + + /* + * Only one of the source or destination port can be flagged as a + * wildcard. When filling it in, fill in a copy of the matched entry + * if it has the cloning flag set. + */ + if ((fin->fin_flx & FI_IGNORE) != 0) { + fin->fin_rev = rev; + return is; + } + + if ((flags & (SI_W_SPORT|SI_W_DPORT))) { + if ((flags & SI_CLONE) != 0) { + ipstate_t *clone; + + clone = fr_stclone(fin, tcp, is); + if (clone == NULL) + return NULL; + is = clone; + } else { + ATOMIC_DECL(ips_stats.iss_wild); + } + + if ((flags & SI_W_SPORT) != 0) { + if (rev == 0) { + is->is_sport = sp; + is->is_send = ntohl(tcp->th_seq); + } else { + is->is_sport = dp; + is->is_send = ntohl(tcp->th_ack); + } + is->is_maxsend = is->is_send + 1; + } else if ((flags & SI_W_DPORT) != 0) { + if (rev == 0) { + is->is_dport = dp; + is->is_dend = ntohl(tcp->th_ack); + } else { + is->is_dport = sp; + is->is_dend = ntohl(tcp->th_seq); + } + is->is_maxdend = is->is_dend + 1; + } + is->is_flags &= ~(SI_W_SPORT|SI_W_DPORT); + if ((flags & SI_CLONED) && ipstate_logging) + ipstate_log(is, ISL_CLONE); + } + + ret = -1; + + if (is->is_flx[out][rev] == 0) { + is->is_flx[out][rev] = flx; + is->is_opt[rev] = fin->fin_optmsk; + if (is->is_v == 6) { + is->is_opt[rev] &= ~0x8; + is->is_optmsk[rev] &= ~0x8; + } + } + + /* + * Check if the interface name for this "direction" is set and if not, + * fill it in. + */ + if (is->is_ifp[idx] == NULL && + (*is->is_ifname[idx] == '\0' || *is->is_ifname[idx] == '*')) { + is->is_ifp[idx] = ifp; + COPYIFNAME(is->is_v, ifp, is->is_ifname[idx]); + } + fin->fin_rev = rev; + return is; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_checkicmpmatchingstate */ +/* Returns: Nil */ +/* Parameters: fin(I) - pointer to packet information */ +/* */ +/* If we've got an ICMP error message, using the information stored in the */ +/* ICMP packet, look for a matching state table entry. */ +/* */ +/* If we return NULL then no lock on ipf_state is held. */ +/* If we return non-null then a read-lock on ipf_state is held. */ +/* ------------------------------------------------------------------------ */ +static ipstate_t *fr_checkicmpmatchingstate(fin) +fr_info_t *fin; +{ + ipstate_t *is, **isp; + u_short sport, dport; + u_char pr; + int backward, i, oi; + i6addr_t dst, src; + struct icmp *ic; + u_short savelen; + icmphdr_t *icmp; + fr_info_t ofin; + tcphdr_t *tcp; + int type, len; + ip_t *oip; + u_int hv; + + /* + * Does it at least have the return (basic) IP header ? + * Is it an actual recognised ICMP error type? + * Only a basic IP header (no options) should be with + * an ICMP error header. + */ + if ((fin->fin_v != 4) || (fin->fin_hlen != sizeof(ip_t)) || + (fin->fin_plen < ICMPERR_MINPKTLEN) || + !(fin->fin_flx & FI_ICMPERR)) + return NULL; + ic = fin->fin_dp; + type = ic->icmp_type; + + oip = (ip_t *)((char *)ic + ICMPERR_ICMPHLEN); + /* + * Check if the at least the old IP header (with options) and + * 8 bytes of payload is present. + */ + if (fin->fin_plen < ICMPERR_MAXPKTLEN + ((IP_HL(oip) - 5) << 2)) + return NULL; + + /* + * Sanity Checks. + */ + len = fin->fin_dlen - ICMPERR_ICMPHLEN; + if ((len <= 0) || ((IP_HL(oip) << 2) > 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 guarantee it is + * all here now. + */ +#ifdef _KERNEL + { + mb_t *m; + + m = fin->fin_m; +# if defined(MENTAT) + if ((char *)oip + len > (char *)m->b_wptr) + return NULL; +# else + if ((char *)oip + len > (char *)fin->fin_ip + m->m_len) + return NULL; +# endif + } +#endif + bcopy((char *)fin, (char *)&ofin, sizeof(*fin)); + + /* + * in the IPv4 case we must zero the i6addr union otherwise + * the IP6_EQ and IP6_NEQ 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)); + + /* + * 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, like + * oip->ip_off - it is still in network byte order so fix it. + */ + savelen = oip->ip_len; + oip->ip_len = len; + oip->ip_off = ntohs(oip->ip_off); + + ofin.fin_flx = FI_NOCKSUM; + ofin.fin_v = 4; + ofin.fin_ip = oip; + ofin.fin_m = NULL; /* if dereferenced, panic XXX */ + ofin.fin_mp = NULL; /* if dereferenced, panic XXX */ + (void) fr_makefrip(IP_HL(oip) << 2, oip, &ofin); + ofin.fin_ifp = fin->fin_ifp; + ofin.fin_out = !fin->fin_out; + /* + * Reset the short and bad flag here because in fr_matchsrcdst() + * the flags for the current packet (fin_flx) are compared against + * those for the existing session. + */ + ofin.fin_flx &= ~(FI_BAD|FI_SHORT); + + /* + * Put old values of ip_len and ip_off back as we don't know + * if we have to forward the packet (or process it again. + */ + oip->ip_len = savelen; + oip->ip_off = htons(oip->ip_off); + + switch (oip->ip_p) + { + case IPPROTO_ICMP : + /* + * an 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 ((ofin.fin_flx & FI_ICMPERR) != 0) + return NULL; + + /* + * perform a lookup of the ICMP packet in the state table + */ + icmp = (icmphdr_t *)((char *)oip + (IP_HL(oip) << 2)); + 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 = DOUBLE_HASH(hv); + + READ_ENTER(&ipf_state); + for (isp = &ips_table[hv]; ((is = *isp) != NULL); ) { + isp = &is->is_hnext; + if ((is->is_p != pr) || (is->is_v != 4)) + continue; + if (is->is_pass & FR_NOICMPERR) + continue; + is = fr_matchsrcdst(&ofin, is, &src, &dst, + NULL, FI_ICMPCMP); + if (is != NULL) { + /* + * i : the index of this packet (the icmp + * unreachable) + * oi : the index of the original packet found + * in the icmp header (i.e. the packet + * causing this icmp) + * backward : original packet was backward + * compared to the state + */ + backward = IP6_NEQ(&is->is_src, &src); + fin->fin_rev = !backward; + i = (!backward << 1) + fin->fin_out; + oi = (backward << 1) + ofin.fin_out; + if (is->is_icmppkts[i] > is->is_pkts[oi]) + continue; + ips_stats.iss_hits++; + is->is_icmppkts[i]++; + return is; + } + } + RWLOCK_EXIT(&ipf_state); + return NULL; + case IPPROTO_TCP : + case IPPROTO_UDP : + break; + default : + return NULL; + } + + tcp = (tcphdr_t *)((char *)oip + (IP_HL(oip) << 2)); + 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 = DOUBLE_HASH(hv); + + READ_ENTER(&ipf_state); + for (isp = &ips_table[hv]; ((is = *isp) != NULL); ) { + 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. Only the + * ports are known to be present and can be even if the + * short flag is