X-Git-Url: https://git.saurik.com/apple/xnu.git/blobdiff_plain/fe8ab488e9161c46dd9885d58fc52996dc0249ff..4ba76501152d51ccb5647018f3192c6096367d48:/bsd/net/pf.c diff --git a/bsd/net/pf.c b/bsd/net/pf.c index 0a74fe5d2..87a392533 100644 --- a/bsd/net/pf.c +++ b/bsd/net/pf.c @@ -1,5 +1,5 @@ -/* - * Copyright (c) 2007-2013 Apple Inc. All rights reserved. +/* + * Copyright (c) 2007-2019 Apple Inc. All rights reserved. * * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ * @@ -31,7 +31,8 @@ /* * Copyright (c) 2001 Daniel Hartmeier - * Copyright (c) 2002,2003 Henning Brauer + * Copyright (c) 2002 - 2013 Henning Brauer + * NAT64 - Copyright (c) 2010 Viagenie Inc. (http://www.viagenie.ca) * All rights reserved. * * Redistribution and use in source and binary forms, with or without @@ -67,7 +68,6 @@ #include #include #include -#include #include #include #include @@ -107,6 +107,7 @@ #include #include #include +#include #include #include @@ -127,12 +128,12 @@ #endif /* DUMMYNET */ /* - * For RandomULong(), to get a 32 bits random value + * For RandomULong(), to get a 32 bits random value * Note that random() returns a 31 bits value, see rdar://11159750 */ #include -#define DPFPRINTF(n, x) (pf_status.debug >= (n) ? printf x : ((void)0)) +#define DPFPRINTF(n, x) (pf_status.debug >= (n) ? printf x : ((void)0)) /* * On Mac OS X, the rtableid value is treated as the interface scope @@ -143,51 +144,40 @@ * the test against INT_MAX to handle userland apps which initialize * the field with a negative number. */ -#define PF_RTABLEID_IS_VALID(r) \ +#define PF_RTABLEID_IS_VALID(r) \ ((r) > IFSCOPE_NONE && (r) <= INT_MAX) /* * Global variables */ -decl_lck_mtx_data(,pf_lock_data); -decl_lck_rw_data(,pf_perim_lock_data); +decl_lck_mtx_data(, pf_lock_data); +decl_lck_rw_data(, pf_perim_lock_data); lck_mtx_t *pf_lock = &pf_lock_data; lck_rw_t *pf_perim_lock = &pf_perim_lock_data; /* state tables */ -struct pf_state_tree_lan_ext pf_statetbl_lan_ext; -struct pf_state_tree_ext_gwy pf_statetbl_ext_gwy; - -struct pf_palist pf_pabuf; -struct pf_status pf_status; - -#if PF_ALTQ -struct pf_altqqueue pf_altqs[2]; -struct pf_altqqueue *pf_altqs_active; -struct pf_altqqueue *pf_altqs_inactive; -u_int32_t ticket_altqs_active; -u_int32_t ticket_altqs_inactive; -int altqs_inactive_open; -#endif /* PF_ALTQ */ -u_int32_t ticket_pabuf; - -static MD5_CTX pf_tcp_secret_ctx; -static u_char pf_tcp_secret[16]; -static int pf_tcp_secret_init; -static int pf_tcp_iss_off; +struct pf_state_tree_lan_ext pf_statetbl_lan_ext; +struct pf_state_tree_ext_gwy pf_statetbl_ext_gwy; + +struct pf_palist pf_pabuf; +struct pf_status pf_status; + +u_int32_t ticket_pabuf; + +static MD5_CTX pf_tcp_secret_ctx; +static u_char pf_tcp_secret[16]; +static int pf_tcp_secret_init; +static int pf_tcp_iss_off; static struct pf_anchor_stackframe { - struct pf_ruleset *rs; - struct pf_rule *r; - struct pf_anchor_node *parent; - struct pf_anchor *child; + struct pf_ruleset *rs; + struct pf_rule *r; + struct pf_anchor_node *parent; + struct pf_anchor *child; } pf_anchor_stack[64]; -struct pool pf_src_tree_pl, pf_rule_pl, pf_pooladdr_pl; -struct pool pf_state_pl, pf_state_key_pl; -#if PF_ALTQ -struct pool pf_altq_pl; -#endif /* PF_ALTQ */ +struct pool pf_src_tree_pl, pf_rule_pl, pf_pooladdr_pl; +struct pool pf_state_pl, pf_state_key_pl; typedef void (*hook_fn_t)(void *); @@ -197,179 +187,188 @@ struct hook_desc { void *hd_arg; }; -#define HOOK_REMOVE 0x01 -#define HOOK_FREE 0x02 -#define HOOK_ABORT 0x04 +#define HOOK_REMOVE 0x01 +#define HOOK_FREE 0x02 +#define HOOK_ABORT 0x04 -static void *hook_establish(struct hook_desc_head *, int, - hook_fn_t, void *); -static void hook_runloop(struct hook_desc_head *, int flags); +static void *hook_establish(struct hook_desc_head *, int, + hook_fn_t, void *); +static void hook_runloop(struct hook_desc_head *, int flags); -struct pool pf_app_state_pl; -static void pf_print_addr(struct pf_addr *addr, sa_family_t af); -static void pf_print_sk_host(struct pf_state_host *, u_int8_t, int, - u_int8_t); +struct pool pf_app_state_pl; +static void pf_print_addr(struct pf_addr *addr, sa_family_t af); +static void pf_print_sk_host(struct pf_state_host *, u_int8_t, int, + u_int8_t); -static void pf_print_host(struct pf_addr *, u_int16_t, u_int8_t); +static void pf_print_host(struct pf_addr *, u_int16_t, u_int8_t); -static void pf_init_threshold(struct pf_threshold *, u_int32_t, - u_int32_t); -static void pf_add_threshold(struct pf_threshold *); -static int pf_check_threshold(struct pf_threshold *); +static void pf_init_threshold(struct pf_threshold *, u_int32_t, + u_int32_t); +static void pf_add_threshold(struct pf_threshold *); +static int pf_check_threshold(struct pf_threshold *); -static void pf_change_ap(int, struct mbuf *, struct pf_addr *, - u_int16_t *, u_int16_t *, u_int16_t *, - struct pf_addr *, u_int16_t, u_int8_t, sa_family_t); -static int pf_modulate_sack(struct mbuf *, int, struct pf_pdesc *, - struct tcphdr *, struct pf_state_peer *); +static void pf_change_ap(int, pbuf_t *, struct pf_addr *, + u_int16_t *, u_int16_t *, u_int16_t *, + struct pf_addr *, u_int16_t, u_int8_t, sa_family_t, + sa_family_t, int); +static int pf_modulate_sack(pbuf_t *, int, struct pf_pdesc *, + struct tcphdr *, struct pf_state_peer *); #if INET6 -static void pf_change_a6(struct pf_addr *, u_int16_t *, - struct pf_addr *, u_int8_t); +static void pf_change_a6(struct pf_addr *, u_int16_t *, + struct pf_addr *, u_int8_t); +void pf_change_addr(struct pf_addr *a, u_int16_t *c, + struct pf_addr *an, u_int8_t u, + sa_family_t af, sa_family_t afn); #endif /* INET6 */ -static void pf_change_icmp(struct pf_addr *, u_int16_t *, - struct pf_addr *, struct pf_addr *, u_int16_t, - u_int16_t *, u_int16_t *, u_int16_t *, - u_int16_t *, u_int8_t, sa_family_t); -static void pf_send_tcp(const struct pf_rule *, sa_family_t, - const struct pf_addr *, const struct pf_addr *, - u_int16_t, u_int16_t, u_int32_t, u_int32_t, - u_int8_t, u_int16_t, u_int16_t, u_int8_t, int, - u_int16_t, struct ether_header *, struct ifnet *); -static void pf_send_icmp(struct mbuf *, u_int8_t, u_int8_t, - sa_family_t, struct pf_rule *); -static struct pf_rule *pf_match_translation(struct pf_pdesc *, struct mbuf *, - int, int, struct pfi_kif *, struct pf_addr *, - union pf_state_xport *, struct pf_addr *, - union pf_state_xport *, int); -static struct pf_rule *pf_get_translation_aux(struct pf_pdesc *, - struct mbuf *, int, int, struct pfi_kif *, - struct pf_src_node **, struct pf_addr *, - union pf_state_xport *, struct pf_addr *, - union pf_state_xport *, struct pf_addr *, - union pf_state_xport *); -static void pf_attach_state(struct pf_state_key *, - struct pf_state *, int); -static void pf_detach_state(struct pf_state *, int); -static u_int32_t pf_tcp_iss(struct pf_pdesc *); -static int pf_test_rule(struct pf_rule **, struct pf_state **, - int, struct pfi_kif *, struct mbuf *, int, - void *, struct pf_pdesc *, struct pf_rule **, - struct pf_ruleset **, struct ifqueue *); +static void pf_change_icmp(struct pf_addr *, u_int16_t *, + struct pf_addr *, struct pf_addr *, u_int16_t, + u_int16_t *, u_int16_t *, u_int16_t *, + u_int16_t *, u_int8_t, sa_family_t); +static void pf_send_tcp(const struct pf_rule *, sa_family_t, + const struct pf_addr *, const struct pf_addr *, + u_int16_t, u_int16_t, u_int32_t, u_int32_t, + u_int8_t, u_int16_t, u_int16_t, u_int8_t, int, + u_int16_t, struct ether_header *, struct ifnet *); +static void pf_send_icmp(pbuf_t *, u_int8_t, u_int8_t, + sa_family_t, struct pf_rule *); +static struct pf_rule *pf_match_translation(struct pf_pdesc *, pbuf_t *, + int, int, struct pfi_kif *, struct pf_addr *, + union pf_state_xport *, struct pf_addr *, + union pf_state_xport *, int); +static struct pf_rule *pf_get_translation_aux(struct pf_pdesc *, + pbuf_t *, int, int, struct pfi_kif *, + struct pf_src_node **, struct pf_addr *, + union pf_state_xport *, struct pf_addr *, + union pf_state_xport *, union pf_state_xport * + ); +static void pf_attach_state(struct pf_state_key *, + struct pf_state *, int); +static void pf_detach_state(struct pf_state *, int); +static u_int32_t pf_tcp_iss(struct pf_pdesc *); +static int pf_test_rule(struct pf_rule **, struct pf_state **, + int, struct pfi_kif *, pbuf_t *, int, + void *, struct pf_pdesc *, struct pf_rule **, + struct pf_ruleset **, struct ifqueue *); #if DUMMYNET -static int pf_test_dummynet(struct pf_rule **, int, - struct pfi_kif *, struct mbuf **, - struct pf_pdesc *, struct ip_fw_args *); +static int pf_test_dummynet(struct pf_rule **, int, + struct pfi_kif *, pbuf_t **, + struct pf_pdesc *, struct ip_fw_args *); #endif /* DUMMYNET */ -static int pf_test_fragment(struct pf_rule **, int, - struct pfi_kif *, struct mbuf *, void *, - struct pf_pdesc *, struct pf_rule **, - struct pf_ruleset **); -static int pf_test_state_tcp(struct pf_state **, int, - struct pfi_kif *, struct mbuf *, int, - void *, struct pf_pdesc *, u_short *); -static int pf_test_state_udp(struct pf_state **, int, - struct pfi_kif *, struct mbuf *, int, - void *, struct pf_pdesc *, u_short *); -static int pf_test_state_icmp(struct pf_state **, int, - struct pfi_kif *, struct mbuf *, int, - void *, struct pf_pdesc *, u_short *); -static int pf_test_state_other(struct pf_state **, int, - struct pfi_kif *, struct pf_pdesc *); -static int pf_match_tag(struct mbuf *, struct pf_rule *, - struct pf_mtag *, int *); -static void pf_hash(struct pf_addr *, struct pf_addr *, - struct pf_poolhashkey *, sa_family_t); -static int pf_map_addr(u_int8_t, struct pf_rule *, - struct pf_addr *, struct pf_addr *, - struct pf_addr *, struct pf_src_node **); -static int pf_get_sport(struct pf_pdesc *, struct pfi_kif *, - struct pf_rule *, struct pf_addr *, - union pf_state_xport *, struct pf_addr *, - union pf_state_xport *, struct pf_addr *, - union pf_state_xport *, struct pf_src_node **); -static void pf_route(struct mbuf **, struct pf_rule *, int, - struct ifnet *, struct pf_state *, - struct pf_pdesc *); +static int pf_test_fragment(struct pf_rule **, int, + struct pfi_kif *, pbuf_t *, void *, + struct pf_pdesc *, struct pf_rule **, + struct pf_ruleset **); +static int pf_test_state_tcp(struct pf_state **, int, + struct pfi_kif *, pbuf_t *, int, + void *, struct pf_pdesc *, u_short *); +static int pf_test_state_udp(struct pf_state **, int, + struct pfi_kif *, pbuf_t *, int, + void *, struct pf_pdesc *, u_short *); +static int pf_test_state_icmp(struct pf_state **, int, + struct pfi_kif *, pbuf_t *, int, + void *, struct pf_pdesc *, u_short *); +static int pf_test_state_other(struct pf_state **, int, + struct pfi_kif *, struct pf_pdesc *); +static int pf_match_tag(struct pf_rule *, + struct pf_mtag *, int *); +static void pf_hash(struct pf_addr *, struct pf_addr *, + struct pf_poolhashkey *, sa_family_t); +static int pf_map_addr(u_int8_t, struct pf_rule *, + struct pf_addr *, struct pf_addr *, + struct pf_addr *, struct pf_src_node **); +static int pf_get_sport(struct pf_pdesc *, struct pfi_kif *, + struct pf_rule *, struct pf_addr *, + union pf_state_xport *, struct pf_addr *, + union pf_state_xport *, struct pf_addr *, + union pf_state_xport *, struct pf_src_node ** + ); +static void pf_route(pbuf_t **, struct pf_rule *, int, + struct ifnet *, struct pf_state *, + struct pf_pdesc *); #if INET6 -static void pf_route6(struct mbuf **, struct pf_rule *, int, - struct ifnet *, struct pf_state *, - struct pf_pdesc *); +static void pf_route6(pbuf_t **, struct pf_rule *, int, + struct ifnet *, struct pf_state *, + struct pf_pdesc *); #endif /* INET6 */ -static u_int8_t pf_get_wscale(struct mbuf *, int, u_int16_t, - sa_family_t); -static u_int16_t pf_get_mss(struct mbuf *, int, u_int16_t, - sa_family_t); -static u_int16_t pf_calc_mss(struct pf_addr *, sa_family_t, - u_int16_t); -static void pf_set_rt_ifp(struct pf_state *, - struct pf_addr *); -static int pf_check_proto_cksum(struct mbuf *, int, int, - u_int8_t, sa_family_t); -static int pf_addr_wrap_neq(struct pf_addr_wrap *, - struct pf_addr_wrap *); -static struct pf_state *pf_find_state(struct pfi_kif *, - struct pf_state_key_cmp *, u_int); -static int pf_src_connlimit(struct pf_state **); -static void pf_stateins_err(const char *, struct pf_state *, - struct pfi_kif *); -static int pf_check_congestion(struct ifqueue *); +static u_int8_t pf_get_wscale(pbuf_t *, int, u_int16_t, + sa_family_t); +static u_int16_t pf_get_mss(pbuf_t *, int, u_int16_t, + sa_family_t); +static u_int16_t pf_calc_mss(struct pf_addr *, sa_family_t, + u_int16_t); +static void pf_set_rt_ifp(struct pf_state *, + struct pf_addr *, sa_family_t af); +static int pf_check_proto_cksum(pbuf_t *, int, int, + u_int8_t, sa_family_t); +static int pf_addr_wrap_neq(struct pf_addr_wrap *, + struct pf_addr_wrap *); +static struct pf_state *pf_find_state(struct pfi_kif *, + struct pf_state_key_cmp *, u_int); +static int pf_src_connlimit(struct pf_state **); +static void pf_stateins_err(const char *, struct pf_state *, + struct pfi_kif *); +static int pf_check_congestion(struct ifqueue *); #if 0 static const char *pf_pptp_ctrl_type_name(u_int16_t code); #endif -static void pf_pptp_handler(struct pf_state *, int, int, - struct pf_pdesc *, struct pfi_kif *); -static void pf_pptp_unlink(struct pf_state *); -static void pf_grev1_unlink(struct pf_state *); -static int pf_test_state_grev1(struct pf_state **, int, - struct pfi_kif *, int, struct pf_pdesc *); -static int pf_ike_compare(struct pf_app_state *, - struct pf_app_state *); -static int pf_test_state_esp(struct pf_state **, int, - struct pfi_kif *, int, struct pf_pdesc *); +static void pf_pptp_handler(struct pf_state *, int, int, + struct pf_pdesc *, struct pfi_kif *); +static void pf_pptp_unlink(struct pf_state *); +static void pf_grev1_unlink(struct pf_state *); +static int pf_test_state_grev1(struct pf_state **, int, + struct pfi_kif *, int, struct pf_pdesc *); +static int pf_ike_compare(struct pf_app_state *, + struct pf_app_state *); +static int pf_test_state_esp(struct pf_state **, int, + struct pfi_kif *, int, struct pf_pdesc *); extern struct pool pfr_ktable_pl; extern struct pool pfr_kentry_pl; extern int path_mtu_discovery; struct pf_pool_limit pf_pool_limits[PF_LIMIT_MAX] = { - { &pf_state_pl, PFSTATE_HIWAT }, - { &pf_app_state_pl, PFAPPSTATE_HIWAT }, - { &pf_src_tree_pl, PFSNODE_HIWAT }, - { &pf_frent_pl, PFFRAG_FRENT_HIWAT }, - { &pfr_ktable_pl, PFR_KTABLE_HIWAT }, - { &pfr_kentry_pl, PFR_KENTRY_HIWAT }, + { .pp = &pf_state_pl, .limit = PFSTATE_HIWAT }, + { .pp = &pf_app_state_pl, .limit = PFAPPSTATE_HIWAT }, + { .pp = &pf_src_tree_pl, .limit = PFSNODE_HIWAT }, + { .pp = &pf_frent_pl, .limit = PFFRAG_FRENT_HIWAT }, + { .pp = &pfr_ktable_pl, .limit = PFR_KTABLE_HIWAT }, + { .pp = &pfr_kentry_pl, .limit = PFR_KENTRY_HIWAT }, }; -struct mbuf * -pf_lazy_makewritable(struct pf_pdesc *pd, struct mbuf *m, int len) +void * +pf_lazy_makewritable(struct pf_pdesc *pd, pbuf_t *pbuf, int len) { - if (pd->lmw < 0) - return (0); + void *p; + + if (pd->lmw < 0) { + return NULL; + } - VERIFY(m == pd->mp); + VERIFY(pbuf == pd->mp); + p = pbuf->pb_data; if (len > pd->lmw) { - if (m_makewritable(&m, 0, len, M_DONTWAIT)) + if ((p = pbuf_ensure_writable(pbuf, len)) == NULL) { len = -1; + } pd->lmw = len; - if (len >= 0 && m != pd->mp) { - pd->mp = m; - pd->pf_mtag = pf_find_mtag(m); + if (len >= 0) { + pd->pf_mtag = pf_find_mtag_pbuf(pbuf); switch (pd->af) { case AF_INET: { - struct ip *h = mtod(m, struct ip *); - pd->src = (struct pf_addr *)&h->ip_src; - pd->dst = (struct pf_addr *)&h->ip_dst; + struct ip *h = p; + pd->src = (struct pf_addr *)(uintptr_t)&h->ip_src; + pd->dst = (struct pf_addr *)(uintptr_t)&h->ip_dst; pd->ip_sum = &h->ip_sum; break; } #if INET6 case AF_INET6: { - struct ip6_hdr *h = mtod(m, struct ip6_hdr *); - pd->src = (struct pf_addr *)&h->ip6_src; - pd->dst = (struct pf_addr *)&h->ip6_dst; + struct ip6_hdr *h = p; + pd->src = (struct pf_addr *)(uintptr_t)&h->ip6_src; + pd->dst = (struct pf_addr *)(uintptr_t)&h->ip6_dst; break; } #endif /* INET6 */ @@ -377,16 +376,16 @@ pf_lazy_makewritable(struct pf_pdesc *pd, struct mbuf *m, int len) } } - return (len < 0 ? 0 : m); + return len < 0 ? NULL : p; } static const int * pf_state_lookup_aux(struct pf_state **state, struct pfi_kif *kif, - int direction, int *action) + int direction, int *action) { if (*state == NULL || (*state)->timeout == PFTM_PURGE) { *action = PF_DROP; - return (action); + return action; } if (direction == PF_OUT && @@ -396,82 +395,83 @@ pf_state_lookup_aux(struct pf_state **state, struct pfi_kif *kif, (*state)->rule.ptr->direction == PF_IN)) && (*state)->rt_kif != NULL && (*state)->rt_kif != kif) { *action = PF_PASS; - return (action); + return action; } - return (0); + return 0; } -#define STATE_LOOKUP() \ - do { \ - int action; \ - *state = pf_find_state(kif, &key, direction); \ - if (*state != NULL && pd != NULL && \ - !(pd->pktflags & PKTF_FLOW_ID)) { \ - pd->flowsrc = (*state)->state_key->flowsrc; \ - pd->flowhash = (*state)->state_key->flowhash; \ - if (pd->flowhash != 0) { \ - pd->pktflags |= PKTF_FLOW_ID; \ - pd->pktflags &= ~PKTF_FLOW_ADV; \ - } \ - } \ - if (pf_state_lookup_aux(state, kif, direction, &action)) \ - return (action); \ +#define STATE_LOOKUP() \ + do { \ + int action; \ + *state = pf_find_state(kif, &key, direction); \ + if (*state != NULL && pd != NULL && \ + !(pd->pktflags & PKTF_FLOW_ID)) { \ + pd->flowsrc = (*state)->state_key->flowsrc; \ + pd->flowhash = (*state)->state_key->flowhash; \ + if (pd->flowhash != 0) { \ + pd->pktflags |= PKTF_FLOW_ID; \ + pd->pktflags &= ~PKTF_FLOW_ADV; \ + } \ + } \ + if (pf_state_lookup_aux(state, kif, direction, &action)) \ + return (action); \ } while (0) -#define STATE_ADDR_TRANSLATE(sk) \ - (sk)->lan.addr.addr32[0] != (sk)->gwy.addr.addr32[0] || \ - ((sk)->af == AF_INET6 && \ - ((sk)->lan.addr.addr32[1] != (sk)->gwy.addr.addr32[1] || \ - (sk)->lan.addr.addr32[2] != (sk)->gwy.addr.addr32[2] || \ +#define STATE_ADDR_TRANSLATE(sk) \ + (sk)->lan.addr.addr32[0] != (sk)->gwy.addr.addr32[0] || \ + ((sk)->af_lan == AF_INET6 && \ + ((sk)->lan.addr.addr32[1] != (sk)->gwy.addr.addr32[1] || \ + (sk)->lan.addr.addr32[2] != (sk)->gwy.addr.addr32[2] || \ (sk)->lan.addr.addr32[3] != (sk)->gwy.addr.addr32[3])) -#define STATE_TRANSLATE(sk) \ - (STATE_ADDR_TRANSLATE(sk) || \ +#define STATE_TRANSLATE(sk) \ + ((sk)->af_lan != (sk)->af_gwy || \ + STATE_ADDR_TRANSLATE(sk) || \ (sk)->lan.xport.port != (sk)->gwy.xport.port) -#define STATE_GRE_TRANSLATE(sk) \ - (STATE_ADDR_TRANSLATE(sk) || \ +#define STATE_GRE_TRANSLATE(sk) \ + (STATE_ADDR_TRANSLATE(sk) || \ (sk)->lan.xport.call_id != (sk)->gwy.xport.call_id) #define BOUND_IFACE(r, k) \ ((r)->rule_flag & PFRULE_IFBOUND) ? (k) : pfi_all -#define STATE_INC_COUNTERS(s) \ - do { \ - s->rule.ptr->states++; \ - VERIFY(s->rule.ptr->states != 0); \ - if (s->anchor.ptr != NULL) { \ - s->anchor.ptr->states++; \ - VERIFY(s->anchor.ptr->states != 0); \ - } \ - if (s->nat_rule.ptr != NULL) { \ - s->nat_rule.ptr->states++; \ - VERIFY(s->nat_rule.ptr->states != 0); \ - } \ +#define STATE_INC_COUNTERS(s) \ + do { \ + s->rule.ptr->states++; \ + VERIFY(s->rule.ptr->states != 0); \ + if (s->anchor.ptr != NULL) { \ + s->anchor.ptr->states++; \ + VERIFY(s->anchor.ptr->states != 0); \ + } \ + if (s->nat_rule.ptr != NULL) { \ + s->nat_rule.ptr->states++; \ + VERIFY(s->nat_rule.ptr->states != 0); \ + } \ } while (0) -#define STATE_DEC_COUNTERS(s) \ - do { \ - if (s->nat_rule.ptr != NULL) { \ - VERIFY(s->nat_rule.ptr->states > 0); \ - s->nat_rule.ptr->states--; \ - } \ - if (s->anchor.ptr != NULL) { \ - VERIFY(s->anchor.ptr->states > 0); \ - s->anchor.ptr->states--; \ - } \ - VERIFY(s->rule.ptr->states > 0); \ - s->rule.ptr->states--; \ +#define STATE_DEC_COUNTERS(s) \ + do { \ + if (s->nat_rule.ptr != NULL) { \ + VERIFY(s->nat_rule.ptr->states > 0); \ + s->nat_rule.ptr->states--; \ + } \ + if (s->anchor.ptr != NULL) { \ + VERIFY(s->anchor.ptr->states > 0); \ + s->anchor.ptr->states--; \ + } \ + VERIFY(s->rule.ptr->states > 0); \ + s->rule.ptr->states--; \ } while (0) static __inline int pf_src_compare(struct pf_src_node *, struct pf_src_node *); static __inline int pf_state_compare_lan_ext(struct pf_state_key *, - struct pf_state_key *); + struct pf_state_key *); static __inline int pf_state_compare_ext_gwy(struct pf_state_key *, - struct pf_state_key *); + struct pf_state_key *); static __inline int pf_state_compare_id(struct pf_state *, - struct pf_state *); + struct pf_state *); struct pf_src_tree tree_src_tracking; @@ -486,177 +486,178 @@ RB_GENERATE(pf_state_tree_ext_gwy, pf_state_key, RB_GENERATE(pf_state_tree_id, pf_state, entry_id, pf_state_compare_id); -#define PF_DT_SKIP_LANEXT 0x01 -#define PF_DT_SKIP_EXTGWY 0x02 +#define PF_DT_SKIP_LANEXT 0x01 +#define PF_DT_SKIP_EXTGWY 0x02 static const u_int16_t PF_PPTP_PORT = 1723; static const u_int32_t PF_PPTP_MAGIC_NUMBER = 0x1A2B3C4D; struct pf_pptp_hdr { - u_int16_t length; - u_int16_t type; - u_int32_t magic; + u_int16_t length; + u_int16_t type; + u_int32_t magic; }; struct pf_pptp_ctrl_hdr { - u_int16_t type; - u_int16_t reserved_0; + u_int16_t type; + u_int16_t reserved_0; }; struct pf_pptp_ctrl_generic { - u_int16_t data[0]; + u_int16_t data[0]; }; -#define PF_PPTP_CTRL_TYPE_START_REQ 1 +#define PF_PPTP_CTRL_TYPE_START_REQ 1 struct pf_pptp_ctrl_start_req { - u_int16_t protocol_version; - u_int16_t reserved_1; - u_int32_t framing_capabilities; - u_int32_t bearer_capabilities; - u_int16_t maximum_channels; - u_int16_t firmware_revision; - u_int8_t host_name[64]; - u_int8_t vendor_string[64]; + u_int16_t protocol_version; + u_int16_t reserved_1; + u_int32_t framing_capabilities; + u_int32_t bearer_capabilities; + u_int16_t maximum_channels; + u_int16_t firmware_revision; + u_int8_t host_name[64]; + u_int8_t vendor_string[64]; }; -#define PF_PPTP_CTRL_TYPE_START_RPY 2 +#define PF_PPTP_CTRL_TYPE_START_RPY 2 struct pf_pptp_ctrl_start_rpy { - u_int16_t protocol_version; - u_int8_t result_code; - u_int8_t error_code; - u_int32_t framing_capabilities; - u_int32_t bearer_capabilities; - u_int16_t maximum_channels; - u_int16_t firmware_revision; - u_int8_t host_name[64]; - u_int8_t vendor_string[64]; + u_int16_t protocol_version; + u_int8_t result_code; + u_int8_t error_code; + u_int32_t framing_capabilities; + u_int32_t bearer_capabilities; + u_int16_t maximum_channels; + u_int16_t firmware_revision; + u_int8_t host_name[64]; + u_int8_t vendor_string[64]; }; -#define PF_PPTP_CTRL_TYPE_STOP_REQ 3 +#define PF_PPTP_CTRL_TYPE_STOP_REQ 3 struct pf_pptp_ctrl_stop_req { - u_int8_t reason; - u_int8_t reserved_1; - u_int16_t reserved_2; + u_int8_t reason; + u_int8_t reserved_1; + u_int16_t reserved_2; }; -#define PF_PPTP_CTRL_TYPE_STOP_RPY 4 +#define PF_PPTP_CTRL_TYPE_STOP_RPY 4 struct pf_pptp_ctrl_stop_rpy { - u_int8_t reason; - u_int8_t error_code; - u_int16_t reserved_1; + u_int8_t reason; + u_int8_t error_code; + u_int16_t reserved_1; }; -#define PF_PPTP_CTRL_TYPE_ECHO_REQ 5 +#define PF_PPTP_CTRL_TYPE_ECHO_REQ 5 struct pf_pptp_ctrl_echo_req { - u_int32_t identifier; + u_int32_t identifier; }; -#define PF_PPTP_CTRL_TYPE_ECHO_RPY 6 +#define PF_PPTP_CTRL_TYPE_ECHO_RPY 6 struct pf_pptp_ctrl_echo_rpy { - u_int32_t identifier; - u_int8_t result_code; - u_int8_t error_code; - u_int16_t reserved_1; + u_int32_t identifier; + u_int8_t result_code; + u_int8_t error_code; + u_int16_t reserved_1; }; -#define PF_PPTP_CTRL_TYPE_CALL_OUT_REQ 7 +#define PF_PPTP_CTRL_TYPE_CALL_OUT_REQ 7 struct pf_pptp_ctrl_call_out_req { - u_int16_t call_id; - u_int16_t call_sernum; - u_int32_t min_bps; - u_int32_t bearer_type; - u_int32_t framing_type; - u_int16_t rxwindow_size; - u_int16_t proc_delay; - u_int8_t phone_num[64]; - u_int8_t sub_addr[64]; + u_int16_t call_id; + u_int16_t call_sernum; + u_int32_t min_bps; + u_int32_t bearer_type; + u_int32_t framing_type; + u_int16_t rxwindow_size; + u_int16_t proc_delay; + u_int8_t phone_num[64]; + u_int8_t sub_addr[64]; }; -#define PF_PPTP_CTRL_TYPE_CALL_OUT_RPY 8 +#define PF_PPTP_CTRL_TYPE_CALL_OUT_RPY 8 struct pf_pptp_ctrl_call_out_rpy { - u_int16_t call_id; - u_int16_t peer_call_id; - u_int8_t result_code; - u_int8_t error_code; - u_int16_t cause_code; - u_int32_t connect_speed; - u_int16_t rxwindow_size; - u_int16_t proc_delay; - u_int32_t phy_channel_id; + u_int16_t call_id; + u_int16_t peer_call_id; + u_int8_t result_code; + u_int8_t error_code; + u_int16_t cause_code; + u_int32_t connect_speed; + u_int16_t rxwindow_size; + u_int16_t proc_delay; + u_int32_t phy_channel_id; }; -#define PF_PPTP_CTRL_TYPE_CALL_IN_1ST 9 +#define PF_PPTP_CTRL_TYPE_CALL_IN_1ST 9 struct pf_pptp_ctrl_call_in_1st { - u_int16_t call_id; - u_int16_t call_sernum; - u_int32_t bearer_type; - u_int32_t phy_channel_id; - u_int16_t dialed_number_len; - u_int16_t dialing_number_len; - u_int8_t dialed_num[64]; - u_int8_t dialing_num[64]; - u_int8_t sub_addr[64]; + u_int16_t call_id; + u_int16_t call_sernum; + u_int32_t bearer_type; + u_int32_t phy_channel_id; + u_int16_t dialed_number_len; + u_int16_t dialing_number_len; + u_int8_t dialed_num[64]; + u_int8_t dialing_num[64]; + u_int8_t sub_addr[64]; }; -#define PF_PPTP_CTRL_TYPE_CALL_IN_2ND 10 +#define PF_PPTP_CTRL_TYPE_CALL_IN_2ND 10 struct pf_pptp_ctrl_call_in_2nd { - u_int16_t call_id; - u_int16_t peer_call_id; - u_int8_t result_code; - u_int8_t error_code; - u_int16_t rxwindow_size; - u_int16_t txdelay; - u_int16_t reserved_1; + u_int16_t call_id; + u_int16_t peer_call_id; + u_int8_t result_code; + u_int8_t error_code; + u_int16_t rxwindow_size; + u_int16_t txdelay; + u_int16_t reserved_1; }; -#define