set. + */ + if ((is->is_p == pr) && (is->is_v == 4) && + (is = fr_matchsrcdst(&ofin, is, &src, &dst, + tcp, FI_ICMPCMP))) { + /* + * i : the index of this packet (the icmp unreachable) + * oi : the index of the original packet found in the + * icmp header (i.e. the packet causing this icmp) + * backward : original packet was backward compared to + * the state + */ + backward = IP6_NEQ(&is->is_src, &src); + fin->fin_rev = !backward; + i = (!backward << 1) + fin->fin_out; + oi = (backward << 1) + ofin.fin_out; + + if (((is->is_pass & FR_NOICMPERR) != 0) || + (is->is_icmppkts[i] > is->is_pkts[oi])) + break; + ips_stats.iss_hits++; + is->is_icmppkts[i]++; + /* + * we deliberately do not touch the timeouts + * for the accompanying state table entry. + * It remains to be seen if that is correct. XXX + */ + return is; + } + } + RWLOCK_EXIT(&ipf_state); + return NULL; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_ipsmove */ +/* Returns: Nil */ +/* Parameters: is(I) - pointer to state table entry */ +/* hv(I) - new hash value for state table entry */ +/* Write Locks: ipf_state */ +/* */ +/* Move a state entry from one position in the hash table to another. */ +/* ------------------------------------------------------------------------ */ +static void fr_ipsmove(is, hv) +ipstate_t *is; +u_int hv; +{ + ipstate_t **isp; + u_int hvm; + + hvm = is->is_hv; + /* + * Remove the hash from the old location... + */ + isp = is->is_phnext; + if (is->is_hnext) + is->is_hnext->is_phnext = isp; + *isp = is->is_hnext; + if (ips_table[hvm] == NULL) + ips_stats.iss_inuse--; + ips_stats.iss_bucketlen[hvm]--; + + /* + * ...and put the hash in the new one. + */ + hvm = DOUBLE_HASH(hv); + is->is_hv = hvm; + isp = &ips_table[hvm]; + if (*isp) + (*isp)->is_phnext = &is->is_hnext; + else + ips_stats.iss_inuse++; + ips_stats.iss_bucketlen[hvm]++; + is->is_phnext = isp; + is->is_hnext = *isp; + *isp = is; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_stlookup */ +/* Returns: ipstate_t* - NULL == no matching state found, */ +/* else pointer to state information is returned */ +/* Parameters: fin(I) - pointer to packet information */ +/* tcp(I) - pointer to TCP/UDP header. */ +/* */ +/* Search the state table for a matching entry to the packet described by */ +/* the contents of *fin. */ +/* */ +/* If we return NULL then no lock on ipf_state is held. */ +/* If we return non-null then a read-lock on ipf_state is held. */ +/* ------------------------------------------------------------------------ */ +ipstate_t *fr_stlookup(fin, tcp, ifqp) +fr_info_t *fin; +tcphdr_t *tcp; +ipftq_t **ifqp; +{ + u_int hv, hvm, pr, v, tryagain; + ipstate_t *is, **isp; + u_short dport, sport; + i6addr_t src, dst; + struct icmp *ic; + ipftq_t *ifq; + int oow; + + is = NULL; + ifq = NULL; + tcp = fin->fin_dp; + 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; + + 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 + if ((v == 4) && + (fin->fin_flx & (FI_MULTICAST|FI_BROADCAST|FI_MBCAST))) { + if (fin->fin_out == 0) { + hv -= src.in4.s_addr; + } else { + hv -= dst.in4.s_addr; + } + } + + /* + * Search the hash table for matching packet header info. + */ + switch (pr) + { +#ifdef USE_INET6 + case IPPROTO_ICMPV6 : + tryagain = 0; + if (v == 6) { + if ((ic->icmp_type == ICMP6_ECHO_REQUEST) || + (ic->icmp_type == ICMP6_ECHO_REPLY)) { + hv += ic->icmp_id; + } + } + READ_ENTER(&ipf_state); +icmp6again: + hvm = DOUBLE_HASH(hv); + for (isp = &ips_table[hvm]; ((is = *isp) != NULL); ) { + isp = &is->is_hnext; + /* + * If a connection is about to be deleted, no packets + * are allowed to match it. + */ + if (is->is_sti.tqe_ifq == &ips_deletetq) + continue; + + if ((is->is_p != pr) || (is->is_v != v)) + continue; + is = fr_matchsrcdst(fin, is, &src, &dst, NULL, FI_CMP); + if (is != NULL && + fr_matchicmpqueryreply(v, &is->is_icmp, + ic, fin->fin_rev)) { + if (fin->fin_rev) + ifq = &ips_icmpacktq; + else + ifq = &ips_icmptq; + break; + } + } + + if (is != NULL) { + if ((tryagain != 0) && !(is->is_flags & SI_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(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. + * + * XXX With some ICMP6 packets, the "other" address is already + * in the packet, after the ICMP6 header, and this could be + * used in place of the multicast address. However, taking + * advantage of this requires some significant code changes + * to handle the specific types where that is the case. + */ + if ((ips_stats.iss_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; + } + + is = fr_checkicmp6matchingstate(fin); + if (is != NULL) + return is; + break; +#endif + + case IPPROTO_ICMP : + if (v == 4) { + hv += ic->icmp_id; + } + hv = DOUBLE_HASH(hv); + READ_ENTER(&ipf_state); + for (isp = &ips_table[hv]; ((is = *isp) != NULL); ) { + isp = &is->is_hnext; + if ((is->is_p != pr) || (is->is_v != v)) + continue; + is = fr_matchsrcdst(fin, is, &src, &dst, NULL, FI_CMP); + if ((is != NULL) && + (ic->icmp_id == is->is_icmp.ici_id) && + fr_matchicmpqueryreply(v, &is->is_icmp, + ic, fin->fin_rev)) { + if (fin->fin_rev) + ifq = &ips_icmpacktq; + else + ifq = &ips_icmptq; + break; + } + } + if (is == NULL) { + RWLOCK_EXIT(&ipf_state); + } + break; + + case IPPROTO_TCP : + case IPPROTO_UDP : + ifqp = NULL; + sport = htons(fin->fin_data[0]); + hv += sport; + dport = htons(fin->fin_data[1]); + hv += dport; + oow = 0; + tryagain = 0; + READ_ENTER(&ipf_state); +retry_tcpudp: + hvm = DOUBLE_HASH(hv); + for (isp = &ips_table[hvm]; ((is = *isp) != NULL); ) { + isp = &is->is_hnext; + if ((is->is_p != pr) || (is->is_v != v)) + continue; + fin->fin_flx &= ~FI_OOW; + is = fr_matchsrcdst(fin, is, &src, &dst, tcp, FI_CMP); + if (is != NULL) { + if (pr == IPPROTO_TCP) { + if (!fr_tcpstate(fin, tcp, is)) { + oow |= fin->fin_flx & FI_OOW; + continue; + } + } + break; + } + } + if (is != NULL) { + if (tryagain && + !(is->is_flags & (SI_CLONE|SI_WILDP|SI_WILDA))) { + hv += dport; + hv += sport; + fr_ipsmove(is, hv); + MUTEX_DOWNGRADE(&ipf_state); + } + break; + } + RWLOCK_EXIT(&ipf_state); + + if (ips_stats.iss_wild) { + if (tryagain == 0) { + hv -= dport; + hv -= sport; + } else if (tryagain == 1) { + hv = fin->fin_fi.fi_p; + /* + * If we try to pretend this is a reply to a + * multicast/broadcast packet then we need to + * exclude part of the address from the hash + * calculation. + */ + if (fin->fin_out == 0) { + hv += src.in4.s_addr; + } else { + hv += dst.in4.s_addr; + } + hv += dport; + hv += sport; + } + tryagain++; + if (tryagain <= 2) { + WRITE_ENTER(&ipf_state); + goto retry_tcpudp; + } + } + fin->fin_flx |= oow; + break; + +#if 0 + case IPPROTO_GRE : + gre = fin->fin_dp; + if (GRE_REV(gre->gr_flags) == 1) { + hv += gre->gr_call; + } + /* FALLTHROUGH */ +#endif + default : + ifqp = NULL; + hvm = DOUBLE_HASH(hv); + READ_ENTER(&ipf_state); + for (isp = &ips_table[hvm]; ((is = *isp) != NULL); ) { + isp = &is->is_hnext; + if ((is->is_p != pr) || (is->is_v != v)) + continue; + is = fr_matchsrcdst(fin, is, &src, &dst, NULL, FI_CMP); + if (is != NULL) { + ifq = &ips_iptq; + break; + } + } + if (is == NULL) { + RWLOCK_EXIT(&ipf_state); + } + break; + } + + if (is != NULL) { + if (((is->is_sti.tqe_flags & TQE_RULEBASED) != 0) && + (is->is_tqehead[fin->fin_rev] != NULL)) + ifq = is->is_tqehead[fin->fin_rev]; + if (ifq != NULL && ifqp != NULL) + *ifqp = ifq; + } + return is; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_updatestate */ +/* Returns: Nil */ +/* Parameters: fin(I) - pointer to packet information */ +/* is(I) - pointer to state table entry */ +/* Read Locks: ipf_state */ +/* */ +/* Updates packet and byte counters for a newly received packet. Seeds the */ +/* fragment cache with a new entry as required. */ +/* ------------------------------------------------------------------------ */ +void fr_updatestate(fin, is, ifq) +fr_info_t *fin; +ipstate_t *is; +ipftq_t *ifq; +{ + ipftqent_t *tqe; + int i, pass; + + i = (fin->fin_rev << 1) + fin->fin_out; + + /* + * For TCP packets, ifq == NULL. For all others, check if this new + * queue is different to the last one it was on and move it if so. + */ + tqe = &is->is_sti; + MUTEX_ENTER(&is->is_lock); + if ((tqe->tqe_flags & TQE_RULEBASED) != 0) + ifq = is->is_tqehead[fin->fin_rev]; + + if (ifq != NULL) + fr_movequeue(tqe, tqe->tqe_ifq, ifq); + + is->is_pkts[i]++; + is->is_bytes[i] += fin->fin_plen; + MUTEX_EXIT(&is->is_lock); + +#ifdef IPFILTER_SYNC + if (is->is_flags & IS_STATESYNC) + ipfsync_update(SMC_STATE, fin, is->is_sync); +#endif + + ATOMIC_INCL(ips_stats.iss_hits); + + fin->fin_fr = is->is_rule; + + /* + * If this packet is a fragment and the rule says to track fragments, + * then create a new fragment cache entry. + */ + pass = is->is_pass; + if ((fin->fin_flx & FI_FRAG) && FR_ISPASS(pass)) + (void) fr_newfrag(fin, pass ^ FR_KEEPSTATE); +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_checkstate */ +/* Returns: frentry_t* - NULL == search failed, */ +/* else pointer to rule for matching state */ +/* Parameters: ifp(I) - pointer to interface */ +/* passp(I) - pointer to filtering result flags */ +/* */ +/* Check if a packet is associated with an entry in the state table. */ +/* ------------------------------------------------------------------------ */ +frentry_t *fr_checkstate(fin, passp) +fr_info_t *fin; +u_32_t *passp; +{ + ipstate_t *is; + frentry_t *fr; + tcphdr_t *tcp; + ipftq_t *ifq; + u_int pass; + + if (fr_state_lock || (ips_list == NULL) || + (fin->fin_flx & (FI_SHORT|FI_STATE|FI_FRAGBODY|FI_BAD))) + return NULL; + + is = NULL; + if ((fin->fin_flx & FI_TCPUDP) || + (fin->fin_fi.fi_p == IPPROTO_ICMP) +#ifdef USE_INET6 + || (fin->fin_fi.fi_p == IPPROTO_ICMPV6) +#endif + ) + tcp = fin->fin_dp; + else + tcp = NULL; + + /* + * Search the hash table for matching packet header info. + */ + ifq = NULL; + is = fin->fin_state; + if (is == NULL) + is = fr_stlookup(fin, tcp, &ifq); + switch (fin->fin_p) + { +#ifdef USE_INET6 + case IPPROTO_ICMPV6 : + if (is != NULL) + break; + if (fin->fin_v == 6) { + is = fr_checkicmp6matchingstate(fin); + if (is != NULL) + goto matched; + } + break; +#endif + case IPPROTO_ICMP : + if (is != NULL) + break; + /* + * No matching icmp state entry. Perhaps this is a + * response to another state entry. + */ + is = fr_checkicmpmatchingstate(fin); + if (is != NULL) + goto matched; + break; + case IPPROTO_TCP : + if (is == NULL) + break; + + if (is->is_pass & FR_NEWISN) { + if (fin->fin_out == 0) + fr_fixinisn(fin, is); + else if (fin->fin_out == 1) + fr_fixoutisn(fin, is); + } + break; + default : + if (fin->fin_rev) + ifq = &ips_udpacktq; + else + ifq = &ips_udptq; + break; + } + if (is == NULL) { + ATOMIC_INCL(ips_stats.iss_miss); + return NULL; + } + +matched: + fr = is->is_rule; + if (fr != NULL) { + if ((fin->fin_out == 0) && (fr->fr_nattag.ipt_num[0] != 0)) { + if (fin->fin_nattag == NULL) + return NULL; + if (fr_matchtag(&fr->fr_nattag, fin->fin_nattag) != 0) + return NULL; + } + (void) strncpy(fin->fin_group, fr->fr_group, FR_GROUPLEN); + fin->fin_icode = fr->fr_icode; + } + + fin->fin_rule = is->is_rulen; + pass = is->is_pass; + fr_updatestate(fin, is, ifq); + + fin->fin_state = is; + is->is_touched = fr_ticks; + MUTEX_ENTER(&is->is_lock); + is->is_ref++; + MUTEX_EXIT(&is->is_lock); + RWLOCK_EXIT(&ipf_state); + fin->fin_flx |= FI_STATE; + if ((pass & FR_LOGFIRST) != 0) + pass &= ~(FR_LOGFIRST|FR_LOG); + *passp = pass; + return fr; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_fixoutisn */ +/* Returns: Nil */ +/* Parameters: fin(I) - pointer to packet information */ +/* is(I) - pointer to master state structure */ +/* */ +/* Called only for outbound packets, adjusts the sequence number and the */ +/* TCP checksum to match that change. */ +/* ------------------------------------------------------------------------ */ +static void fr_fixoutisn(fin, is) +fr_info_t *fin; +ipstate_t *is; +{ + tcphdr_t *tcp; + int rev; + u_32_t seq; + + tcp = fin->fin_dp; + rev = fin->fin_rev; + if ((is->is_flags & IS_ISNSYN) != 0) { + if (rev == 0) { + seq = ntohl(tcp->th_seq); + seq += is->is_isninc[0]; + tcp->th_seq = htonl(seq); + fix_outcksum(fin, &tcp->th_sum, is->is_sumd[0]); + } + } + if ((is->is_flags & IS_ISNACK) != 0) { + if (rev == 1) { + seq = ntohl(tcp->th_seq); + seq += is->is_isninc[1]; + tcp->th_seq = htonl(seq); + fix_outcksum(fin, &tcp->th_sum, is->is_sumd[1]); + } + } +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_fixinisn */ +/* Returns: Nil */ +/* Parameters: fin(I) - pointer to packet information */ +/* is(I) - pointer to master state structure */ +/* */ +/* Called only