PF_PPTP_CTRL_TYPE_CALL_IN_3RD 11 +#define PF_PPTP_CTRL_TYPE_CALL_IN_3RD 11 struct pf_pptp_ctrl_call_in_3rd { - u_int16_t call_id; - u_int16_t reserved_1; - u_int32_t connect_speed; - u_int16_t rxwindow_size; - u_int16_t txdelay; - u_int32_t framing_type; + u_int16_t call_id; + u_int16_t reserved_1; + u_int32_t connect_speed; + u_int16_t rxwindow_size; + u_int16_t txdelay; + u_int32_t framing_type; }; -#define PF_PPTP_CTRL_TYPE_CALL_CLR 12 +#define PF_PPTP_CTRL_TYPE_CALL_CLR 12 struct pf_pptp_ctrl_call_clr { - u_int16_t call_id; - u_int16_t reserved_1; + u_int16_t call_id; + u_int16_t reserved_1; }; -#define PF_PPTP_CTRL_TYPE_CALL_DISC 13 +#define PF_PPTP_CTRL_TYPE_CALL_DISC 13 struct pf_pptp_ctrl_call_disc { - u_int16_t call_id; - u_int8_t result_code; - u_int8_t error_code; - u_int16_t cause_code; - u_int16_t reserved_1; - u_int8_t statistics[128]; + u_int16_t call_id; + u_int8_t result_code; + u_int8_t error_code; + u_int16_t cause_code; + u_int16_t reserved_1; + u_int8_t statistics[128]; }; -#define PF_PPTP_CTRL_TYPE_ERROR 14 +#define PF_PPTP_CTRL_TYPE_ERROR 14 struct pf_pptp_ctrl_error { - u_int16_t peer_call_id; - u_int16_t reserved_1; - u_int32_t crc_errors; - u_int32_t fr_errors; - u_int32_t hw_errors; - u_int32_t buf_errors; - u_int32_t tim_errors; - u_int32_t align_errors; + u_int16_t peer_call_id; + u_int16_t reserved_1; + u_int32_t crc_errors; + u_int32_t fr_errors; + u_int32_t hw_errors; + u_int32_t buf_errors; + u_int32_t tim_errors; + u_int32_t align_errors; }; -#define PF_PPTP_CTRL_TYPE_SET_LINKINFO 15 +#define PF_PPTP_CTRL_TYPE_SET_LINKINFO 15 struct pf_pptp_ctrl_set_linkinfo { - u_int16_t peer_call_id; - u_int16_t reserved_1; - u_int32_t tx_accm; - u_int32_t rx_accm; + u_int16_t peer_call_id; + u_int16_t reserved_1; + u_int32_t tx_accm; + u_int32_t rx_accm; }; #if 0 -static const char *pf_pptp_ctrl_type_name(u_int16_t code) +static const char * +pf_pptp_ctrl_type_name(u_int16_t code) { code = ntohs(code); @@ -665,7 +666,7 @@ static const char *pf_pptp_ctrl_type_name(u_int16_t code) static char reserved[] = "reserved-00"; sprintf(&reserved[9], "%02x", code); - return (reserved); + return reserved; } else { static const char *name[] = { "start_req", "start_rpy", "stop_req", "stop_rpy", @@ -674,59 +675,42 @@ static const char *pf_pptp_ctrl_type_name(u_int16_t code) "call_clr", "call_disc", "error", "set_linkinfo" }; - return (name[code - 1]); + return name[code - 1]; } }; #endif static const size_t PF_PPTP_CTRL_MSG_MINSIZE = - sizeof (struct pf_pptp_hdr) + - sizeof (struct pf_pptp_ctrl_hdr) + - MIN(sizeof (struct pf_pptp_ctrl_start_req), - MIN(sizeof (struct pf_pptp_ctrl_start_rpy), - MIN(sizeof (struct pf_pptp_ctrl_stop_req), - MIN(sizeof (struct pf_pptp_ctrl_stop_rpy), - MIN(sizeof (struct pf_pptp_ctrl_echo_req), - MIN(sizeof (struct pf_pptp_ctrl_echo_rpy), - MIN(sizeof (struct pf_pptp_ctrl_call_out_req), - MIN(sizeof (struct pf_pptp_ctrl_call_out_rpy), - MIN(sizeof (struct pf_pptp_ctrl_call_in_1st), - MIN(sizeof (struct pf_pptp_ctrl_call_in_2nd), - MIN(sizeof (struct pf_pptp_ctrl_call_in_3rd), - MIN(sizeof (struct pf_pptp_ctrl_call_clr), - MIN(sizeof (struct pf_pptp_ctrl_call_disc), - MIN(sizeof (struct pf_pptp_ctrl_error), - sizeof (struct pf_pptp_ctrl_set_linkinfo) - )))))))))))))); + sizeof(struct pf_pptp_hdr) + sizeof(struct pf_pptp_ctrl_hdr); union pf_pptp_ctrl_msg_union { - struct pf_pptp_ctrl_start_req start_req; - struct pf_pptp_ctrl_start_rpy start_rpy; - struct pf_pptp_ctrl_stop_req stop_req; - struct pf_pptp_ctrl_stop_rpy stop_rpy; - struct pf_pptp_ctrl_echo_req echo_req; - struct pf_pptp_ctrl_echo_rpy echo_rpy; - struct pf_pptp_ctrl_call_out_req call_out_req; - struct pf_pptp_ctrl_call_out_rpy call_out_rpy; - struct pf_pptp_ctrl_call_in_1st call_in_1st; - struct pf_pptp_ctrl_call_in_2nd call_in_2nd; - struct pf_pptp_ctrl_call_in_3rd call_in_3rd; - struct pf_pptp_ctrl_call_clr call_clr; - struct pf_pptp_ctrl_call_disc call_disc; - struct pf_pptp_ctrl_error error; - struct pf_pptp_ctrl_set_linkinfo set_linkinfo; - u_int8_t data[0]; + struct pf_pptp_ctrl_start_req start_req; + struct pf_pptp_ctrl_start_rpy start_rpy; + struct pf_pptp_ctrl_stop_req stop_req; + struct pf_pptp_ctrl_stop_rpy stop_rpy; + struct pf_pptp_ctrl_echo_req echo_req; + struct pf_pptp_ctrl_echo_rpy echo_rpy; + struct pf_pptp_ctrl_call_out_req call_out_req; + struct pf_pptp_ctrl_call_out_rpy call_out_rpy; + struct pf_pptp_ctrl_call_in_1st call_in_1st; + struct pf_pptp_ctrl_call_in_2nd call_in_2nd; + struct pf_pptp_ctrl_call_in_3rd call_in_3rd; + struct pf_pptp_ctrl_call_clr call_clr; + struct pf_pptp_ctrl_call_disc call_disc; + struct pf_pptp_ctrl_error error; + struct pf_pptp_ctrl_set_linkinfo set_linkinfo; + u_int8_t data[0]; }; struct pf_pptp_ctrl_msg { - struct pf_pptp_hdr hdr; - struct pf_pptp_ctrl_hdr ctrl; - union pf_pptp_ctrl_msg_union msg; + struct pf_pptp_hdr hdr; + struct pf_pptp_ctrl_hdr ctrl; + union pf_pptp_ctrl_msg_union msg; }; -#define PF_GRE_FLAG_CHECKSUM_PRESENT 0x8000 -#define PF_GRE_FLAG_VERSION_MASK 0x0007 -#define PF_GRE_PPP_ETHERTYPE 0x880B +#define PF_GRE_FLAG_CHECKSUM_PRESENT 0x8000 +#define PF_GRE_FLAG_VERSION_MASK 0x0007 +#define PF_GRE_PPP_ETHERTYPE 0x880B struct pf_grev1_hdr { u_int16_t flags; @@ -734,9 +718,9 @@ struct pf_grev1_hdr { u_int16_t payload_length; u_int16_t call_id; /* - u_int32_t seqno; - u_int32_t ackno; - */ + * u_int32_t seqno; + * u_int32_t ackno; + */ }; static const u_int16_t PF_IKE_PORT = 500; @@ -747,24 +731,24 @@ struct pf_ike_hdr { u_int32_t message_id, length; }; -#define PF_IKE_PACKET_MINSIZE (sizeof (struct pf_ike_hdr)) - -#define PF_IKEv1_EXCHTYPE_BASE 1 -#define PF_IKEv1_EXCHTYPE_ID_PROTECT 2 -#define PF_IKEv1_EXCHTYPE_AUTH_ONLY 3 -#define PF_IKEv1_EXCHTYPE_AGGRESSIVE 4 -#define PF_IKEv1_EXCHTYPE_INFORMATIONAL 5 -#define PF_IKEv2_EXCHTYPE_SA_INIT 34 -#define PF_IKEv2_EXCHTYPE_AUTH 35 -#define PF_IKEv2_EXCHTYPE_CREATE_CHILD_SA 36 -#define PF_IKEv2_EXCHTYPE_INFORMATIONAL 37 - -#define PF_IKEv1_FLAG_E 0x01 -#define PF_IKEv1_FLAG_C 0x02 -#define PF_IKEv1_FLAG_A 0x04 -#define PF_IKEv2_FLAG_I 0x08 -#define PF_IKEv2_FLAG_V 0x10 -#define PF_IKEv2_FLAG_R 0x20 +#define PF_IKE_PACKET_MINSIZE (sizeof (struct pf_ike_hdr)) + +#define PF_IKEv1_EXCHTYPE_BASE 1 +#define PF_IKEv1_EXCHTYPE_ID_PROTECT 2 +#define PF_IKEv1_EXCHTYPE_AUTH_ONLY 3 +#define PF_IKEv1_EXCHTYPE_AGGRESSIVE 4 +#define PF_IKEv1_EXCHTYPE_INFORMATIONAL 5 +#define PF_IKEv2_EXCHTYPE_SA_INIT 34 +#define PF_IKEv2_EXCHTYPE_AUTH 35 +#define PF_IKEv2_EXCHTYPE_CREATE_CHILD_SA 36 +#define PF_IKEv2_EXCHTYPE_INFORMATIONAL 37 + +#define PF_IKEv1_FLAG_E 0x01 +#define PF_IKEv1_FLAG_C 0x02 +#define PF_IKEv1_FLAG_A 0x04 +#define PF_IKEv2_FLAG_I 0x08 +#define PF_IKEv2_FLAG_V 0x10 +#define PF_IKEv2_FLAG_R 0x20 struct pf_esp_hdr { u_int32_t spi; @@ -773,156 +757,168 @@ struct pf_esp_hdr { }; static __inline int -pf_src_compare(struct pf_src_node *a, struct pf_src_node *b) +pf_addr_compare(struct pf_addr *a, struct pf_addr *b, sa_family_t af) { - int diff; - - if (a->rule.ptr > b->rule.ptr) - return (1); - if (a->rule.ptr < b->rule.ptr) - return (-1); - if ((diff = a->af - b->af) != 0) - return (diff); - switch (a->af) { -#if INET + switch (af) { +#ifdef INET case AF_INET: - if (a->addr.addr32[0] > b->addr.addr32[0]) - return (1); - if (a->addr.addr32[0] < b->addr.addr32[0]) - return (-1); + if (a->addr32[0] > b->addr32[0]) { + return 1; + } + if (a->addr32[0] < b->addr32[0]) { + return -1; + } break; #endif /* INET */ -#if INET6 +#ifdef INET6 case AF_INET6: - if (a->addr.addr32[3] > b->addr.addr32[3]) - return (1); - if (a->addr.addr32[3] < b->addr.addr32[3]) - return (-1); - if (a->addr.addr32[2] > b->addr.addr32[2]) - return (1); - if (a->addr.addr32[2] < b->addr.addr32[2]) - return (-1); - if (a->addr.addr32[1] > b->addr.addr32[1]) - return (1); - if (a->addr.addr32[1] < b->addr.addr32[1]) - return (-1); - if (a->addr.addr32[0] > b->addr.addr32[0]) - return (1); - if (a->addr.addr32[0] < b->addr.addr32[0]) - return (-1); + if (a->addr32[3] > b->addr32[3]) { + return 1; + } + if (a->addr32[3] < b->addr32[3]) { + return -1; + } + if (a->addr32[2] > b->addr32[2]) { + return 1; + } + if (a->addr32[2] < b->addr32[2]) { + return -1; + } + if (a->addr32[1] > b->addr32[1]) { + return 1; + } + if (a->addr32[1] < b->addr32[1]) { + return -1; + } + if (a->addr32[0] > b->addr32[0]) { + return 1; + } + if (a->addr32[0] < b->addr32[0]) { + return -1; + } break; #endif /* INET6 */ } - return (0); + return 0; +} + +static __inline int +pf_src_compare(struct pf_src_node *a, struct pf_src_node *b) +{ + int diff; + + if (a->rule.ptr > b->rule.ptr) { + return 1; + } + if (a->rule.ptr < b->rule.ptr) { + return -1; + } + if ((diff = a->af - b->af) != 0) { + return diff; + } + if ((diff = pf_addr_compare(&a->addr, &b->addr, a->af)) != 0) { + return diff; + } + return 0; } static __inline int pf_state_compare_lan_ext(struct pf_state_key *a, struct pf_state_key *b) { - int diff; - int extfilter; + int diff; + int extfilter; - if ((diff = a->proto - b->proto) != 0) - return (diff); - if ((diff = a->af - b->af) != 0) - return (diff); + if ((diff = a->proto - b->proto) != 0) { + return diff; + } + if ((diff = a->af_lan - b->af_lan) != 0) { + return diff; + } extfilter = PF_EXTFILTER_APD; switch (a->proto) { case IPPROTO_ICMP: case IPPROTO_ICMPV6: - if ((diff = a->lan.xport.port - b->lan.xport.port) != 0) - return (diff); + if ((diff = a->lan.xport.port - b->lan.xport.port) != 0) { + return diff; + } break; case IPPROTO_TCP: - if ((diff = a->lan.xport.port - b->lan.xport.port) != 0) - return (diff); - if ((diff = a->ext.xport.port - b->ext.xport.port) != 0) - return (diff); + if ((diff = a->lan.xport.port - b->lan.xport.port) != 0) { + return diff; + } + if ((diff = a->ext_lan.xport.port - b->ext_lan.xport.port) != 0) { + return diff; + } break; case IPPROTO_UDP: - if ((diff = a->proto_variant - b->proto_variant)) - return (diff); + if ((diff = a->proto_variant - b->proto_variant)) { + return diff; + } extfilter = a->proto_variant; - if ((diff = a->lan.xport.port - b->lan.xport.port) != 0) - return (diff); + if ((diff = a->lan.xport.port - b->lan.xport.port) != 0) { + return diff; + } if ((extfilter < PF_EXTFILTER_AD) && - (diff = a->ext.xport.port - b->ext.xport.port) != 0) - return (diff); + (diff = a->ext_lan.xport.port - b->ext_lan.xport.port) != 0) { + return diff; + } break; case IPPROTO_GRE: if (a->proto_variant == PF_GRE_PPTP_VARIANT && a->proto_variant == b->proto_variant) { - if (!!(diff = a->ext.xport.call_id - - b->ext.xport.call_id)) - return (diff); + if (!!(diff = a->ext_lan.xport.call_id - + b->ext_lan.xport.call_id)) { + return diff; + } } break; case IPPROTO_ESP: - if (!!(diff = a->ext.xport.spi - b->ext.xport.spi)) - return (diff); + if (!!(diff = a->ext_lan.xport.spi - b->ext_lan.xport.spi)) { + return diff; + } break; default: break; } - switch (a->af) { + switch (a->af_lan) { #if INET case AF_INET: - if (a->lan.addr.addr32[0] > b->lan.addr.addr32[0]) - return (1); - if (a->lan.addr.addr32[0] < b->lan.addr.addr32[0]) - return (-1); + if ((diff = pf_addr_compare(&a->lan.addr, &b->lan.addr, + a->af_lan)) != 0) { + return diff; + } + if (extfilter < PF_EXTFILTER_EI) { - if (a->ext.addr.addr32[0] > b->ext.addr.addr32[0]) - return (1); - if (a->ext.addr.addr32[0] < b->ext.addr.addr32[0]) - return (-1); + if ((diff = pf_addr_compare(&a->ext_lan.addr, + &b->ext_lan.addr, + a->af_lan)) != 0) { + return diff; + } } break; #endif /* INET */ #if INET6 case AF_INET6: - if (a->lan.addr.addr32[3] > b->lan.addr.addr32[3]) - return (1); - if (a->lan.addr.addr32[3] < b->lan.addr.addr32[3]) - return (-1); - if (a->lan.addr.addr32[2] > b->lan.addr.addr32[2]) - return (1); - if (a->lan.addr.addr32[2] < b->lan.addr.addr32[2]) - return (-1); - if (a->lan.addr.addr32[1] > b->lan.addr.addr32[1]) - return (1); - if (a->lan.addr.addr32[1] < b->lan.addr.addr32[1]) - return (-1); - if (a->lan.addr.addr32[0] > b->lan.addr.addr32[0]) - return (1); - if (a->lan.addr.addr32[0] < b->lan.addr.addr32[0]) - return (-1); + if ((diff = pf_addr_compare(&a->lan.addr, &b->lan.addr, + a->af_lan)) != 0) { + return diff; + } + if (extfilter < PF_EXTFILTER_EI || - !PF_AZERO(&b->ext.addr, AF_INET6)) { - if (a->ext.addr.addr32[3] > b->ext.addr.addr32[3]) - return (1); - if (a->ext.addr.addr32[3] < b->ext.addr.addr32[3]) - return (-1); - if (a->ext.addr.addr32[2] > b->ext.addr.addr32[2]) - return (1); - if (a->ext.addr.addr32[2] < b->ext.addr.addr32[2]) - return (-1); - if (a->ext.addr.addr32[1] > b->ext.addr.addr32[1]) - return (1); - if (a->ext.addr.addr32[1] < b->ext.addr.addr32[1]) - return (-1); - if (a->ext.addr.addr32[0] > b->ext.addr.addr32[0]) - return (1); - if (a->ext.addr.addr32[0] < b->ext.addr.addr32[0]) - return (-1); + !PF_AZERO(&b->ext_lan.addr, AF_INET6)) { + if ((diff = pf_addr_compare(&a->ext_lan.addr, + &b->ext_lan.addr, + a->af_lan)) != 0) { + return diff; + } } break; #endif /* INET6 */ @@ -933,126 +929,116 @@ pf_state_compare_lan_ext(struct pf_state_key *a, struct pf_state_key *b) b->app_state->compare_lan_ext) { diff = (const char *)b->app_state->compare_lan_ext - (const char *)a->app_state->compare_lan_ext; - if (diff != 0) - return (diff); + if (diff != 0) { + return diff; + } diff = a->app_state->compare_lan_ext(a->app_state, b->app_state); - if (diff != 0) - return (diff); + if (diff != 0) { + return diff; + } } } - return (0); + return 0; } static __inline int pf_state_compare_ext_gwy(struct pf_state_key *a, struct pf_state_key *b) { - int diff; - int extfilter; + int diff; + int extfilter; - if ((diff = a->proto - b->proto) != 0) - return (diff); + if ((diff = a->proto - b->proto) != 0) { + return diff; + } - if ((diff = a->af - b->af) != 0) - return (diff); + if ((diff = a->af_gwy - b->af_gwy) != 0) { + return diff; + } extfilter = PF_EXTFILTER_APD; switch (a->proto) { case IPPROTO_ICMP: case IPPROTO_ICMPV6: - if ((diff = a->gwy.xport.port - b->gwy.xport.port) != 0) - return (diff); + if ((diff = a->gwy.xport.port - b->gwy.xport.port) != 0) { + return diff; + } break; case IPPROTO_TCP: - if ((diff = a->ext.xport.port - b->ext.xport.port) != 0) - return (diff); - if ((diff = a->gwy.xport.port - b->gwy.xport.port) != 0) - return (diff); + if ((diff = a->ext_gwy.xport.port - b->ext_gwy.xport.port) != 0) { + return diff; + } + if ((diff = a->gwy.xport.port - b->gwy.xport.port) != 0) { + return diff; + } break; case IPPROTO_UDP: - if ((diff = a->proto_variant - b->proto_variant)) - return (diff); + if ((diff = a->proto_variant - b->proto_variant)) { + return diff; + } extfilter = a->proto_variant; - if ((diff = a->gwy.xport.port - b->gwy.xport.port) != 0) - return (diff); + if ((diff = a->gwy.xport.port - b->gwy.xport.port) != 0) { + return diff; + } if ((extfilter < PF_EXTFILTER_AD) && - (diff = a->ext.xport.port - b->ext.xport.port) != 0) - return (diff); + (diff = a->ext_gwy.xport.port - b->ext_gwy.xport.port) != 0) { + return diff; + } break; case IPPROTO_GRE: if (a->proto_variant == PF_GRE_PPTP_VARIANT && a->proto_variant == b->proto_variant) { if (!!(diff = a->gwy.xport.call_id - - b->gwy.xport.call_id)) - return (diff); + b->gwy.xport.call_id)) { + return diff; + } } break; case IPPROTO_ESP: - if (!!(diff = a->gwy.xport.spi - b->gwy.xport.spi)) - return (diff); + if (!!(diff = a->gwy.xport.spi - b->gwy.xport.spi)) { + return diff; + } break; default: break; } - switch (a->af) { + switch (a->af_gwy) { #if INET case AF_INET: - if (a->gwy.addr.addr32[0] > b->gwy.addr.addr32[0]) - return (1); - if (a->gwy.addr.addr32[0] < b->gwy.addr.addr32[0]) - return (-1); + if ((diff = pf_addr_compare(&a->gwy.addr, &b->gwy.addr, + a->af_gwy)) != 0) { + return diff; + } + if (extfilter < PF_EXTFILTER_EI) { - if (a->ext.addr.addr32[0] > b->ext.addr.addr32[0]) - return (1); - if (a->ext.addr.addr32[0] < b->ext.addr.addr32[0]) - return (-1); + if ((diff = pf_addr_compare(&a->ext_gwy.addr, &b->ext_gwy.addr, + a->af_gwy)) != 0) { + return diff; + } } break; #endif /* INET */ #if INET6 case AF_INET6: - if (a->gwy.addr.addr32[3] > b->gwy.addr.addr32[3]) - return (1); - if (a->gwy.addr.addr32[3] < b->gwy.addr.addr32[3]) - return (-1); - if (a->gwy.addr.addr32[2] > b->gwy.addr.addr32[2]) - return (1); - if (a->gwy.addr.addr32[2] < b->gwy.addr.addr32[2]) - return (-1); - if (a->gwy.addr.addr32[1] > b->gwy.addr.addr32[1]) - return (1); - if (a->gwy.addr.addr32[1] < b->gwy.addr.addr32[1]) - return (-1); - if (a->gwy.addr.addr32[0] > b->gwy.addr.addr32[0]) - return (1); - if (a->gwy.addr.addr32[0] < b->gwy.addr.addr32[0]) - return (-1); + if ((diff = pf_addr_compare(&a->gwy.addr, &b->gwy.addr, + a->af_gwy)) != 0) { + return diff; + } + if (extfilter < PF_EXTFILTER_EI || - !PF_AZERO(&b->ext.addr, AF_INET6)) { - if (a->ext.addr.addr32[3] > b->ext.addr.addr32[3]) - return (1); - if (a->ext.addr.addr32[3] < b->ext.addr.addr32[3]) - return (-1); - if (a->ext.addr.addr32[2] > b->ext.addr.addr32[2]) - return (1); - if (a->ext.addr.addr32[2] < b->ext.addr.addr32[2]) - return (-1); - if (a->ext.addr.addr32[1] > b->ext.addr.addr32[1]) - return (1); - if (a->ext.addr.addr32[1] < b->ext.addr.addr32[1]) - return (-1); - if (a->ext.addr.addr32[0] > b->ext.addr.addr32[0]) - return (1); - if (a->ext.addr.addr32[0] < b->ext.addr.addr32[0]) - return (-1); + !PF_AZERO(&b->ext_gwy.addr, AF_INET6)) { + if ((diff = pf_addr_compare(&a->ext_gwy.addr, &b->ext_gwy.addr, + a->af_gwy)) != 0) { + return diff; + } } break; #endif /* INET6 */ @@ -1063,31 +1049,37 @@ pf_state_compare_ext_gwy(struct pf_state_key *a, struct pf_state_key *b) b->app_state->compare_ext_gwy) { diff = (const char *)b->app_state->compare_ext_gwy - (const char *)a->app_state->compare_ext_gwy; - if (diff != 0) - return (diff); + if (diff != 0) { + return diff; + } diff = a->app_state->compare_ext_gwy(a->app_state, b->app_state); - if (diff != 0) - return (diff); + if (diff != 0) { + return diff; + } } } - return (0); + return 0; } static __inline int pf_state_compare_id(struct pf_state *a, struct pf_state *b) { - if (a->id > b->id) - return (1); - if (a->id < b->id) - return (-1); - if (a->creatorid > b->creatorid) - return (1); - if (a->creatorid < b->creatorid) - return (-1); - - return (0); + if (a->id > b->id) { + return 1; + } + if (a->id < b->id) { + return -1; + } + if (a->creatorid > b->creatorid) { + return 1; + } + if (a->creatorid < b->creatorid) { + return -1; + } + + return 0; } #if INET6 @@ -1115,15 +1107,15 @@ pf_find_state_byid(struct pf_state_cmp *key) { pf_status.fcounters[FCNT_STATE_SEARCH]++; - return (RB_FIND(pf_state_tree_id, &tree_id, - (struct pf_state *)(void *)key)); + return RB_FIND(pf_state_tree_id, &tree_id, + (struct pf_state *)(void *)key); } static struct pf_state * pf_find_state(struct pfi_kif *kif, struct pf_state_key_cmp *key, u_int dir) { - struct pf_state_key *sk = NULL; - struct pf_state *s; + struct pf_state_key *sk = NULL; + struct pf_state *s; pf_status.fcounters[FCNT_STATE_SEARCH]++; @@ -1135,25 +1127,39 @@ pf_find_state(struct pfi_kif *kif, struct pf_state_key_cmp *key, u_int dir) case PF_IN: sk = RB_FIND(pf_state_tree_ext_gwy, &pf_statetbl_ext_gwy, (struct pf_state_key *)key); + /* + * NAT64 is done only on input, for packets coming in from + * from the LAN side, need to lookup the lan_ext tree. + */ + if (sk == NULL) { + sk = RB_FIND(pf_state_tree_lan_ext, + &pf_statetbl_lan_ext, + (struct pf_state_key *)key); + if (sk && sk->af_lan == sk->af_gwy) { + sk = NULL; + } + } break; default: panic("pf_find_state"); } /* list is sorted, if-bound states before floating ones */ - if (sk != NULL) + if (sk != NULL) { TAILQ_FOREACH(s, &sk->states, next) - if (s->kif == pfi_all || s->kif == kif) - return (s); + if (s->kif == pfi_all || s->kif == kif) { + return s; + } + } - return (NULL); + return NULL; } struct pf_state * pf_find_state_all(struct pf_state_key_cmp *key, u_int dir, int *more) { - struct pf_state_key *sk = NULL; - struct pf_state *s, *ret = NULL; + struct pf_state_key *sk = NULL; + struct pf_state *s, *ret = NULL; pf_status.fcounters[FCNT_STATE_SEARCH]++; @@ -1165,6 +1171,18 @@ pf_find_state_all(struct pf_state_key_cmp *key, u_int dir, int *more) case PF_IN: sk = RB_FIND(pf_state_tree_ext_gwy, &pf_statetbl_ext_gwy, (struct pf_state_key *)key); + /* + * NAT64 is done only on input, for packets coming in from + * from the LAN side, need to lookup the lan_ext tree. + */ + if ((sk == NULL) && pf_nat64_configured) { + sk = RB_FIND(pf_state_tree_lan_ext, + &pf_statetbl_lan_ext, + (struct pf_state_key *)key); + if (sk && sk->af_lan == sk->af_gwy) { + sk = NULL; + } + } break; default: panic("pf_find_state_all"); @@ -1172,14 +1190,15 @@ pf_find_state_all(struct pf_state_key_cmp *key, u_int dir, int *more) if (sk != NULL) { ret = TAILQ_FIRST(&sk->states); - if (more == NULL) - return (ret); + if (more == NULL) { + return ret; + } TAILQ_FOREACH(s, &sk->states, next) - (*more)++; + (*more)++; } - return (ret); + return ret; } static void @@ -1197,11 +1216,12 @@ pf_add_threshold(struct pf_threshold *threshold) { u_int32_t t = pf_time_second(), diff = t - threshold->last; - if (diff >= threshold->seconds) + if (diff >= threshold->seconds) { threshold->count = 0; - else + } else { threshold->count -= threshold->count * diff / threshold->seconds; + } threshold->count += PF_THRESHOLD_MULT; threshold->last = t; } @@ -1209,14 +1229,13 @@ pf_add_threshold(struct pf_threshold *threshold) static int pf_check_threshold(struct pf_threshold *threshold) { - return (threshold->count > threshold->limit); + return threshold->count > threshold->limit; } static int pf_src_connlimit(struct pf_state **state) { int bad = 0; - (*state)->src_node->conn++; VERIFY((*state)->src_node->conn != 0); (*state)->src.tcp_est = 1; @@ -1235,33 +1254,34 @@ pf_src_connlimit(struct pf_state **state) bad++; } - if (!bad) - return (0); + if (!bad) { + return 0; + } if ((*state)->rule.ptr->overload_tbl) { struct pfr_addr p; - u_int32_t killed = 0; + u_int32_t killed = 0; pf_status.lcounters[LCNT_OVERLOAD_TABLE]++; if (pf_status.debug >= PF_DEBUG_MISC) { printf("pf_src_connlimit: blocking address "); pf_print_host(&(*state)->src_node->addr, 0, - (*state)->state_key->af); + (*state)->state_key->af_lan); } - bzero(&p, sizeof (p)); - p.pfra_af = (*state)->state_key->af; - switch ((*state)->state_key->af) { + bzero(&p, sizeof(p)); + p.pfra_af = (*state)->state_key->af_lan; + switch ((*state)->state_key->af_lan) { #if INET case AF_INET: p.pfra_net = 32; - p.pfra_ip4addr = (*state)->src_node->addr.v4; + p.pfra_ip4addr = (*state)->src_node->addr.v4addr; break; #endif /* INET */ #if INET6 case AF_INET6: p.pfra_net = 128; - p.pfra_ip6addr = (*state)->src_node->addr.v6; + p.pfra_ip6addr = (*state)->src_node->addr.v6addr; break; #endif /* INET6 */ } @@ -1282,15 +1302,15 @@ pf_src_connlimit(struct pf_state **state) * from the same rule if PF_FLUSH_GLOBAL is not * set) */ - if (sk->af == - (*state)->state_key->af && + if (sk->af_lan == + (*state)->state_key->af_lan && (((*state)->state_key->direction == - PF_OUT && + PF_OUT && PF_AEQ(&(*state)->src_node->addr, - &sk->lan.addr, sk->af)) || + &sk->lan.addr, sk->af_lan)) || ((*state)->state_key->direction == PF_IN && PF_AEQ(&(*state)->src_node->addr, - &sk->ext.addr, sk->af))) && + &sk->ext_lan.addr, sk->af_lan))) && ((*state)->rule.ptr->flush & PF_FLUSH_GLOBAL || (*state)->rule.ptr == st->rule.ptr)) { @@ -1300,45 +1320,50 @@ pf_src_connlimit(struct pf_state **state) killed++; } } - if (pf_status.debug >= PF_DEBUG_MISC) + if (pf_status.debug >= PF_DEBUG_MISC) { printf(", %u states killed", killed); + } } - if (pf_status.debug >= PF_DEBUG_MISC) + if (pf_status.debug >= PF_DEBUG_MISC) { printf("\n"); + } } /* kill this state */ (*state)->timeout = PFTM_PURGE; (*state)->src.state = (*state)->dst.state = TCPS_CLOSED; - return (1); + return 1; } int pf_insert_src_node(struct pf_src_node **sn, struct pf_rule *rule, struct pf_addr *src, sa_family_t af) { - struct pf_src_node k; + struct pf_src_node k; if (*sn == NULL) { k.af = af; PF_ACPY(&k.addr, src, af); if (rule->rule_flag & PFRULE_RULESRCTRACK || - rule->rpool.opts & PF_POOL_STICKYADDR) + rule->rpool.opts & PF_POOL_STICKYADDR) { k.rule.ptr = rule; - else + } else { k.rule.ptr = NULL; + } pf_status.scounters[SCNT_SRC_NODE_SEARCH]++; *sn = RB_FIND(pf_src_tree, &tree_src_tracking, &k); } if (*sn == NULL) { if (!rule->max_src_nodes || - rule->src_nodes < rule->max_src_nodes) + rule->src_nodes < rule->max_src_nodes) { (*sn) = pool_get(&pf_src_tree_pl, PR_WAITOK); - else + } else { pf_status.lcounters[LCNT_SRCNODES]++; - if ((*sn) == NULL) - return (-1); - bzero(*sn, sizeof (struct pf_src_node)); + } + if ((*sn) == NULL) { + return -1; + } + bzero(*sn, sizeof(struct pf_src_node)); pf_init_threshold(&(*sn)->conn_rate, rule->max_src_conn_rate.limit, @@ -1346,10 +1371,11 @@ pf_insert_src_node(struct pf_src_node **sn, struct pf_rule *rule, (*sn)->af = af; if (rule->rule_flag & PFRULE_RULESRCTRACK || - rule->rpool.opts & PF_POOL_STICKYADDR) + rule->rpool.opts & PF_POOL_STICKYADDR) { (*sn)->rule.ptr = rule; - else + } else { (*sn)->rule.ptr = NULL; + } PF_ACPY(&(*sn)->addr, src, af); if (RB_INSERT(pf_src_tree, &tree_src_tracking, *sn) != NULL) { @@ -1359,28 +1385,29 @@ pf_insert_src_node(struct pf_src_node **sn, struct pf_rule *rule, printf("\n"); } pool_put(&pf_src_tree_pl, *sn); - return (-1); + return -1; } (*sn)->creation = pf_time_second(); (*sn)->ruletype = rule->action; - if ((*sn)->rule.ptr != NULL) + if ((*sn)->rule.ptr != NULL) { (*sn)->rule.ptr->src_nodes++; + } pf_status.scounters[SCNT_SRC_NODE_INSERT]++; pf_status.src_nodes++; } else { if (rule->max_src_states && (*sn)->states >= rule->max_src_states) { pf_status.lcounters[LCNT_SRCSTATES]++; - return (-1); + return -1; } } - return (0); + return 0; } static void pf_stateins_err(const char *tree, struct pf_state *s, struct pfi_kif *kif) { - struct pf_state_key *sk = s->state_key; + struct pf_state_key *sk = s->state_key; if (pf_status.debug >= PF_DEBUG_MISC) { printf("pf: state insert failed: %s %s ", tree, kif->pfik_name); @@ -1402,16 +1429,20 @@ pf_stateins_err(const char *tree, struct pf_state *s, struct pfi_kif *kif) break; } printf(" lan: "); - pf_print_sk_host(&sk->lan, sk->af, sk->proto, + pf_print_sk_host(&sk->lan, sk->af_lan, sk->proto, sk->proto_variant); printf(" gwy: "); - pf_print_sk_host(&sk->gwy, sk->af, sk->proto, + pf_print_sk_host(&sk->gwy, sk->af_gwy, sk->proto, + sk->proto_variant); + printf(" ext_lan: "); + pf_print_sk_host(&sk->ext_lan, sk->af_lan, sk->proto, sk->proto_variant); - printf(" ext: "); - pf_print_sk_host(&sk->ext, sk->af, sk->proto, + printf(" ext_gwy: "); + pf_print_sk_host(&sk->ext_gwy, sk->af_gwy, sk->proto, sk->proto_variant); - if (s->sync_flags & PFSTATE_FROMSYNC) + if (s->sync_flags & PFSTATE_FROMSYNC) { printf(" (from sync)"); + } printf("\n"); } } @@ -1419,8 +1450,8 @@ pf_stateins_err(const char *tree, struct pf_state *s, struct pfi_kif *kif) int pf_insert_state(struct pfi_kif *kif, struct pf_state *s) { - struct pf_state_key *cur; - struct pf_state *sp; + struct pf_state_key *cur; + struct pf_state *sp; VERIFY(s->state_key != NULL); s->kif = kif; @@ -1429,13 +1460,13 @@ pf_insert_state(struct pfi_kif *kif, struct pf_state *s) s->state_key)) != NULL) { /* key exists. check for same kif, if none, add to key */ TAILQ_FOREACH(sp, &cur->states, next) - if (sp->kif == kif) { /* collision! */ - pf_stateins_err("tree_lan_ext", s, kif); - pf_detach_state(s, - PF_DT_SKIP_LANEXT|PF_DT_SKIP_EXTGWY); - return (-1); - } - pf_detach_state(s, PF_DT_SKIP_LANEXT|PF_DT_SKIP_EXTGWY); + if (sp->kif == kif) { /* collision! */ + pf_stateins_err("tree_lan_ext", s, kif); + pf_detach_state(s, + PF_DT_SKIP_LANEXT | PF_DT_SKIP_EXTGWY); + return -1; + } + pf_detach_state(s, PF_DT_SKIP_LANEXT | PF_DT_SKIP_EXTGWY); pf_attach_state(cur, s, kif == pfi_all ? 1 : 0); } @@ -1445,7 +1476,7 @@ pf_insert_state(struct pfi_kif *kif, struct pf_state *s) /* must not happen. we must have found the sk above! */ pf_stateins_err("tree_ext_gwy", s, kif); pf_detach_state(s, PF_DT_SKIP_EXTGWY); - return (-1); + return -1; } if (s->id == 0 && s->creatorid == 0) { @@ -1457,12 +1488,13 @@ pf_insert_state(struct pfi_kif *kif, struct pf_state *s) printf("pf: state insert failed: " "id: %016llx creatorid: %08x", be64toh(s->id), ntohl(s->creatorid)); - if (s->sync_flags & PFSTATE_FROMSYNC) + if (s->sync_flags & PFSTATE_FROMSYNC) { printf(" (from sync)"); + } printf("\n"); } pf_detach_state(s, 0); - return (-1); + return -1; } TAILQ_INSERT_TAIL(&state_list, s, entry_list); pf_status.fcounters[FCNT_STATE_INSERT]++; @@ -1472,7 +1504,7 @@ pf_insert_state(struct pfi_kif *kif, struct pf_state *s) #if NPFSYNC pfsync_insert_state(s); #endif - return (0); + return 0; } static int @@ -1480,7 +1512,7 @@ pf_purge_thread_cont(int err) { #pragma unused(err) static u_int32_t nloops = 0; - int t = 1; /* 1 second */ + int t = 1; /* 1 second */ /* * Update coarse-grained networking timestamp (in sec.); the idea @@ -1506,7 +1538,7 @@ pf_purge_thread_cont(int err) thread_deallocate(current_thread()); thread_terminate(current_thread()); /* NOTREACHED */ - return (0); + return 0; } else { /* if there's nothing left, sleep w/o timeout */ if (pf_status.states == 0 && @@ -1538,7 +1570,7 @@ done: /* NOTREACHED */ VERIFY(0); - return (0); + return 0; } void @@ -1557,22 +1589,24 @@ pf_purge_thread_fn(void *v, wait_result_t w) u_int64_t pf_state_expires(const struct pf_state *state) { - u_int32_t t; - u_int32_t start; - u_int32_t end; - u_int32_t states; + u_int32_t t; + u_int32_t start; + u_int32_t end; + u_int32_t states; - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); /* handle all PFTM_* > PFTM_MAX here */ - if (state->timeout == PFTM_PURGE) - return (pf_time_second()); + if (state->timeout == PFTM_PURGE) { + return pf_time_second(); + } VERIFY(state->timeout != PFTM_UNLINKED); VERIFY(state->timeout < PFTM_MAX); t = state->rule.ptr->timeout[state->timeout]; - if (!t) + if (!t) { t = pf_default_rule.timeout[state->timeout]; + } start = state->rule.ptr->timeout[PFTM_ADAPTIVE_START]; if (start) { end = state->rule.ptr->timeout[PFTM_ADAPTIVE_END]; @@ -1583,21 +1617,22 @@ pf_state_expires(const struct pf_state *state) states = pf_status.states; } if (end && states > start && start < end) { - if (states < end) - return (state->expire + t * (end - states) / - (end - start)); - else - return (pf_time_second()); + if (states < end) { + return state->expire + t * (end - states) / + (end - start); + } else { + return pf_time_second(); + } } - return (state->expire + t); + return state->expire + t; } void pf_purge_expired_src_nodes(void) { - struct pf_src_node *cur, *next; + struct pf_src_node *cur, *next; - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); for (cur = RB_MIN(pf_src_tree, &tree_src_tracking); cur; cur = next) { next = RB_NEXT(pf_src_tree, &tree_src_tracking, cur); @@ -1606,8 +1641,9 @@ pf_purge_expired_src_nodes(void) if (cur->rule.ptr != NULL) { cur->rule.ptr->src_nodes--; if (cur->rule.ptr->states <= 0 && - cur->rule.ptr->max_src_nodes <= 0) + cur->rule.ptr->max_src_nodes <= 0) { pf_rm_rule(NULL, cur->rule.ptr); + } } RB_REMOVE(pf_src_tree, &tree_src_tracking, cur); pf_status.scounters[SCNT_SRC_NODE_REMOVALS]++; @@ -1622,7 +1658,7 @@ pf_src_tree_remove_state(struct pf_state *s) { u_int32_t t; - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); if (s->src_node != NULL) { if (s->src.tcp_est) { @@ -1632,8 +1668,9 @@ pf_src_tree_remove_state(struct pf_state *s) VERIFY(s->src_node->states > 0); if (--s->src_node->states <= 0) { t = s->rule.ptr->timeout[PFTM_SRC_NODE]; - if (!t) + if (!t) { t = pf_default_rule.timeout[PFTM_SRC_NODE]; + } s->src_node->expire = pf_time_second() + t; } } @@ -1641,8 +1678,9 @@ pf_src_tree_remove_state(struct pf_state *s) VERIFY(s->nat_src_node->states > 0); if (--s->nat_src_node->states <= 0) { t = s->rule.ptr->timeout[PFTM_SRC_NODE]; - if (!t) + if (!t) { t = pf_default_rule.timeout[PFTM_SRC_NODE]; + } s->nat_src_node->expire = pf_time_second() + t; } } @@ -1652,22 +1690,23 @@ pf_src_tree_remove_state(struct pf_state *s) void pf_unlink_state(struct pf_state *cur) { - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); if (cur->src.state == PF_TCPS_PROXY_DST) { - pf_send_tcp(cur->rule.ptr, cur->state_key->af, - &cur->state_key->ext.addr, &cur->state_key->lan.addr, - cur->state_key->ext.xport.port, + pf_send_tcp(cur->rule.ptr, cur->state_key->af_lan, + &cur->state_key->ext_lan.addr, &cur->state_key->lan.addr, + cur->state_key->ext_lan.xport.port, cur->state_key->lan.xport.port, cur->src.seqhi, cur->src.seqlo + 1, - TH_RST|TH_ACK, 0, 0, 0, 1, cur->tag, NULL, NULL); + TH_RST | TH_ACK, 0, 0, 0, 1, cur->tag, NULL, NULL); } - hook_runloop(&cur->unlink_hooks, HOOK_REMOVE|HOOK_FREE); + hook_runloop(&cur->unlink_hooks, HOOK_REMOVE | HOOK_FREE); RB_REMOVE(pf_state_tree_id, &tree_id, cur); #if NPFSYNC - if (cur->creatorid == pf_status.hostid) + if (cur->creatorid == pf_status.hostid) { pfsync_delete_state(cur); + } #endif cur->timeout = PFTM_UNLINKED; pf_src_tree_remove_state(cur); @@ -1679,34 +1718,39 @@ pf_unlink_state(struct pf_state *cur) void pf_free_state(struct pf_state *cur) { - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); #if NPFSYNC if (pfsyncif != NULL && (pfsyncif->sc_bulk_send_next == cur || - pfsyncif->sc_bulk_terminator == cur)) + pfsyncif->sc_bulk_terminator == cur)) { return; + } #endif VERIFY(cur->timeout == PFTM_UNLINKED); VERIFY(cur->rule.ptr->states > 0); if (--cur->rule.ptr->states <= 0 && - cur->rule.ptr->src_nodes <= 0) + cur->rule.ptr->src_nodes <= 0) { pf_rm_rule(NULL, cur->rule.ptr); + } if (cur->nat_rule.ptr != NULL) { VERIFY(cur->nat_rule.ptr->states > 0); if (--cur->nat_rule.ptr->states <= 0 && - cur->nat_rule.ptr->src_nodes <= 0) + cur->nat_rule.ptr->src_nodes <= 0) { pf_rm_rule(NULL, cur->nat_rule.ptr); + } } if (cur->anchor.ptr != NULL) { VERIFY(cur->anchor.ptr->states > 0); - if (--cur->anchor.ptr->states <= 0) + if (--cur->anchor.ptr->states <= 0) { pf_rm_rule(NULL, cur->anchor.ptr); + } } pf_normalize_tcp_cleanup(cur); pfi_kif_unref(cur->kif, PFI_KIF_REF_STATE); TAILQ_REMOVE(&state_list, cur, entry_list); - if (cur->tag) + if (cur->tag) { pf_tag_unref(cur->tag); + } pool_put(&pf_state_pl, cur); pf_status.fcounters[FCNT_STATE_REMOVALS]++; VERIFY(pf_status.states > 0); @@ -1716,17 +1760,18 @@ pf_free_state(struct pf_state *cur) void pf_purge_expired_states(u_int32_t maxcheck) { - static struct pf_state *cur = NULL; - struct pf_state *next; + static struct pf_state *cur = NULL; + struct pf_state *next; - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); while (maxcheck--) { /* wrap to start of list when we hit the end */ if (cur == NULL) { cur = TAILQ_FIRST(&state_list); - if (cur == NULL) - break; /* list empty */ + if (cur == NULL) { + break; /* list empty */ + } } /* get next state, as cur may get deleted */ @@ -1746,22 +1791,25 @@ pf_purge_expired_states(u_int32_t maxcheck) int pf_tbladdr_setup(struct pf_ruleset *rs, struct pf_addr_wrap *aw) { - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); - if (aw->type != PF_ADDR_TABLE) - return (0); - if ((aw->p.tbl = pfr_attach_table(rs, aw->v.tblname)) == NULL) - return (1); - return (0); + if (aw->type != PF_ADDR_TABLE) { + return 0; + } + if ((aw->p.tbl = pfr_attach_table(rs, aw->v.tblname)) == NULL) { + return 1; + } + return 0; } void pf_tbladdr_remove(struct pf_addr_wrap *aw) { - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); - if (aw->type != PF_ADDR_TABLE || aw->p.tbl == NULL) + if (aw->type != PF_ADDR_TABLE || aw->p.tbl == NULL) { return; + } pfr_detach_table(aw->p.tbl); aw->p.tbl = NULL; } @@ -1771,12 +1819,14 @@ pf_tbladdr_copyout(struct pf_addr_wrap *aw) { struct pfr_ktable *kt = aw->p.tbl; - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); - if (aw->type != PF_ADDR_TABLE || kt == NULL) + if (aw->type != PF_ADDR_TABLE || kt == NULL) { return; - if (!(kt->pfrkt_flags & PFR_TFLAG_ACTIVE) && kt->pfrkt_root != NULL) + } + if (!(kt->pfrkt_flags & PFR_TFLAG_ACTIVE) && kt->pfrkt_root != NULL) { kt = kt->pfrkt_root; + } aw->p.tbl = NULL; aw->p.tblcnt = (kt->pfrkt_flags & PFR_TFLAG_ACTIVE) ? kt->pfrkt_cnt : -1; @@ -1789,8 +1839,8 @@ pf_print_addr(struct pf_addr *addr, sa_family_t af) #if INET case AF_INET: { u_int32_t a = ntohl(addr->addr32[0]); - printf("%u.%u.%u.%u", (a>>24)&255, (a>>16)&255, - (a>>8)&255, a&255); + printf("%u.%u.%u.%u", (a >> 24) & 255, (a >> 16) & 255, + (a >> 8) & 255, a & 255); break; } #endif /* INET */ @@ -1801,10 +1851,11 @@ pf_print_addr(struct pf_addr *addr, sa_family_t af) maxstart = 0, maxend = 0; for (i = 0; i < 8; i++) { if (!addr->addr16[i]) { - if (curstart == 255) + if (curstart == 255) { curstart = i; - else + } else { curend = i; + } } else { if (curstart) { if ((curend - curstart) > @@ -1819,17 +1870,20 @@ pf_print_addr(struct pf_addr *addr, sa_family_t af) for (i = 0; i < 8; i++) { if (i >= maxstart && i <= maxend) { if (maxend != 7) { - if (i == maxstart) + if (i == maxstart) { printf(":"); + } } else { - if (i == maxend) + if (i == maxend) { printf(":"); + } } } else { b = ntohs(addr->addr16[i]); printf("%x", b); - if (i < 7) + if (i < 7) { printf(":"); + } } } break; @@ -1840,19 +1894,21 @@ pf_print_addr(struct pf_addr *addr, sa_family_t af) static void pf_print_sk_host(struct pf_state_host *sh, sa_family_t af, int proto, - u_int8_t proto_variant) + u_int8_t proto_variant) { pf_print_addr(&sh->addr, af); switch (proto) { case IPPROTO_ESP: - if (sh->xport.spi) + if (sh->xport.spi) { printf("[%08x]", ntohl(sh->xport.spi)); + } break; case IPPROTO_GRE: - if (proto_variant == PF_GRE_PPTP_VARIANT) + if (proto_variant == PF_GRE_PPTP_VARIANT) { printf("[%u]", ntohs(sh->xport.call_id)); + } break; case IPPROTO_TCP: @@ -1869,8 +1925,9 @@ static void pf_print_host(struct pf_addr *addr, u_int16_t p, sa_family_t af) { pf_print_addr(addr, af); - if (p) + if (p) { printf("[%u]", ntohs(p)); + } } void @@ -1900,20 +1957,26 @@ pf_print_state(struct pf_state *s) printf("%u ", sk->proto); break; } - pf_print_sk_host(&sk->lan, sk->af, sk->proto, sk->proto_variant); + pf_print_sk_host(&sk->lan, sk->af_lan, sk->proto, sk->proto_variant); + printf(" "); + pf_print_sk_host(&sk->gwy, sk->af_gwy, sk->proto, sk->proto_variant); printf(" "); - pf_print_sk_host(&sk->gwy, sk->af, sk->proto, sk->proto_variant); + pf_print_sk_host(&sk->ext_lan, sk->af_lan, sk->proto, + sk->proto_variant); printf(" "); - pf_print_sk_host(&sk->ext, sk->af, sk->proto, sk->proto_variant); + pf_print_sk_host(&sk->ext_gwy, sk->af_gwy, sk->proto, + sk->proto_variant); printf(" [lo=%u