for inbound packets, adjusts the acknowledge number and the */ +/* TCP checksum to match that change. */ +/* ------------------------------------------------------------------------ */ +static void fr_fixinisn(fin, is) +fr_info_t *fin; +ipstate_t *is; +{ + tcphdr_t *tcp; + int rev; + u_32_t ack; + + tcp = fin->fin_dp; + rev = fin->fin_rev; + if ((is->is_flags & IS_ISNSYN) != 0) { + if (rev == 1) { + ack = ntohl(tcp->th_ack); + ack -= is->is_isninc[0]; + tcp->th_ack = htonl(ack); + fix_incksum(fin, &tcp->th_sum, is->is_sumd[0]); + } + } + if ((is->is_flags & IS_ISNACK) != 0) { + if (rev == 0) { + ack = ntohl(tcp->th_ack); + ack -= is->is_isninc[1]; + tcp->th_ack = htonl(ack); + fix_incksum(fin, &tcp->th_sum, is->is_sumd[1]); + } + } +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_statesync */ +/* Returns: Nil */ +/* Parameters: ifp(I) - pointer to interface */ +/* */ +/* Walk through all state entries and if an interface pointer match is */ +/* found then look it up again, based on its name in case the pointer has */ +/* changed since last time. */ +/* */ +/* If ifp is passed in as being non-null then we are only doing updates for */ +/* existing, matching, uses of it. */ +/* ------------------------------------------------------------------------ */ +void fr_statesync(ifp) +void *ifp; +{ + ipstate_t *is; + int i; + + if (fr_running <= 0) + return; + + WRITE_ENTER(&ipf_state); + + if (fr_running <= 0) { + RWLOCK_EXIT(&ipf_state); + return; + } + + for (is = ips_list; is; is = is->is_next) { + /* + * Look up all the interface names in the state entry. + */ + for (i = 0; i < 4; i++) { + if (ifp == NULL || ifp == is->is_ifp[i]) + is->is_ifp[i] = fr_resolvenic(is->is_ifname[i], + is->is_v); + } + } + RWLOCK_EXIT(&ipf_state); +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_delstate */ +/* Returns: int - 0 = entry deleted, else reference count on struct */ +/* Parameters: is(I) - pointer to state structure to delete */ +/* why(I) - if not 0, log reason why it was deleted */ +/* Write Locks: ipf_state */ +/* */ +/* Deletes a state entry from the enumerated list as well as the hash table */ +/* and timeout queue lists. Make adjustments to hash table statistics and */ +/* global counters as required. */ +/* ------------------------------------------------------------------------ */ +static int fr_delstate(is, why) +ipstate_t *is; +int why; +{ + + /* + * Since we want to delete this, remove it from the state table, + * where it can be found & used, first. + */ + if (is->is_phnext != NULL) { + *is->is_phnext = is->is_hnext; + if (is->is_hnext != NULL) + is->is_hnext->is_phnext = is->is_phnext; + if (ips_table[is->is_hv] == NULL) + ips_stats.iss_inuse--; + ips_stats.iss_bucketlen[is->is_hv]--; + + is->is_phnext = NULL; + is->is_hnext = NULL; + } + + /* + * Because ips_stats.iss_wild is a count of entries in the state + * table that have wildcard flags set, only decerement it once + * and do it here. + */ + if (is->is_flags & (SI_WILDP|SI_WILDA)) { + if (!(is->is_flags & SI_CLONED)) { + ATOMIC_DECL(ips_stats.iss_wild); + } + is->is_flags &= ~(SI_WILDP|SI_WILDA); + } + + /* + * Next, remove it from the timeout queue it is in. + */ + if (is->is_sti.tqe_ifq != NULL) + fr_deletequeueentry(&is->is_sti); + + if (is->is_me != NULL) { + *is->is_me = NULL; + is->is_me = NULL; + } + + /* + * If it is still in use by something else, do not go any further, + * but note that at this point it is now an orphan. How can this + * be? fr_state_flush() calls fr_delete() directly because it wants + * to empty the table out and if something has a hold on a state + * entry (such as ipfstat), it'll do the deref path that'll bring + * us back here to do the real delete & free. + */ + MUTEX_ENTER(&is->is_lock); + if (is->is_ref > 1) { + is->is_ref--; + MUTEX_EXIT(&is->is_lock); + return is->is_ref; + } + MUTEX_EXIT(&is->is_lock); + + is->is_ref = 0; + + if (is->is_tqehead[0] != NULL) { + if (fr_deletetimeoutqueue(is->is_tqehead[0]) == 0) + fr_freetimeoutqueue(is->is_tqehead[0]); + } + if (is->is_tqehead[1] != NULL) { + if (fr_deletetimeoutqueue(is->is_tqehead[1]) == 0) + fr_freetimeoutqueue(is->is_tqehead[1]); + } + +#ifdef IPFILTER_SYNC + if (is->is_sync) + ipfsync_del(is->is_sync); +#endif +#ifdef IPFILTER_SCAN + (void) ipsc_detachis(is); +#endif + + /* + * Now remove it from the linked list of known states + */ + if (is->is_pnext != NULL) { + *is->is_pnext = is->is_next; + + if (is->is_next != NULL) + is->is_next->is_pnext = is->is_pnext; + + is->is_pnext = NULL; + is->is_next = NULL; + } + + if (ipstate_logging != 0 && why != 0) + ipstate_log(is, why); + + if (is->is_p == IPPROTO_TCP) + ips_stats.iss_fin++; + else + ips_stats.iss_expire++; + + if (is->is_rule != NULL) { + is->is_rule->fr_statecnt--; + (void) fr_derefrule(&is->is_rule); + } + +#if defined(NEED_LOCAL_RAND) && defined(_KERNEL) + ipf_rand_push(is, sizeof(*is)); +#endif + + MUTEX_DESTROY(&is->is_lock); + KFREE(is); + ips_num--; + + return 0; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_timeoutstate */ +/* Returns: Nil */ +/* Parameters: Nil */ +/* */ +/* Slowly expire held state for thingslike UDP and ICMP. The algorithm */ +/* used here is to keep the queue sorted with the oldest things at the top */ +/* and the youngest at the bottom. So if the top one doesn't need to be */ +/* expired then neither will any under it. */ +/* ------------------------------------------------------------------------ */ +void fr_timeoutstate() +{ + ipftq_t *ifq, *ifqnext; + ipftqent_t *tqe, *tqn; + ipstate_t *is; + SPL_INT(s); + + SPL_NET(s); + WRITE_ENTER(&ipf_state); + for (ifq = ips_tqtqb; ifq != NULL; ifq = ifq->ifq_next) + for (tqn = ifq->ifq_head; ((tqe = tqn) != NULL); ) { + if (tqe->tqe_die > fr_ticks) + break; + tqn = tqe->tqe_next; + is = tqe->tqe_parent; + fr_delstate(is, ISL_EXPIRE); + } + + for (ifq = ips_utqe; ifq != NULL; ifq = ifqnext) { + ifqnext = ifq->ifq_next; + + for (tqn = ifq->ifq_head; ((tqe = tqn) != NULL); ) { + if (tqe->tqe_die > fr_ticks) + break; + tqn = tqe->tqe_next; + is = tqe->tqe_parent; + fr_delstate(is, ISL_EXPIRE); + } + } + + for (ifq = ips_utqe; ifq != NULL; ifq = ifqnext) { + ifqnext = ifq->ifq_next; + + if (((ifq->ifq_flags & IFQF_DELETE) != 0) && + (ifq->ifq_ref == 0)) { + fr_freetimeoutqueue(ifq); + } + } + + if (fr_state_doflush) { + (void) fr_state_flush(2, 0); + fr_state_doflush = 0; + } + + RWLOCK_EXIT(&ipf_state); + SPL_X(s); +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_state_flush */ +/* Returns: int - 0 == success, -1 == failure */ +/* Parameters: Nil */ +/* Write Locks: ipf_state */ +/* */ +/* Flush state tables. Three 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. If this fails to free enough */ +/* slots then work backwards in half hour slots to 30 minutes. */ +/* If that too fails, then work backwards in 30 second intervals */ +/* for the last 30 minutes to at worst 30 seconds idle. */ +/* ------------------------------------------------------------------------ */ +static int fr_state_flush(which, proto) +int which, proto; +{ + ipftq_t *ifq, *ifqnext; + ipftqent_t *tqe, *tqn; + ipstate_t *is, **isp; + int removed; + SPL_INT(s); + + removed = 0; + + SPL_NET(s); + + switch (which) + { + case 0 : + /* + * Style 0 flush removes everything... + */ + for (isp = &ips_list; ((is = *isp) != NULL); ) { + if ((proto != 0) && (is->is_v != proto)) { + isp = &is->is_next; + continue; + } + if (fr_delstate(is, ISL_FLUSH) == 0) + removed++; + else + isp = &is->is_next; + } + break; + + case 1 : + /* + * Since we're only interested in things that are closing, + * we can start with the appropriate timeout queue. + */ + for (ifq = ips_tqtqb + IPF_TCPS_CLOSE_WAIT; ifq != NULL; + ifq = ifq->ifq_next) { + + for (tqn = ifq->ifq_head; ((tqe = tqn) != NULL); ) { + tqn = tqe->tqe_next; + is = tqe->tqe_parent; + if (is->is_p != IPPROTO_TCP) + break; + if (fr_delstate(is, ISL_EXPIRE) == 0) + removed++; + } + } + + /* + * Also need to look through the user defined queues. + */ + for (ifq = ips_utqe; ifq != NULL; ifq = ifqnext) { + ifqnext = ifq->ifq_next; + for (tqn = ifq->ifq_head; ((tqe = tqn) != NULL); ) { + tqn = tqe->tqe_next; + is = tqe->tqe_parent; + if (is->is_p != IPPROTO_TCP) + continue; + + if ((is->is_state[0] > IPF_TCPS_ESTABLISHED) && + (is->is_state[1] > IPF_TCPS_ESTABLISHED)) { + if (fr_delstate(is, ISL_EXPIRE) == 0) + removed++; + } + } + } + break; + + case 2 : + break; + + /* + * Args 5-11 correspond to flushing those particular states + * for TCP connections. + */ + case IPF_TCPS_CLOSE_WAIT : + case IPF_TCPS_FIN_WAIT_1 : + case IPF_TCPS_CLOSING : + case IPF_TCPS_LAST_ACK : + case IPF_TCPS_FIN_WAIT_2 : + case IPF_TCPS_TIME_WAIT : + case IPF_TCPS_CLOSED : + tqn = ips_tqtqb[which].ifq_head; + while (tqn != NULL) { + tqe = tqn; + tqn = tqe->tqe_next; + is = tqe->tqe_parent; + if (fr_delstate(is, ISL_FLUSH) == 0) + removed++; + } + break; + + default : + if (which < 30) + break; + + /* + * Take a large arbitrary number to mean the number of seconds + * for which which consider to be the maximum value we'll allow + * the expiration to be. + */ + which = IPF_TTLVAL(which); + for (isp = &ips_list; ((is = *isp) != NULL); ) { + if ((proto == 0) || (is->is_v == proto)) { + if (fr_ticks - is->is_touched > which) { + if (fr_delstate(is, ISL_FLUSH) == 0) { + removed++; + continue; + } + } + } + isp = &is->is_next; + } + break; + } + + if (which != 2) { + SPL_X(s); + return removed; + } + + /* + * Asked to remove inactive entries because the table is full. + */ + if (fr_ticks - ips_last_force_flush > IPF_TTLVAL(5)) { + ips_last_force_flush = fr_ticks; + removed = ipf_queueflush(fr_state_flush_entry, ips_tqtqb, + ips_utqe); + } + + SPL_X(s); + return removed; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_state_flush_entry */ +/* Returns: int - 0 = entry deleted, else not deleted */ +/* Parameters: entry(I) - pointer to state structure to delete */ +/* Write Locks: ipf_state */ +/* */ +/* This function is a stepping stone between ipf_queueflush() and */ +/* fr_delstate(). It is used so we can provide a uniform interface via the */ +/* ipf_queueflush() function. */ +/* ------------------------------------------------------------------------ */ +static int fr_state_flush_entry(entry) +void *entry; +{ + return fr_delstate(entry, ISL_FLUSH); +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_tcp_age */ +/* Returns: int - 1 == state transition made, 0 == no change (rejected) */ +/* Parameters: tq(I) - pointer to timeout queue information */ +/* fin(I) - pointer to packet information */ +/* tqtab(I) - TCP timeout queue table this is in */ +/* flags(I) - flags from state/NAT entry */ +/* */ +/* Rewritten by Arjan de Vet <Arjan.deVet@adv.iae.nl>, 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. */ +/* */ +/* Internal 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 */ +/* */ +/* A typical procession for a connection is as follows: */ +/* */ +/* +--------------+-------------------+ */ +/* | Side '0' | Side '1' | */ +/* +--------------+-------------------+ */ +/* | 0 -> 1 (SYN) | | */ +/* | | 0 -> 2 (SYN-ACK) | */ +/* | 1 -> 3 (ACK) | | */ +/* | | 2 -> 4 (ACK-PUSH) | */ +/* | 3 -> 4 (ACK) | | */ +/* | ... | ... | */ +/* | | 4 -> 6 (FIN-ACK) | */ +/* | 4 -> 5 (ACK) | | */ +/* | | 6 -> 6 (ACK-PUSH) | */ +/* | 5 -> 5 (ACK) | | */ +/* | 5 -> 8 (FIN) | | */ +/* | | 6 -> 10 (ACK) | */ +/* +--------------+-------------------+ */ +/* */ +/* Locking: it is assumed that the parent of the tqe structure is locked. */ +/* ------------------------------------------------------------------------ */ +int fr_tcp_age(tqe, fin, tqtab, flags) +ipftqent_t *tqe; +fr_info_t *fin; +ipftq_t *tqtab; +int flags; +{ + int dlen, ostate, nstate, rval, dir; + u_char tcpflags; + tcphdr_t *tcp; + + tcp = fin->fin_dp; + + rval = 0; + dir = fin->fin_rev; + tcpflags = tcp->th_flags; + dlen = fin->fin_dlen - (TCP_OFF(tcp) << 2); + + if (tcpflags & TH_RST) { + if (!(tcpflags & TH_PUSH) && !dlen) + nstate = IPF_TCPS_CLOSED; + else + nstate = IPF_TCPS_CLOSE_WAIT; + rval = 1; + } else { + ostate = tqe->tqe_state[1 - dir]; + nstate = tqe->tqe_state[dir]; + + switch (nstate) + { + case IPF_TCPS_LISTEN: /* 0 */ + if ((tcpflags & TH_OPENING) == TH_OPENING) { + /* + * 'dir' received an S and sends SA in + * response, LISTEN -> SYN_RECEIVED + */ + nstate = IPF_TCPS_SYN_RECEIVED; + rval = 1; + } else if ((tcpflags & TH_OPENING) == TH_SYN) { + /* 'dir' sent S, LISTEN -> SYN_SENT */ + nstate = IPF_TCPS_SYN_SENT; + rval = 1; + } + /* + * 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 (((flags & IS_TCPFSM) == 0) && + ((tcpflags & TH_ACKMASK) == TH_ACK)) { + /* + * we saw an A, guess 'dir' is in ESTABLISHED + * mode + */ + switch (ostate) + { + case IPF_TCPS_LISTEN : + case IPF_TCPS_SYN_RECEIVED : + nstate = IPF_TCPS_HALF_ESTAB; + rval = 1; + break; + case IPF_TCPS_HALF_ESTAB : + case IPF_TCPS_ESTABLISHED : + nstate = IPF_TCPS_ESTABLISHED; + rval = 1; + break; + default : + break; + } + } + /* + * 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 IPF_TCPS_SYN_SENT: /* 1 */ + if ((tcpflags & ~(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 rval so as to indicate the packet has + * passed the check for its flags being valid + * in the TCP FSM. Setting rval to 2 has the + * result of not resetting the timeout. + */ + rval = 2; + } else if ((tcpflags & (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 + */ + nstate = IPF_TCPS_ESTABLISHED; + rval = 1; + } else if (tcpflags & 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 + */ + nstate = IPF_TCPS_FIN_WAIT_1; + rval = 1; + } else if ((tcpflags & 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 + */ + nstate = IPF_TCPS_SYN_RECEIVED; + rval = 1; + } + break; + + case IPF_TCPS_SYN_RECEIVED: /* 2 */ + if ((tcpflags & (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 + */ + nstate = IPF_TCPS_ESTABLISHED; + rval = 1; + } else if ((tcpflags & ~(TH_ECN|TH_CWR)) == + TH_OPENING) { + /* + * We see an SA from 'dir' which is already in + * SYN_RECEIVED state. + */ + rval = 2; + } else if (tcpflags & 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 + */ + nstate = IPF_TCPS_FIN_WAIT_1; + rval = 1; + } + break; + + case IPF_TCPS_HALF_ESTAB: /* 3 */ + if (tcpflags & TH_FIN) { + nstate = IPF_TCPS_FIN_WAIT_1; + rval = 1; + } else if ((tcpflags & TH_ACKMASK) == TH_ACK) { + /* + * If we've picked up a connection in mid + * flight, we could be looking at a follow on + * packet from the same direction as the one + * that created this state. Recognise it but + * do not advance the entire connection's + * state. + */ + switch (ostate) + { + case IPF_TCPS_LISTEN : + case IPF_TCPS_SYN_SENT : + case IPF_TCPS_SYN_RECEIVED : + rval = 1; + break; + case IPF_TCPS_HALF_ESTAB : + case IPF_TCPS_ESTABLISHED : + nstate = IPF_TCPS_ESTABLISHED; + rval = 1; + break; + default : + break; + } + } + break; + + case IPF_TCPS_ESTABLISHED: /* 4 */ + rval = 1; + if (tcpflags & TH_FIN) { + /* + * 'dir' closed its side of the connection; + * this gives us a half-closed connection; + * ESTABLISHED -> FIN_WAIT_1 + */ + if (ostate == IPF_TCPS_FIN_WAIT_1) { + nstate = IPF_TCPS_CLOSING; + } else { + nstate = IPF_TCPS_FIN_WAIT_1; + } + } else if (tcpflags & TH_ACK) { + /* + * an ACK, should we exclude other flags here? + */ + if (ostate == IPF_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 + */ + nstate = IPF_TCPS_CLOSE_WAIT; + } else if (ostate < IPF_TCPS_CLOSE_WAIT) { + /* + * still a fully established + * connection reset timeout + */ + nstate = IPF_TCPS_ESTABLISHED; + } + } + break; + + case IPF_TCPS_CLOSE_WAIT: /* 5 */ + rval = 1; + if (tcpflags & TH_FIN) { + /* + * application closed and 'dir' sent a FIN, + * we're now going into LAST_ACK state + */ + nstate = IPF_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 + */ + nstate = IPF_TCPS_CLOSE_WAIT; + } + break; + + case IPF_TCPS_FIN_WAIT_1: /* 6 */ + rval = 1; + if ((tcpflags & TH_ACK) && + ostate > IPF_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? + */ + nstate = IPF_TCPS_TIME_WAIT; + } 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 + */ + nstate = IPF_TCPS_FIN_WAIT_1; + } + break; + + case IPF_TCPS_CLOSING: /* 7 */ + if ((tcpflags & (TH_FIN|TH_ACK)) == TH_ACK) { + nstate = IPF_TCPS_TIME_WAIT; + } + rval = 2; + break; + + case IPF_TCPS_LAST_ACK: /* 8 */ + if (tcpflags & TH_ACK) { + if ((tcpflags & TH_PUSH) || dlen) + /* + * there is still data to be delivered, + * reset timeout + */ + rval = 1; + else + rval = 2; + } + /* + * 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 IPF_TCPS_FIN_WAIT_2: /* 9 */ + /* NOT USED */ + break; + + case IPF_TCPS_TIME_WAIT: /* 10 */ + /* we're in 2MSL timeout now */ + if (ostate == IPF_TCPS_LAST_ACK) { + nstate = IPF_TCPS_CLOSED; + } + rval = 1; + break; + + case IPF_TCPS_CLOSED: /* 11 */ + rval = 2; + break; + + default : +#if defined(_KERNEL) +# if SOLARIS + cmn_err(CE_NOTE, + "tcp %lx flags %x si %lx nstate %d ostate %d\n", + (u_long)tcp, tcpflags, (u_long)tqe, + nstate, ostate); +# else + printf("tcp %lx flags %x si %lx nstate %d ostate %d\n", + (u_long)tcp, tcpflags, (u_long)tqe, + nstate, ostate); +# endif +#else + abort(); +#endif + break; + } + } + + /* + * If rval == 2 then do not update the queue position, but treat the + * packet as being ok. + */ + if (rval == 2) + rval = 1; + else if (rval == 1) { + tqe->tqe_state[dir] = nstate; + if ((tqe->tqe_flags & TQE_RULEBASED) == 0) + fr_movequeue(tqe, tqe->tqe_ifq, tqtab + nstate); + } + + return rval; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: ipstate_log */ +/* Returns: Nil */ +/* Parameters: is(I) - pointer to state structure */ +/* type(I) - type of log entry to create */ +/* */ +/* Creates a state table log entry using the state structure and type info. */ +/* passed in. Log packet/byte counts, source/destination address and other */ +/* protocol specific information. */ +/* ------------------------------------------------------------------------ */ +void ipstate_log(is, type) +struct ipstate *is; +u_int type; +{ +#ifdef IPFILTER_LOG + struct ipslog ipsl; + size_t sizes[1]; + void *items[1]; + int types[1]; + + /* + * Copy information out of the ipstate_t structure and into the + * structure used for logging. + */ + ipsl.isl_type = type; + ipsl.isl_pkts[0] = is->is_pkts[0] + is->is_icmppkts[0]; + ipsl.isl_bytes[0] = is->is_bytes[0]; + ipsl.isl_pkts[1] = is->is_pkts[1] + is->is_icmppkts[1]; + ipsl.isl_bytes[1] = is->is_bytes[1]; + ipsl.isl_pkts[2] = is->is_pkts[2] + is->is_icmppkts[2]; + ipsl.isl_bytes[2] = is->is_bytes[2]; + ipsl.isl_pkts[3] = is->is_pkts[3] + is->is_icmppkts[3]; + ipsl.isl_bytes[3] = is->is_bytes[3]; + 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; + ipsl.isl_tag = is->is_tag; + ipsl.isl_rulen = is->is_rulen; + (void) strncpy(ipsl.isl_group, is->is_group, FR_GROUPLEN); + + 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.ici_type; + } else if (ipsl.isl_p == IPPROTO_ICMPV6) { + ipsl.isl_itype = is->is_icmp.ici_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; + + if (ipllog(IPL_LOGSTATE, NULL, items, sizes, types, 1)) { + ATOMIC_INCL(ips_stats.iss_logged); + } else { + ATOMIC_INCL(ips_stats.iss_logfail); + } +#endif +} + + +#ifdef USE_INET6 +/* ------------------------------------------------------------------------ */ +/* Function: fr_checkicmp6matchingstate */ +/* Returns: ipstate_t* - NULL == no match found, */ +/* else pointer to matching state entry */ +/* Parameters: fin(I) - pointer to packet information */ +/* Locks: NULL == no locks, else Read Lock on ipf_state */ +/* */ +/* If we've got an ICMPv6 error message, using the information stored in */ +/* the ICMPv6 packet, look for a matching state table entry. */ +/* ------------------------------------------------------------------------ */ +static ipstate_t *fr_checkicmp6matchingstate(fin) +fr_info_t *fin; +{ + struct icmp6_hdr *ic6, *oic; + int type, backward, i; + ipstate_t *is, **isp; + u_short sport, dport; + i6addr_t dst, src; + u_short savelen; + icmpinfo_t *ic; + fr_info_t ofin; + tcphdr_t *tcp; + ip6_t *oip6; + u_char pr; + u_int hv; + + /* + * Does it at least have the return (basic) IP header ? + * Is it an actual recognised ICMP error type? + * Only a basic IP header (no options) should be with + * an ICMP error header. + */ + if ((fin->fin_v != 6) || (fin->fin_plen < ICMP6ERR_MINPKTLEN) || + !(fin->fin_flx & FI_ICMPERR)) + return NULL; + + ic6 = fin->fin_dp; + type = ic6->icmp6_type; + + oip6 = (ip6_t *)((char *)ic6 + ICMPERR_ICMPHLEN); + if (fin->fin_plen < sizeof(*oip6)) + return NULL; + + bcopy((char *)fin, (char *)&ofin, sizeof(*fin)); + ofin.fin_v = 6; + ofin.fin_ifp = fin->fin_ifp; + ofin.fin_out = !fin->fin_out; + ofin.fin_m = NULL; /* if dereferenced, panic XXX */ + ofin.fin_mp = NULL; /* if dereferenced, panic XXX */ + + /* + * We make a fin entry to be able to feed it to + * matchsrcdst. Note that not all fields are necessary + * 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 oip6 in network + * order. Any change we make must be undone afterwards. + */ + savelen = oip6->ip6_plen; + oip6->ip6_plen = fin->fin_dlen - ICMPERR_ICMPHLEN; + ofin.fin_flx = FI_NOCKSUM; + ofin.fin_ip = (ip_t *)oip6; + (void) fr_makefrip(sizeof(*oip6), (ip_t *)oip6, &ofin); + ofin.fin_flx &= ~(FI_BAD|FI_SHORT); + oip6->ip6_plen = savelen; + + if (oip6->ip6_nxt == IPPROTO_ICMPV6) { + oic = (struct icmp6_hdr *)(oip6 + 1); + /* + * an 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 = oip6->ip6_nxt); + src.in6 = oip6->ip6_src; + hv += src.in4.s_addr; + dst.in6 = oip6->ip6_dst; + hv += dst.in4.s_addr; + hv += oic->icmp6_id; + hv += oic->icmp6_seq; + hv = DOUBLE_HASH(hv); + + READ_ENTER(&ipf_state); + for (isp = &ips_table[hv]; ((is = *isp) != NULL); ) { + ic = &is->is_icmp; + isp = &is->is_hnext; + if ((is->is_p == pr) && + !(is->is_pass & FR_NOICMPERR) && + (oic->icmp6_id == ic->ici_id) && + (oic->icmp6_seq == ic->ici_seq) && + (is = fr_matchsrcdst(&ofin, is, &src, + &dst, NULL, FI_ICMPCMP))) { + /* + * in the state table ICMP query's are stored + * with the type of the corresponding ICMP + * response. Correct here + */ + if (((ic->ici_type == ICMP6_ECHO_REPLY) && + (oic->icmp6_type == ICMP6_ECHO_REQUEST)) || + (ic->ici_type - 1 == oic->icmp6_type )) { + ips_stats.iss_hits++; + backward = IP6_NEQ(&is->is_dst, &src); + fin->fin_rev = !backward; + i = (backward << 1) + fin->fin_out; + is->is_icmppkts[i]++; + return is; + } + } + } + RWLOCK_EXIT(&ipf_state); + return NULL; + } + + hv = (pr = oip6->ip6_nxt); + src.in6 = oip6->ip6_src; + hv += src.i6[0]; + hv += src.i6[1]; + hv += src.i6[2]; + hv += src.i6[3]; + dst.in6 = oip6->ip6_dst; + hv += dst.i6[0]; + hv += dst.i6[1]; + hv += dst.i6[2]; + hv += dst.i6[3]; + + if ((oip6->ip6_nxt == IPPROTO_TCP) || (oip6->ip6_nxt == IPPROTO_UDP)) { + tcp = (tcphdr_t *)(oip6 + 1); + dport = tcp->th_dport; + sport = tcp->th_sport; + hv += dport; + hv += sport; + } else + tcp = NULL; + hv = DOUBLE_HASH(hv); + + READ_ENTER(&ipf_state); + for (isp = &ips_table[hv]; ((is = *isp) != NULL); ) { + 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) || + (is->is_pass & FR_NOICMPERR)) + continue; + is = fr_matchsrcdst(&ofin, is, &src, &dst, tcp, FI_ICMPCMP); + if (is != NULL) { + ips_stats.iss_hits++; + backward = IP6_NEQ(&is->is_dst, &src); + fin->fin_rev = !backward; + i = (backward << 1) + fin->fin_out; + is->is_icmppkts[i]++; + /* + * we deliberately do not touch the timeouts + * for the accompanying state table entry. + * It remains to be seen if that is correct. XXX + */ + return is; + } + } + RWLOCK_EXIT(&ipf_state); + return NULL; +} +#endif + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_sttab_init */ +/* Returns: Nil */ +/* Parameters: tqp(I) - pointer to an array of timeout queues for TCP */ +/* */ +/* Initialise the array of timeout queues for TCP. */ +/* ------------------------------------------------------------------------ */ +void fr_sttab_init(tqp) +ipftq_t *tqp; +{ + int i; + + for (i = IPF_TCP_NSTATES - 1; i >= 0; i--) { + tqp[i].ifq_ttl = 0; + tqp[i].ifq_ref = 1; + tqp[i].ifq_head = NULL; + tqp[i].ifq_tail = &tqp[i].ifq_head; + tqp[i].ifq_next = tqp + i + 1; + MUTEX_INIT(&tqp[i].ifq_lock, "ipftq tcp tab"); + } + tqp[IPF_TCP_NSTATES - 1].ifq_next = NULL; + tqp[IPF_TCPS_CLOSED].ifq_ttl = fr_tcpclosed; + tqp[IPF_TCPS_LISTEN].ifq_ttl = fr_tcptimeout; + tqp[IPF_TCPS_SYN_SENT].ifq_ttl = fr_tcptimeout; + tqp[IPF_TCPS_SYN_RECEIVED].ifq_ttl = fr_tcptimeout; + tqp[IPF_TCPS_ESTABLISHED].ifq_ttl = fr_tcpidletimeout; + tqp[IPF_TCPS_CLOSE_WAIT].ifq_ttl = fr_tcphalfclosed; + tqp[IPF_TCPS_FIN_WAIT_1].ifq_ttl = fr_tcphalfclosed; + tqp[IPF_TCPS_CLOSING].ifq_ttl = fr_tcptimeout; + tqp[IPF_TCPS_LAST_ACK].ifq_ttl = fr_tcplastack; + tqp[IPF_TCPS_FIN_WAIT_2].ifq_ttl = fr_tcpclosewait; + tqp[IPF_TCPS_TIME_WAIT].ifq_ttl = fr_tcptimewait; + tqp[IPF_TCPS_HALF_ESTAB].ifq_ttl = fr_tcptimeout; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_sttab_destroy */ +/* Returns: Nil */ +/* Parameters: tqp(I) - pointer to an array of timeout queues for TCP */ +/* */ +/* Do whatever is necessary to "destroy" each of the entries in the array */ +/* of timeout queues for TCP. */ +/* ------------------------------------------------------------------------ */ +void fr_sttab_destroy(tqp) +ipftq_t *tqp; +{ + int i; + + for (i = IPF_TCP_NSTATES - 1; i >= 0; i--) + MUTEX_DESTROY(&tqp[i].ifq_lock); +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_statederef */ +/* Returns: Nil */ +/* Parameters: isp(I) - pointer to pointer to state table entry */ +/* */ +/* Decrement the reference counter for this state table entry and free it */ +/* if there are no more things using it. */ +/* */ +/* This function is only called when cleaning up after increasing is_ref by */ +/* one earlier in the 'code path' so if is_ref is 1 when entering, we do */ +/* have an orphan, otherwise not. However there is a possible race between */ +/* the entry being deleted via flushing with an ioctl call (that calls the */ +/* delete function directly) and the tail end of packet processing so we */ +/* need to grab is_lock before doing the check to synchronise the two code */ +/* paths. */ +/* */ +/* When operating in userland (ipftest), we have no timers to clear a state */ +/* entry. Therefore, we make a few simple tests before deleting an entry */ +/* outright. We compare states on each side looking for a combination of */ +/* TIME_WAIT (should really be FIN_WAIT_2?) and LAST_ACK. Then we factor */ +/* in packet direction with the interface list to make sure we don't */ +/* prematurely delete an entry on a final inbound packet that's we're also */ +/* supposed to route elsewhere. */ +/* */ +/* Internal 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 */ +/* ------------------------------------------------------------------------ */ +void fr_statederef(isp) +ipstate_t **isp; +{ + ipstate_t *is; + + is = *isp; + *isp = NULL; + + MUTEX_ENTER(&is->is_lock); + if (is->is_ref > 1) { + is->is_ref--; + MUTEX_EXIT(&is->is_lock); +#ifndef _KERNEL + if ((is->is_sti.tqe_state[0] > IPF_TCPS_ESTABLISHED) || + (is->is_sti.tqe_state[1] > IPF_TCPS_ESTABLISHED)) { + fr_delstate(is, ISL_ORPHAN); + } +#endif + return; + } + MUTEX_EXIT(&is->is_lock); + + WRITE_ENTER(&ipf_state); + fr_delstate(is, ISL_EXPIRE); + RWLOCK_EXIT(&ipf_state); +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_setstatequeue */ +/* Returns: Nil */ +/* Parameters: is(I) - pointer to state structure */ +/* rev(I) - forward(0) or reverse(1) direction */ +/* Locks: ipf_state (read or write) */ +/* */ +/* Put the state entry on its default queue entry, using rev as a helped in */ +/* determining which queue it should be placed on. */ +/* ------------------------------------------------------------------------ */ +void fr_setstatequeue(is, rev) +ipstate_t *is; +int rev; +{ + ipftq_t *oifq, *nifq; + + + if ((is->is_sti.tqe_flags & TQE_RULEBASED) != 0) + nifq = is->is_tqehead[rev]; + else + nifq = NULL; + + if (nifq == NULL) { + switch (is->is_p) + { +#ifdef USE_INET6 + case IPPROTO_ICMPV6 : + if (rev == 1) + nifq = &ips_icmpacktq; + else + nifq = &ips_icmptq; + break; +#endif + case IPPROTO_ICMP : + if (rev == 1) + nifq = &ips_icmpacktq; + else + nifq = &ips_icmptq; + break; + case IPPROTO_TCP : + nifq = ips_tqtqb + is->is_state[rev]; + break; + + case IPPROTO_UDP : + if (rev == 1) + nifq = &ips_udpacktq; + else + nifq = &ips_udptq; + break; + + default : + nifq = &ips_iptq; + break; + } + } + + oifq = is->is_sti.tqe_ifq; + /* + * If it's currently on a timeout queue, move it from one queue to + * another, else put it on the end of the newly determined queue. + */ + if (oifq != NULL) + fr_movequeue(&is->is_sti, oifq, nifq); + else + fr_queueappend(&is->is_sti, nifq, is); + return; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_stateiter */ +/* Returns: int - 0 == success, else error */ +/* Parameters: token(I) - pointer to ipftoken structure */ +/* itp(I) - pointer to ipfgeniter structure */ +/* */ +/* This function handles the SIOCGENITER ioctl for the state tables and */ +/* walks through the list of entries in the state table list (ips_list.) */ +/* ------------------------------------------------------------------------ */ +static int fr_stateiter(token, itp) +ipftoken_t *token; +ipfgeniter_t *itp; +{ + ipstate_t *is, *next, zero; + int error, count; + char *dst; + + if (itp->igi_data == NULL) + return EFAULT; + + if (itp->igi_nitems < 1) + return ENOSPC; + + if (itp->igi_type != IPFGENITER_STATE) + return EINVAL; + + is = token->ipt_data; + if (is == (void *)-1) { + ipf_freetoken(token); + return ESRCH; + } + + error = 0; + dst = itp->igi_data; + + READ_ENTER(&ipf_state); + if (is == NULL) { + next = ips_list; + } else { + next = is->is_next; + } + + count = itp->igi_nitems; + for (;;) { + if (next != NULL) { + /* + * If we find a state entry to use, bump its + * reference count so that it can be used for + * is_next when we come back. + */ + if (count == 1) { + MUTEX_ENTER(&next->is_lock); + next->is_ref++; + MUTEX_EXIT(&next->is_lock); + token->ipt_data = next; + } + } else { + bzero(&zero, sizeof(zero)); + next = &zero; + count = 1; + token->ipt_data = NULL; + } + RWLOCK_EXIT(&ipf_state); + + /* + * This should arguably be via fr_outobj() so that the state + * structure can (if required) be massaged going out. + */ + error = COPYOUT(next, dst, sizeof(*next)); + if (error != 0) + error = EFAULT; + if ((count == 1) || (error != 0)) + break; + + dst += sizeof(*next); + count--; + + READ_ENTER(&ipf_state); + next = next->is_next; + } + + if (is != NULL) { + fr_statederef(&is); + } + + return error; +} + + +/* ------------------------------------------------------------------------ */ +/* Function: fr_stgettable */ +/* Returns: int - 0 = success, else error */ +/* Parameters: data(I) - pointer to ioctl data */ +/* */ +/* This function handles ioctl requests for tables of state information. */ +/* At present the only table it deals with is the hash bucket statistics. */ +/* ------------------------------------------------------------------------ */ +static int fr_stgettable(data) +char *data; +{ + ipftable_t table; + int error; + + error = fr_inobj(data, &table, IPFOBJ_GTABLE); + if (error != 0) + return error; + + if (table.ita_type != IPFTABLE_BUCKETS) + return EINVAL; + + error = COPYOUT(ips_stats.iss_bucketlen, table.ita_table, + fr_statesize * sizeof(u_long)); + if (error != 0) + error = EFAULT; + return error; +} |