high=%u win=%u modulator=%u", s->src.seqlo, s->src.seqhi, s->src.max_win, s->src.seqdiff); - if (s->src.wscale && s->dst.wscale) + if (s->src.wscale && s->dst.wscale) { printf(" wscale=%u", s->src.wscale & PF_WSCALE_MASK); + } printf("]"); printf(" [lo=%u high=%u win=%u modulator=%u", s->dst.seqlo, s->dst.seqhi, s->dst.max_win, s->dst.seqdiff); - if (s->src.wscale && s->dst.wscale) + if (s->src.wscale && s->dst.wscale) { printf(" wscale=%u", s->dst.wscale & PF_WSCALE_MASK); + } printf("]"); printf(" %u:%u", s->src.state, s->dst.state); } @@ -1921,32 +1984,41 @@ pf_print_state(struct pf_state *s) void pf_print_flags(u_int8_t f) { - if (f) + if (f) { printf(" "); - if (f & TH_FIN) + } + if (f & TH_FIN) { printf("F"); - if (f & TH_SYN) + } + if (f & TH_SYN) { printf("S"); - if (f & TH_RST) + } + if (f & TH_RST) { printf("R"); - if (f & TH_PUSH) + } + if (f & TH_PUSH) { printf("P"); - if (f & TH_ACK) + } + if (f & TH_ACK) { printf("A"); - if (f & TH_URG) + } + if (f & TH_URG) { printf("U"); - if (f & TH_ECE) + } + if (f & TH_ECE) { printf("E"); - if (f & TH_CWR) + } + if (f & TH_CWR) { printf("W"); + } } -#define PF_SET_SKIP_STEPS(i) \ - do { \ - while (head[i] != cur) { \ - head[i]->skip[i].ptr = cur; \ - head[i] = TAILQ_NEXT(head[i], entries); \ - } \ +#define PF_SET_SKIP_STEPS(i) \ + do { \ + while (head[i] != cur) { \ + head[i]->skip[i].ptr = cur; \ + head[i] = TAILQ_NEXT(head[i], entries); \ + } \ } while (0) void @@ -1957,21 +2029,26 @@ pf_calc_skip_steps(struct pf_rulequeue *rules) cur = TAILQ_FIRST(rules); prev = cur; - for (i = 0; i < PF_SKIP_COUNT; ++i) + for (i = 0; i < PF_SKIP_COUNT; ++i) { head[i] = cur; + } while (cur != NULL) { - - if (cur->kif != prev->kif || cur->ifnot != prev->ifnot) + if (cur->kif != prev->kif || cur->ifnot != prev->ifnot) { PF_SET_SKIP_STEPS(PF_SKIP_IFP); - if (cur->direction != prev->direction) + } + if (cur->direction != prev->direction) { PF_SET_SKIP_STEPS(PF_SKIP_DIR); - if (cur->af != prev->af) + } + if (cur->af != prev->af) { PF_SET_SKIP_STEPS(PF_SKIP_AF); - if (cur->proto != prev->proto) + } + if (cur->proto != prev->proto) { PF_SET_SKIP_STEPS(PF_SKIP_PROTO); + } if (cur->src.neg != prev->src.neg || - pf_addr_wrap_neq(&cur->src.addr, &prev->src.addr)) + pf_addr_wrap_neq(&cur->src.addr, &prev->src.addr)) { PF_SET_SKIP_STEPS(PF_SKIP_SRC_ADDR); + } { union pf_rule_xport *cx = &cur->src.xport; union pf_rule_xport *px = &prev->src.xport; @@ -1986,14 +2063,16 @@ pf_calc_skip_steps(struct pf_rulequeue *rules) prev->proto == IPPROTO_ESP || cx->range.op != px->range.op || cx->range.port[0] != px->range.port[0] || - cx->range.port[1] != px->range.port[1]) + cx->range.port[1] != px->range.port[1]) { PF_SET_SKIP_STEPS(PF_SKIP_SRC_PORT); + } break; } } if (cur->dst.neg != prev->dst.neg || - pf_addr_wrap_neq(&cur->dst.addr, &prev->dst.addr)) + pf_addr_wrap_neq(&cur->dst.addr, &prev->dst.addr)) { PF_SET_SKIP_STEPS(PF_SKIP_DST_ADDR); + } { union pf_rule_xport *cx = &cur->dst.xport; union pf_rule_xport *px = &prev->dst.xport; @@ -2001,21 +2080,24 @@ pf_calc_skip_steps(struct pf_rulequeue *rules) switch (cur->proto) { case IPPROTO_GRE: if (cur->proto != prev->proto || - cx->call_id != px->call_id) + cx->call_id != px->call_id) { PF_SET_SKIP_STEPS(PF_SKIP_DST_PORT); + } break; case IPPROTO_ESP: if (cur->proto != prev->proto || - cx->spi != px->spi) + cx->spi != px->spi) { PF_SET_SKIP_STEPS(PF_SKIP_DST_PORT); + } break; default: if (prev->proto == IPPROTO_GRE || prev->proto == IPPROTO_ESP || cx->range.op != px->range.op || cx->range.port[0] != px->range.port[0] || - cx->range.port[1] != px->range.port[1]) + cx->range.port[1] != px->range.port[1]) { PF_SET_SKIP_STEPS(PF_SKIP_DST_PORT); + } break; } } @@ -2023,8 +2105,9 @@ pf_calc_skip_steps(struct pf_rulequeue *rules) prev = cur; cur = TAILQ_NEXT(cur, entries); } - for (i = 0; i < PF_SKIP_COUNT; ++i) + for (i = 0; i < PF_SKIP_COUNT; ++i) { PF_SET_SKIP_STEPS(i); + } } u_int32_t @@ -2033,157 +2116,224 @@ pf_calc_state_key_flowhash(struct pf_state_key *sk) struct pf_flowhash_key fh __attribute__((aligned(8))); uint32_t flowhash = 0; - bzero(&fh, sizeof (fh)); - if (PF_ALEQ(&sk->lan.addr, &sk->ext.addr, sk->af)) { - bcopy(&sk->lan.addr, &fh.ap1.addr, sizeof (fh.ap1.addr)); - bcopy(&sk->ext.addr, &fh.ap2.addr, sizeof (fh.ap2.addr)); + bzero(&fh, sizeof(fh)); + if (PF_ALEQ(&sk->lan.addr, &sk->ext_lan.addr, sk->af_lan)) { + bcopy(&sk->lan.addr, &fh.ap1.addr, sizeof(fh.ap1.addr)); + bcopy(&sk->ext_lan.addr, &fh.ap2.addr, sizeof(fh.ap2.addr)); } else { - bcopy(&sk->ext.addr, &fh.ap1.addr, sizeof (fh.ap1.addr)); - bcopy(&sk->lan.addr, &fh.ap2.addr, sizeof (fh.ap2.addr)); + bcopy(&sk->ext_lan.addr, &fh.ap1.addr, sizeof(fh.ap1.addr)); + bcopy(&sk->lan.addr, &fh.ap2.addr, sizeof(fh.ap2.addr)); } - if (sk->lan.xport.spi <= sk->ext.xport.spi) { + if (sk->lan.xport.spi <= sk->ext_lan.xport.spi) { fh.ap1.xport.spi = sk->lan.xport.spi; - fh.ap2.xport.spi = sk->ext.xport.spi; + fh.ap2.xport.spi = sk->ext_lan.xport.spi; } else { - fh.ap1.xport.spi = sk->ext.xport.spi; + fh.ap1.xport.spi = sk->ext_lan.xport.spi; fh.ap2.xport.spi = sk->lan.xport.spi; } - fh.af = sk->af; + fh.af = sk->af_lan; fh.proto = sk->proto; try_again: - flowhash = net_flowhash(&fh, sizeof (fh), pf_hash_seed); + flowhash = net_flowhash(&fh, sizeof(fh), pf_hash_seed); if (flowhash == 0) { /* try to get a non-zero flowhash */ pf_hash_seed = RandomULong(); goto try_again; } - return (flowhash); + return flowhash; } static int pf_addr_wrap_neq(struct pf_addr_wrap *aw1, struct pf_addr_wrap *aw2) { - if (aw1->type != aw2->type) - return (1); + if (aw1->type != aw2->type) { + return 1; + } switch (aw1->type) { case PF_ADDR_ADDRMASK: case PF_ADDR_RANGE: - if (PF_ANEQ(&aw1->v.a.addr, &aw2->v.a.addr, 0)) - return (1); - if (PF_ANEQ(&aw1->v.a.mask, &aw2->v.a.mask, 0)) - return (1); - return (0); + if (PF_ANEQ(&aw1->v.a.addr, &aw2->v.a.addr, 0)) { + return 1; + } + if (PF_ANEQ(&aw1->v.a.mask, &aw2->v.a.mask, 0)) { + return 1; + } + return 0; case PF_ADDR_DYNIFTL: - return (aw1->p.dyn == NULL || aw2->p.dyn == NULL || - aw1->p.dyn->pfid_kt != aw2->p.dyn->pfid_kt); + return aw1->p.dyn == NULL || aw2->p.dyn == NULL || + aw1->p.dyn->pfid_kt != aw2->p.dyn->pfid_kt; case PF_ADDR_NOROUTE: case PF_ADDR_URPFFAILED: - return (0); + return 0; case PF_ADDR_TABLE: - return (aw1->p.tbl != aw2->p.tbl); + return aw1->p.tbl != aw2->p.tbl; case PF_ADDR_RTLABEL: - return (aw1->v.rtlabel != aw2->v.rtlabel); + return aw1->v.rtlabel != aw2->v.rtlabel; default: printf("invalid address type: %d\n", aw1->type); - return (1); + return 1; } } u_int16_t pf_cksum_fixup(u_int16_t cksum, u_int16_t old, u_int16_t new, u_int8_t udp) { - u_int32_t l; - - if (udp && !cksum) - return (0); - l = cksum + old - new; - l = (l >> 16) + (l & 0xffff); - l = l & 0xffff; - if (udp && !l) - return (0xffff); - return (l); + return nat464_cksum_fixup(cksum, old, new, udp); } +/* + * change ip address & port + * dir : packet direction + * a : address to be changed + * p : port to be changed + * ic : ip header checksum + * pc : protocol checksum + * an : new ip address + * pn : new port + * u : should be 1 if UDP packet else 0 + * af : address family of the packet + * afn : address family of the new address + * ua : should be 1 if ip address needs to be updated in the packet else + * only the checksum is recalculated & updated. + */ static void -pf_change_ap(int dir, struct mbuf *m, struct pf_addr *a, u_int16_t *p, +pf_change_ap(int dir, pbuf_t *pbuf, struct pf_addr *a, u_int16_t *p, u_int16_t *ic, u_int16_t *pc, struct pf_addr *an, u_int16_t pn, - u_int8_t u, sa_family_t af) + u_int8_t u, sa_family_t af, sa_family_t afn, int ua) { - struct pf_addr ao; - u_int16_t po = *p; + struct pf_addr ao; + u_int16_t po = *p; PF_ACPY(&ao, a, af); - PF_ACPY(a, an, af); + if (ua) { + PF_ACPY(a, an, afn); + } *p = pn; switch (af) { #if INET case AF_INET: - *ic = pf_cksum_fixup(pf_cksum_fixup(*ic, - ao.addr16[0], an->addr16[0], 0), - ao.addr16[1], an->addr16[1], 0); - *p = pn; - /* - * If the packet is originated from an ALG on the NAT gateway - * (source address is loopback or local), in which case the - * TCP/UDP checksum field contains the pseudo header checksum - * that's not yet complemented. - */ - if (dir == PF_OUT && m != NULL && - (m->m_flags & M_PKTHDR) && - (m->m_pkthdr.csum_flags & (CSUM_TCP | CSUM_UDP))) { - /* Pseudo-header checksum does not include ports */ - *pc = ~pf_cksum_fixup(pf_cksum_fixup(~*pc, - ao.addr16[0], an->addr16[0], u), - ao.addr16[1], an->addr16[1], u); - } else { - *pc = pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup(*pc, - ao.addr16[0], an->addr16[0], u), - ao.addr16[1], an->addr16[1], u), + switch (afn) { + case AF_INET: + *ic = pf_cksum_fixup(pf_cksum_fixup(*ic, + ao.addr16[0], an->addr16[0], 0), + ao.addr16[1], an->addr16[1], 0); + *p = pn; + /* + * If the packet is originated from an ALG on the NAT gateway + * (source address is loopback or local), in which case the + * TCP/UDP checksum field contains the pseudo header checksum + * that's not yet complemented. + * In that case we do not need to fixup the checksum for port + * translation as the pseudo header checksum doesn't include ports. + * + * A packet generated locally will have UDP/TCP CSUM flag + * set (gets set in protocol output). + * + * It should be noted that the fixup doesn't do anything if the + * checksum is 0. + */ + if (dir == PF_OUT && pbuf != NULL && + (*pbuf->pb_csum_flags & (CSUM_TCP | CSUM_UDP))) { + /* Pseudo-header checksum does not include ports */ + *pc = ~pf_cksum_fixup(pf_cksum_fixup(~*pc, + ao.addr16[0], an->addr16[0], u), + ao.addr16[1], an->addr16[1], u); + } else { + *pc = + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + *pc, ao.addr16[0], an->addr16[0], u), + ao.addr16[1], an->addr16[1], u), + po, pn, u); + } + break; +#ifdef INET6 + case AF_INET6: + *p = pn; + *pc = pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup(*pc, + ao.addr16[0], an->addr16[0], u), + ao.addr16[1], an->addr16[1], u), + 0, an->addr16[2], u), + 0, an->addr16[3], u), + 0, an->addr16[4], u), + 0, an->addr16[5], u), + 0, an->addr16[6], u), + 0, an->addr16[7], u), po, pn, u); + break; +#endif /* INET6 */ } break; #endif /* INET */ #if INET6 case AF_INET6: - /* - * If the packet is originated from an ALG on the NAT gateway - * (source address is loopback or local), in which case the - * TCP/UDP checksum field contains the pseudo header checksum - * that's not yet complemented. - */ - if (dir == PF_OUT && m != NULL && - (m->m_flags & M_PKTHDR) && - (m->m_pkthdr.csum_flags & (CSUM_TCPIPV6 | CSUM_UDPIPV6))) { - /* Pseudo-header checksum does not include ports */ - *pc = ~pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( - pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( - pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup(~*pc, - ao.addr16[0], an->addr16[0], u), - ao.addr16[1], an->addr16[1], u), - ao.addr16[2], an->addr16[2], u), - ao.addr16[3], an->addr16[3], u), - ao.addr16[4], an->addr16[4], u), - ao.addr16[5], an->addr16[5], u), - ao.addr16[6], an->addr16[6], u), - ao.addr16[7], an->addr16[7], u), - po, pn, u); - } else { + switch (afn) { + case AF_INET6: + /* + * If the packet is originated from an ALG on the NAT gateway + * (source address is loopback or local), in which case the + * TCP/UDP checksum field contains the pseudo header checksum + * that's not yet complemented. + * A packet generated locally + * will have UDP/TCP CSUM flag set (gets set in protocol + * output). + */ + if (dir == PF_OUT && pbuf != NULL && + (*pbuf->pb_csum_flags & (CSUM_TCPIPV6 | + CSUM_UDPIPV6))) { + /* Pseudo-header checksum does not include ports */ + *pc = + ~pf_cksum_fixup(pf_cksum_fixup( + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + ~*pc, + ao.addr16[0], an->addr16[0], u), + ao.addr16[1], an->addr16[1], u), + ao.addr16[2], an->addr16[2], u), + ao.addr16[3], an->addr16[3], u), + ao.addr16[4], an->addr16[4], u), + ao.addr16[5], an->addr16[5], u), + ao.addr16[6], an->addr16[6], u), + ao.addr16[7], an->addr16[7], u); + } else { + *pc = + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + *pc, + ao.addr16[0], an->addr16[0], u), + ao.addr16[1], an->addr16[1], u), + ao.addr16[2], an->addr16[2], u), + ao.addr16[3], an->addr16[3], u), + ao.addr16[4], an->addr16[4], u), + ao.addr16[5], an->addr16[5], u), + ao.addr16[6], an->addr16[6], u), + ao.addr16[7], an->addr16[7], u), + po, pn, u); + } + break; +#ifdef INET + case AF_INET: *pc = pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( - pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( - pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup(*pc, - ao.addr16[0], an->addr16[0], u), - ao.addr16[1], an->addr16[1], u), - ao.addr16[2], an->addr16[2], u), - ao.addr16[3], an->addr16[3], u), - ao.addr16[4], an->addr16[4], u), - ao.addr16[5], an->addr16[5], u), - ao.addr16[6], an->addr16[6], u), - ao.addr16[7], an->addr16[7], u), - po, pn, u); + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup(*pc, + ao.addr16[0], an->addr16[0], u), + ao.addr16[1], an->addr16[1], u), + ao.addr16[2], 0, u), + ao.addr16[3], 0, u), + ao.addr16[4], 0, u), + ao.addr16[5], 0, u), + ao.addr16[6], 0, u), + ao.addr16[7], 0, u), + po, pn, u); + break; +#endif /* INET */ } break; #endif /* INET6 */ @@ -2195,10 +2345,10 @@ pf_change_ap(int dir, struct mbuf *m, struct pf_addr *a, u_int16_t *p, void pf_change_a(void *a, u_int16_t *c, u_int32_t an, u_int8_t u) { - u_int32_t ao; + u_int32_t ao; - memcpy(&ao, a, sizeof (ao)); - memcpy(a, &an, sizeof (u_int32_t)); + memcpy(&ao, a, sizeof(ao)); + memcpy(a, &an, sizeof(u_int32_t)); *c = pf_cksum_fixup(pf_cksum_fixup(*c, ao / 65536, an / 65536, u), ao % 65536, an % 65536, u); } @@ -2207,23 +2357,79 @@ pf_change_a(void *a, u_int16_t *c, u_int32_t an, u_int8_t u) static void pf_change_a6(struct pf_addr *a, u_int16_t *c, struct pf_addr *an, u_int8_t u) { - struct pf_addr ao; + struct pf_addr ao; PF_ACPY(&ao, a, AF_INET6); PF_ACPY(a, an, AF_INET6); *c = pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( - pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( - pf_cksum_fixup(pf_cksum_fixup(*c, - ao.addr16[0], an->addr16[0], u), - ao.addr16[1], an->addr16[1], u), - ao.addr16[2], an->addr16[2], u), - ao.addr16[3], an->addr16[3], u), - ao.addr16[4], an->addr16[4], u), - ao.addr16[5], an->addr16[5], u), + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + pf_cksum_fixup(pf_cksum_fixup(*c, + ao.addr16[0], an->addr16[0], u), + ao.addr16[1], an->addr16[1], u), + ao.addr16[2], an->addr16[2], u), + ao.addr16[3], an->addr16[3], u), + ao.addr16[4], an->addr16[4], u), + ao.addr16[5], an->addr16[5], u), ao.addr16[6], an->addr16[6], u), ao.addr16[7], an->addr16[7], u); } + +void +pf_change_addr(struct pf_addr *a, u_int16_t *c, struct pf_addr *an, u_int8_t u, + sa_family_t af, sa_family_t afn) +{ + struct pf_addr ao; + + if (af != afn) { + PF_ACPY(&ao, a, af); + PF_ACPY(a, an, afn); + } + + switch (af) { + case AF_INET: + switch (afn) { + case AF_INET: + pf_change_a(a, c, an->v4addr.s_addr, u); + break; + case AF_INET6: + *c = pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + pf_cksum_fixup(pf_cksum_fixup(*c, + ao.addr16[0], an->addr16[0], u), + ao.addr16[1], an->addr16[1], u), + 0, an->addr16[2], u), + 0, an->addr16[3], u), + 0, an->addr16[4], u), + 0, an->addr16[5], u), + 0, an->addr16[6], u), + 0, an->addr16[7], u); + break; + } + break; + case AF_INET6: + switch (afn) { + case AF_INET: + *c = pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + pf_cksum_fixup(pf_cksum_fixup(*c, + ao.addr16[0], an->addr16[0], u), + ao.addr16[1], an->addr16[1], u), + ao.addr16[2], 0, u), + ao.addr16[3], 0, u), + ao.addr16[4], 0, u), + ao.addr16[5], 0, u), + ao.addr16[6], 0, u), + ao.addr16[7], 0, u); + break; + case AF_INET6: + pf_change_a6(a, c, an, u); + break; + } + break; + } +} + #endif /* INET6 */ static void @@ -2231,31 +2437,34 @@ pf_change_icmp(struct pf_addr *ia, u_int16_t *ip, struct pf_addr *oa, struct pf_addr *na, u_int16_t np, u_int16_t *pc, u_int16_t *h2c, u_int16_t *ic, u_int16_t *hc, u_int8_t u, sa_family_t af) { - struct pf_addr oia, ooa; + struct pf_addr oia, ooa; PF_ACPY(&oia, ia, af); PF_ACPY(&ooa, oa, af); /* Change inner protocol port, fix inner protocol checksum. */ if (ip != NULL) { - u_int16_t oip = *ip; - u_int32_t opc = 0; + u_int16_t oip = *ip; + u_int32_t opc = 0; - if (pc != NULL) + if (pc != NULL) { opc = *pc; + } *ip = np; - if (pc != NULL) + if (pc != NULL) { *pc = pf_cksum_fixup(*pc, oip, *ip, u); + } *ic = pf_cksum_fixup(*ic, oip, *ip, 0); - if (pc != NULL) + if (pc != NULL) { *ic = pf_cksum_fixup(*ic, opc, *pc, 0); + } } /* Change inner ip address, fix inner ip and icmp checksums. */ PF_ACPY(ia, na, af); switch (af) { #if INET case AF_INET: { - u_int32_t oh2c = *h2c; + u_int32_t oh2c = *h2c; *h2c = pf_cksum_fixup(pf_cksum_fixup(*h2c, oia.addr16[0], ia->addr16[0], 0), @@ -2270,14 +2479,14 @@ pf_change_icmp(struct pf_addr *ia, u_int16_t *ip, struct pf_addr *oa, #if INET6 case AF_INET6: *ic = pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( - pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( - pf_cksum_fixup(pf_cksum_fixup(*ic, - oia.addr16[0], ia->addr16[0], u), - oia.addr16[1], ia->addr16[1], u), - oia.addr16[2], ia->addr16[2], u), - oia.addr16[3], ia->addr16[3], u), - oia.addr16[4], ia->addr16[4], u), - oia.addr16[5], ia->addr16[5], u), + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + pf_cksum_fixup(pf_cksum_fixup(*ic, + oia.addr16[0], ia->addr16[0], u), + oia.addr16[1], ia->addr16[1], u), + oia.addr16[2], ia->addr16[2], u), + oia.addr16[3], ia->addr16[3], u), + oia.addr16[4], ia->addr16[4], u), + oia.addr16[5], ia->addr16[5], u), oia.addr16[6], ia->addr16[6], u), oia.addr16[7], ia->addr16[7], u); break; @@ -2296,14 +2505,14 @@ pf_change_icmp(struct pf_addr *ia, u_int16_t *ip, struct pf_addr *oa, #if INET6 case AF_INET6: *ic = pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( - pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( - pf_cksum_fixup(pf_cksum_fixup(*ic, - ooa.addr16[0], oa->addr16[0], u), - ooa.addr16[1], oa->addr16[1], u), - ooa.addr16[2], oa->addr16[2], u), - ooa.addr16[3], oa->addr16[3], u), - ooa.addr16[4], oa->addr16[4], u), - ooa.addr16[5], oa->addr16[5], u), + pf_cksum_fixup(pf_cksum_fixup(pf_cksum_fixup( + pf_cksum_fixup(pf_cksum_fixup(*ic, + ooa.addr16[0], oa->addr16[0], u), + ooa.addr16[1], oa->addr16[1], u), + ooa.addr16[2], oa->addr16[2], u), + ooa.addr16[3], oa->addr16[3], u), + ooa.addr16[4], oa->addr16[4], u), + ooa.addr16[5], oa->addr16[5], u), ooa.addr16[6], oa->addr16[6], u), ooa.addr16[7], oa->addr16[7], u); break; @@ -2317,62 +2526,77 @@ pf_change_icmp(struct pf_addr *ia, u_int16_t *ip, struct pf_addr *oa, * (credits to Krzysztof Pfaff for report and patch) */ static int -pf_modulate_sack(struct mbuf *m, int off, struct pf_pdesc *pd, +pf_modulate_sack(pbuf_t *pbuf, int off, struct pf_pdesc *pd, struct tcphdr *th, struct pf_state_peer *dst) { - int hlen = (th->th_off << 2) - sizeof (*th), thoptlen = hlen; + int hlen = (th->th_off << 2) - sizeof(*th), thoptlen = hlen; u_int8_t opts[MAX_TCPOPTLEN], *opt = opts; int copyback = 0, i, olen; struct sackblk sack; -#define TCPOLEN_SACKLEN (TCPOLEN_SACK + 2) +#define TCPOLEN_SACKLEN (TCPOLEN_SACK + 2) if (hlen < TCPOLEN_SACKLEN || - !pf_pull_hdr(m, off + sizeof (*th), opts, hlen, NULL, NULL, pd->af)) - return (0); + !pf_pull_hdr(pbuf, off + sizeof(*th), opts, hlen, NULL, NULL, pd->af)) { + return 0; + } while (hlen >= TCPOLEN_SACKLEN) { olen = opt[1]; switch (*opt) { - case TCPOPT_EOL: /* FALLTHROUGH */ + case TCPOPT_EOL: /* FALLTHROUGH */ case TCPOPT_NOP: opt++; hlen--; break; case TCPOPT_SACK: - if (olen > hlen) + if (olen > hlen) { olen = hlen; + } if (olen >= TCPOLEN_SACKLEN) { for (i = 2; i + TCPOLEN_SACK <= olen; i += TCPOLEN_SACK) { - memcpy(&sack, &opt[i], sizeof (sack)); + memcpy(&sack, &opt[i], sizeof(sack)); pf_change_a(&sack.start, &th->th_sum, htonl(ntohl(sack.start) - dst->seqdiff), 0); pf_change_a(&sack.end, &th->th_sum, htonl(ntohl(sack.end) - dst->seqdiff), 0); - memcpy(&opt[i], &sack, sizeof (sack)); + memcpy(&opt[i], &sack, sizeof(sack)); } - copyback = off + sizeof (*th) + thoptlen; + copyback = off + sizeof(*th) + thoptlen; } - /* FALLTHROUGH */ + /* FALLTHROUGH */ default: - if (olen < 2) + if (olen < 2) { olen = 2; + } hlen -= olen; opt += olen; } } if (copyback) { - m = pf_lazy_makewritable(pd, m, copyback); - if (!m) - return (-1); - m_copyback(m, off + sizeof (*th), thoptlen, opts); + if (pf_lazy_makewritable(pd, pbuf, copyback) == NULL) { + return -1; + } + pbuf_copy_back(pbuf, off + sizeof(*th), thoptlen, opts); } - return (copyback); + return copyback; } +/* + * XXX + * + * The following functions (pf_send_tcp and pf_send_icmp) are somewhat + * special in that they originate "spurious" packets rather than + * filter/NAT existing packets. As such, they're not a great fit for + * the 'pbuf' shim, which assumes the underlying packet buffers are + * allocated elsewhere. + * + * Since these functions are rarely used, we'll carry on allocating mbufs + * and passing them to the IP stack for eventual routing. + */ static void pf_send_tcp(const struct pf_rule *r, sa_family_t af, const struct pf_addr *saddr, const struct pf_addr *daddr, @@ -2381,32 +2605,33 @@ pf_send_tcp(const struct pf_rule *r, sa_family_t af, u_int16_t rtag, struct ether_header *eh, struct ifnet *ifp) { #pragma unused(eh, ifp) - struct mbuf *m; - int len, tlen; + struct mbuf *m; + int len, tlen; #if INET - struct ip *h = NULL; + struct ip *h = NULL; #endif /* INET */ #if INET6 - struct ip6_hdr *h6 = NULL; + struct ip6_hdr *h6 = NULL; #endif /* INET6 */ - struct tcphdr *th = NULL; - char *opt; - struct pf_mtag *pf_mtag; + struct tcphdr *th = NULL; + char *opt; + struct pf_mtag *pf_mtag; /* maximum segment size tcp option */ - tlen = sizeof (struct tcphdr); - if (mss) + tlen = sizeof(struct tcphdr); + if (mss) { tlen += 4; + } switch (af) { #if INET case AF_INET: - len = sizeof (struct ip) + tlen; + len = sizeof(struct ip) + tlen; break; #endif /* INET */ #if INET6 case AF_INET6: - len = sizeof (struct ip6_hdr) + tlen; + len = sizeof(struct ip6_hdr) + tlen; break; #endif /* INET6 */ default: @@ -2416,25 +2641,22 @@ pf_send_tcp(const struct pf_rule *r, sa_family_t af, /* create outgoing mbuf */ m = m_gethdr(M_DONTWAIT, MT_HEADER); - if (m == NULL) + if (m == NULL) { return; + } if ((pf_mtag = pf_get_mtag(m)) == NULL) { - m_free(m); return; } - if (tag) + if (tag) { pf_mtag->pftag_flags |= PF_TAG_GENERATED; + } pf_mtag->pftag_tag = rtag; - if (r != NULL && PF_RTABLEID_IS_VALID(r->rtableid)) + if (r != NULL && PF_RTABLEID_IS_VALID(r->rtableid)) { pf_mtag->pftag_rtableid = r->rtableid; - -#if PF_ALTQ - if (altq_allowed && r != NULL && r->qid) - pf_mtag->pftag_qid = r->qid; -#endif /* PF_ALTQ */ + } #if PF_ECN /* add hints for ecn */ @@ -2471,10 +2693,10 @@ pf_send_tcp(const struct pf_rule *r, sa_family_t af, /* IP header fields included in the TCP checksum */ h->ip_p = IPPROTO_TCP; h->ip_len = htons(tlen); - h->ip_src.s_addr = saddr->v4.s_addr; - h->ip_dst.s_addr = daddr->v4.s_addr; + h->ip_src.s_addr = saddr->v4addr.s_addr; + h->ip_dst.s_addr = daddr->v4addr.s_addr; - th = (struct tcphdr *)(void *)((caddr_t)h + sizeof (struct ip)); + th = (struct tcphdr *)(void *)((caddr_t)h + sizeof(struct ip)); break; #endif /* INET */ #if INET6 @@ -2484,11 +2706,11 @@ pf_send_tcp(const struct pf_rule *r, sa_family_t af, /* IP header fields included in the TCP checksum */ h6->ip6_nxt = IPPROTO_TCP; h6->ip6_plen = htons(tlen); - memcpy(&h6->ip6_src, &saddr->v6, sizeof (struct in6_addr)); - memcpy(&h6->ip6_dst, &daddr->v6, sizeof (struct in6_addr)); + memcpy(&h6->ip6_src, &saddr->v6addr, sizeof(struct in6_addr)); + memcpy(&h6->ip6_dst, &daddr->v6addr, sizeof(struct in6_addr)); th = (struct tcphdr *)(void *) - ((caddr_t)h6 + sizeof (struct ip6_hdr)); + ((caddr_t)h6 + sizeof(struct ip6_hdr)); break; #endif /* INET6 */ } @@ -2522,7 +2744,7 @@ pf_send_tcp(const struct pf_rule *r, sa_family_t af, /* Finish the IP header */ h->ip_v = 4; - h->ip_hl = sizeof (*h) >> 2; + h->ip_hl = sizeof(*h) >> 2; h->ip_tos = IPTOS_LOWDELAY; /* * ip_output() expects ip_len and ip_off to be in host order. @@ -2532,7 +2754,7 @@ pf_send_tcp(const struct pf_rule *r, sa_family_t af, h->ip_ttl = ttl ? ttl : ip_defttl; h->ip_sum = 0; - bzero(&ro, sizeof (ro)); + bzero(&ro, sizeof(ro)); ip_output(m, NULL, &ro, 0, NULL, NULL); ROUTE_RELEASE(&ro); break; @@ -2544,12 +2766,12 @@ pf_send_tcp(const struct pf_rule *r, sa_family_t af, /* TCP checksum */ th->th_sum = in6_cksum(m, IPPROTO_TCP, - sizeof (struct ip6_hdr), tlen); + sizeof(struct ip6_hdr), tlen); h6->ip6_vfc |= IPV6_VERSION; h6->ip6_hlim = IPV6_DEFHLIM; - bzero(&ro6, sizeof (ro6)); + bzero(&ro6, sizeof(ro6)); ip6_output(m, NULL, &ro6, 0, NULL, NULL, NULL); ROUTE_RELEASE(&ro6); break; @@ -2559,28 +2781,26 @@ pf_send_tcp(const struct pf_rule *r, sa_family_t af, } static void -pf_send_icmp(struct mbuf *m, u_int8_t type, u_int8_t code, sa_family_t af, +pf_send_icmp(pbuf_t *pbuf, u_int8_t type, u_int8_t code, sa_family_t af, struct pf_rule *r) { - struct mbuf *m0; - struct pf_mtag *pf_mtag; + struct mbuf *m0; + struct pf_mtag *pf_mtag; - m0 = m_copy(m, 0, M_COPYALL); - if (m0 == NULL) + m0 = pbuf_clone_to_mbuf(pbuf); + if (m0 == NULL) { return; + } - if ((pf_mtag = pf_get_mtag(m0)) == NULL) + if ((pf_mtag = pf_get_mtag(m0)) == NULL) { return; + } pf_mtag->pftag_flags |= PF_TAG_GENERATED; - if (PF_RTABLEID_IS_VALID(r->rtableid)) + if (PF_RTABLEID_IS_VALID(r->rtableid)) { pf_mtag->pftag_rtableid = r->rtableid; - -#if PF_ALTQ - if (altq_allowed && r->qid) - pf_mtag->pftag_qid = r->qid; -#endif /* PF_ALTQ */ + } #if PF_ECN /* add hints for ecn */ @@ -2626,40 +2846,44 @@ int pf_match_addr(u_int8_t n, struct pf_addr *a, struct pf_addr *m, struct pf_addr *b, sa_family_t af) { - int match = 0; + int match = 0; switch (af) { #if INET case AF_INET: if ((a->addr32[0] & m->addr32[0]) == - (b->addr32[0] & m->addr32[0])) + (b->addr32[0] & m->addr32[0])) { match++; + } break; #endif /* INET */ #if INET6 case AF_INET6: if (((a->addr32[0] & m->addr32[0]) == - (b->addr32[0] & m->addr32[0])) && + (b->addr32[0] & m->addr32[0])) && ((a->addr32[1] & m->addr32[1]) == - (b->addr32[1] & m->addr32[1])) && + (b->addr32[1] & m->addr32[1])) && ((a->addr32[2] & m->addr32[2]) == - (b->addr32[2] & m->addr32[2])) && + (b->addr32[2] & m->addr32[2])) && ((a->addr32[3] & m->addr32[3]) == - (b->addr32[3] & m->addr32[3]))) + (b->addr32[3] & m->addr32[3]))) { match++; + } break; #endif /* INET6 */ } if (match) { - if (n) - return (0); - else - return (1); + if (n) { + return 0; + } else { + return 1; + } } else { - if (n) - return (1); - else - return (0); + if (n) { + return 1; + } else { + return 0; + } } } @@ -2674,31 +2898,36 @@ pf_match_addr_range(struct pf_addr *b, struct pf_addr *e, #if INET case AF_INET: if ((a->addr32[0] < b->addr32[0]) || - (a->addr32[0] > e->addr32[0])) - return (0); + (a->addr32[0] > e->addr32[0])) { + return 0; + } break; #endif /* INET */ #if INET6 case AF_INET6: { - int i; + int i; /* check a >= b */ - for (i = 0; i < 4; ++i) - if (a->addr32[i] > b->addr32[i]) + for (i = 0; i < 4; ++i) { + if (a->addr32[i] > b->addr32[i]) { break; - else if (a->addr32[i] < b->addr32[i]) - return (0); + } else if (a->addr32[i] < b->addr32[i]) { + return 0; + } + } /* check a <= e */ - for (i = 0; i < 4; ++i) - if (a->addr32[i] < e->addr32[i]) + for (i = 0; i < 4; ++i) { + if (a->addr32[i] < e->addr32[i]) { break; - else if (a->addr32[i] > e->addr32[i]) - return (0); - break; + } else if (a->addr32[i] > e->addr32[i]) { + return 0; + } + } + break; } #endif /* INET6 */ } - return (1); + return 1; } int @@ -2706,25 +2935,25 @@ pf_match(u_int8_t op, u_int32_t a1, u_int32_t a2, u_int32_t p) { switch (op) { case PF_OP_IRG: - return ((p > a1) && (p < a2)); + return (p > a1) && (p < a2); case PF_OP_XRG: - return ((p < a1) || (p > a2)); + return (p < a1) || (p > a2); case PF_OP_RRG: - return ((p >= a1) && (p <= a2)); + return (p >= a1) && (p <= a2); case PF_OP_EQ: - return (p == a1); + return p == a1; case PF_OP_NE: - return (p != a1); + return p != a1; case PF_OP_LT: - return (p < a1); + return p < a1; case PF_OP_LE: - return (p <= a1); + return p <= a1; case PF_OP_GT: - return (p > a1); + return p > a1; case PF_OP_GE: - return (p >= a1); + return p >= a1; } - return (0); /* never reached */ + return 0; /* never reached */ } int @@ -2735,7 +2964,7 @@ pf_match_port(u_int8_t op, u_int16_t a1, u_int16_t a2, u_int16_t p) NTOHS(a2); NTOHS(p); #endif - return (pf_match(op, a1, a2, p)); + return pf_match(op, a1, a2, p); } int @@ -2747,8 +2976,9 @@ pf_match_xport(u_int8_t proto, u_int8_t proto_variant, union pf_rule_xport *rx, if (sx) { switch (proto) { case IPPROTO_GRE: - if (proto_variant == PF_GRE_PPTP_VARIANT) + if (proto_variant == PF_GRE_PPTP_VARIANT) { d = (rx->call_id == sx->call_id); + } break; case IPPROTO_ESP: @@ -2759,10 +2989,11 @@ pf_match_xport(u_int8_t proto, u_int8_t proto_variant, union pf_rule_xport *rx, case IPPROTO_UDP: case IPPROTO_ICMP: case IPPROTO_ICMPV6: - if (rx->range.op) + if (rx->range.op) { d = pf_match_port(rx->range.op, rx->range.port[0], rx->range.port[1], sx->port); + } break; default: @@ -2770,78 +3001,86 @@ pf_match_xport(u_int8_t proto, u_int8_t proto_variant, union pf_rule_xport *rx, } } - return (d); + return d; } int pf_match_uid(u_int8_t op, uid_t a1, uid_t a2, uid_t u) { - if (u == UID_MAX && op != PF_OP_EQ && op != PF_OP_NE) - return (0); - return (pf_match(op, a1, a2, u)); + if (u == UID_MAX && op != PF_OP_EQ && op != PF_OP_NE) { + return 0; + } + return pf_match(op, a1, a2, u); } int pf_match_gid(u_int8_t op, gid_t a1, gid_t a2, gid_t g) { - if (g == GID_MAX && op != PF_OP_EQ && op != PF_OP_NE) - return (0); - return (pf_match(op, a1, a2, g)); + if (g == GID_MAX && op != PF_OP_EQ && op != PF_OP_NE) { + return 0; + } + return pf_match(op, a1, a2, g); } static int -pf_match_tag(struct mbuf *m, struct pf_rule *r, struct pf_mtag *pf_mtag, +pf_match_tag(struct pf_rule *r, struct pf_mtag *pf_mtag, int *tag) { -#pragma unused(m) - if (*tag == -1) + if (*tag == -1) { *tag = pf_mtag->pftag_tag; + } - return ((!r->match_tag_not && r->match_tag == *tag) || - (r->match_tag_not && r->match_tag != *tag)); + return (!r->match_tag_not && r->match_tag == *tag) || + (r->match_tag_not && r->match_tag != *tag); } int -pf_tag_packet(struct mbuf *m, struct pf_mtag *pf_mtag, int tag, +pf_tag_packet(pbuf_t *pbuf, struct pf_mtag *pf_mtag, int tag, unsigned int rtableid, struct pf_pdesc *pd) { if (tag <= 0 && !PF_RTABLEID_IS_VALID(rtableid) && - (pd == NULL || !(pd->pktflags & PKTF_FLOW_ID))) - return (0); + (pd == NULL || !(pd->pktflags & PKTF_FLOW_ID))) { + return 0; + } - if (pf_mtag == NULL && (pf_mtag = pf_get_mtag(m)) == NULL) - return (1); + if (pf_mtag == NULL && (pf_mtag = pf_get_mtag_pbuf(pbuf)) == NULL) { + return 1; + } - if (tag > 0) + if (tag > 0) { pf_mtag->pftag_tag = tag; - if (PF_RTABLEID_IS_VALID(rtableid)) + } + if (PF_RTABLEID_IS_VALID(rtableid)) { pf_mtag->pftag_rtableid = rtableid; + } if (pd != NULL && (pd->pktflags & PKTF_FLOW_ID)) { - m->m_pkthdr.pkt_flowsrc = pd->flowsrc; - m->m_pkthdr.pkt_flowid = pd->flowhash; - m->m_pkthdr.pkt_flags |= pd->pktflags; - m->m_pkthdr.pkt_proto = pd->proto; + *pbuf->pb_flowsrc = pd->flowsrc; + *pbuf->pb_flowid = pd->flowhash; + *pbuf->pb_flags |= pd->pktflags; + *pbuf->pb_proto = pd->proto; } - return (0); + return 0; } void pf_step_into_anchor(int *depth, struct pf_ruleset **rs, int n, - struct pf_rule **r, struct pf_rule **a, int *match) + struct pf_rule **r, struct pf_rule **a, int *match) { - struct pf_anchor_stackframe *f; + struct pf_anchor_stackframe *f; (*r)->anchor->match = 0; - if (match) + if (match) { *match = 0; - if (*depth >= (int)sizeof (pf_anchor_stack) / - (int)sizeof (pf_anchor_stack[0])) { + } + if (*depth >= (int)sizeof(pf_anchor_stack) / + (int)sizeof(pf_anchor_stack[0])) { printf("pf_step_into_anchor: stack overflow\n"); *r = TAILQ_NEXT(*r, entries); return; - } else if (*depth == 0 && a != NULL) + } else if (*depth == 0 && a != NULL) { *a = *r; + } f = pf_anchor_stack + (*depth)++; f->rs = *rs; f->r = *r; @@ -2865,39 +3104,45 @@ int pf_step_out_of_anchor(int *depth, struct pf_ruleset **rs, int n, struct pf_rule **r, struct pf_rule **a, int *match) { - struct pf_anchor_stackframe *f; + struct pf_anchor_stackframe *f; int quick = 0; do { - if (*depth <= 0) + if (*depth <= 0) { break; + } f = pf_anchor_stack + *depth - 1; if (f->parent != NULL && f->child != NULL) { if (f->child->match || (match != NULL && *match)) { f->r->anchor->match = 1; - *match = 0; + if (match) { + *match = 0; + } } f->child = RB_NEXT(pf_anchor_node, f->parent, f->child); if (f->child != NULL) { *rs = &f->child->ruleset; *r = TAILQ_FIRST((*rs)->rules[n].active.ptr); - if (*r == NULL) + if (*r == NULL) { continue; - else + } else { break; + } } } (*depth)--; - if (*depth == 0 && a != NULL) + if (*depth == 0 && a != NULL) { *a = NULL; + } *rs = f->rs; - if (f->r->anchor->match || (match != NULL && *match)) + if (f->r->anchor->match || (match != NULL && *match)) { quick = f->r->quick; + } *r = TAILQ_NEXT(f->r, entries); } while (*r == NULL); - return (quick); + return quick; } #if INET6 @@ -2943,31 +3188,34 @@ pf_addr_inc(struct pf_addr *addr, sa_family_t af) addr->addr32[1] = 0; addr->addr32[0] = htonl(ntohl(addr->addr32[0]) + 1); - } else + } else { addr->addr32[1] = htonl(ntohl(addr->addr32[1]) + 1); - } else + } + } else { addr->addr32[2] = htonl(ntohl(addr->addr32[2]) + 1); - } else + } + } else { addr->addr32[3] = htonl(ntohl(addr->addr32[3]) + 1); + } break; } } #endif /* INET6 */ #define mix(a, b, c) \ - do { \ - a -= b; a -= c; a ^= (c >> 13); \ - b -= c; b -= a; b ^= (a << 8); \ - c -= a; c -= b; c ^= (b >> 13); \ - a -= b; a -= c; a ^= (c >> 12); \ - b -= c; b -= a; b ^= (a << 16); \ - c -= a; c -= b; c ^= (b >> 5); \ - a -= b; a -= c; a ^= (c >> 3); \ - b -= c; b -= a; b ^= (a << 10); \ - c -= a; c -= b; c ^= (b >> 15); \ + do { \ + a -= b; a -= c; a ^= (c >> 13); \ + b -= c; b -= a; b ^= (a << 8); \ + c -= a; c -= b; c ^= (b >> 13); \ + a -= b; a -= c; a ^= (c >> 12); \ + b -= c; b -= a; b ^= (a << 16); \ + c -= a; c -= b; c ^= (b >> 5); \ + a -= b; a -= c; a ^= (c >> 3); \ + b -= c; b -= a; b ^= (a << 10); \ + c -= a; c -= b; c ^= (b >> 15); \ } while (0) /* @@ -2977,7 +3225,7 @@ static void pf_hash(struct pf_addr *inaddr, struct pf_addr *hash, struct pf_poolhashkey *key, sa_family_t af) { - u_int32_t a = 0x9e3779b9, b = 0x9e3779b9, c = key->key32[0]; + u_int32_t a = 0x9e3779b9, b = 0x9e3779b9, c = key->key32[0]; switch (af) { #if INET @@ -3018,49 +3266,53 @@ static int pf_map_addr(sa_family_t af, struct pf_rule *r, struct pf_addr *saddr, struct pf_addr *naddr, struct pf_addr *init_addr, struct pf_src_node **sn) { - unsigned char hash[16]; - struct pf_pool *rpool = &r->rpool; - struct pf_addr *raddr = &rpool->cur->addr.v.a.addr; - struct pf_addr *rmask = &rpool->cur->addr.v.a.mask; - struct pf_pooladdr *acur = rpool->cur; - struct pf_src_node k; + unsigned char hash[16]; + struct pf_pool *rpool = &r->rpool; + struct pf_addr *raddr = &rpool->cur->addr.v.a.addr; + struct pf_addr *rmask = &rpool->cur->addr.v.a.mask; + struct pf_pooladdr *acur = rpool->cur; + struct pf_src_node k; if (*sn == NULL && r->rpool.opts & PF_POOL_STICKYADDR && (r->rpool.opts & PF_POOL_TYPEMASK) != PF_POOL_NONE) { k.af = af; PF_ACPY(&k.addr, saddr, af); if (r->rule_flag & PFRULE_RULESRCTRACK || - r->rpool.opts & PF_POOL_STICKYADDR) + r->rpool.opts & PF_POOL_STICKYADDR) { k.rule.ptr = r; - else + } else { k.rule.ptr = NULL; + } pf_status.scounters[SCNT_SRC_NODE_SEARCH]++; *sn = RB_FIND(pf_src_tree, &tree_src_tracking, &k); - if (*sn != NULL && !PF_AZERO(&(*sn)->raddr, af)) { - PF_ACPY(naddr, &(*sn)->raddr, af); + if (*sn != NULL && !PF_AZERO(&(*sn)->raddr, rpool->af)) { + PF_ACPY(naddr, &(*sn)->raddr, rpool->af); if (pf_status.debug >= PF_DEBUG_MISC) { printf("pf_map_addr: src tracking maps "); pf_print_host(&k.addr, 0, af); printf(" to "); - pf_print_host(naddr, 0, af); + pf_print_host(naddr, 0, rpool->af); printf("\n"); } - return (0); + return 0; } } - if (rpool->cur->addr.type == PF_ADDR_NOROUTE) - return (1); + if (rpool->cur->addr.type == PF_ADDR_NOROUTE) { + return 1; + } if (rpool->cur->addr.type == PF_ADDR_DYNIFTL) { - if (rpool->cur->addr.p.dyn == NULL) - return (1); - switch (af) { + if (rpool->cur->addr.p.dyn == NULL) { + return 1; + } + switch (rpool->af) { #if INET case AF_INET: if (rpool->cur->addr.p.dyn->pfid_acnt4 < 1 && (rpool->opts & PF_POOL_TYPEMASK) != - PF_POOL_ROUNDROBIN) - return (1); + PF_POOL_ROUNDROBIN) { + return 1; + } raddr = &rpool->cur->addr.p.dyn->pfid_addr4; rmask = &rpool->cur->addr.p.dyn->pfid_mask4; break; @@ -3069,16 +3321,18 @@ pf_map_addr(sa_family_t af, struct pf_rule *r, struct pf_addr *saddr, case AF_INET6: if (rpool->cur->addr.p.dyn->pfid_acnt6 < 1 && (rpool->opts & PF_POOL_TYPEMASK) != - PF_POOL_ROUNDROBIN) - return (1); + PF_POOL_ROUNDROBIN) { + return 1; + } raddr = &rpool->cur->addr.p.dyn->pfid_addr6; rmask = &rpool->cur->addr.p.dyn->pfid_mask6; break; #endif /* INET6 */ } } else if (rpool->cur->addr.type == PF_ADDR_TABLE) { - if ((rpool->opts & PF_POOL_TYPEMASK) != PF_POOL_ROUNDROBIN) - return (1); /* unsupported */ + if ((rpool->opts & PF_POOL_TYPEMASK) != PF_POOL_ROUNDROBIN) { + return 1; /* unsupported */ + } } else { raddr = &rpool->cur->addr.v.a.addr; rmask = &rpool->cur->addr.v.a.mask; @@ -3086,13 +3340,14 @@ pf_map_addr(sa_family_t af, struct pf_rule *r, struct pf_addr *saddr, switch (rpool->opts & PF_POOL_TYPEMASK) { case PF_POOL_NONE: - PF_ACPY(naddr, raddr, af); + PF_ACPY(naddr, raddr, rpool->af); break; case PF_POOL_BITMASK: + ASSERT(af == rpool->af); PF_POOLMASK(naddr, raddr, rmask, saddr, af); break; case PF_POOL_RANDOM: - if (init_addr != NULL && PF_AZERO(init_addr, af)) { + if (init_addr != NULL && PF_AZERO(init_addr, rpool->af)) { switch (af) { #if INET case AF_INET: @@ -3101,36 +3356,43 @@ pf_map_addr(sa_family_t af, struct pf_rule *r, struct pf_addr *saddr, #endif /* INET */ #if INET6 case AF_INET6: - if (rmask->addr32[3] != 0xffffffff) + if (rmask->addr32[3] != 0xffffffff) { rpool->counter.addr32[3] = RandomULong(); - else + } else { break; - if (rmask->addr32[2] != 0xffffffff) + } + if (rmask->addr32[2] != 0xffffffff) { rpool->counter.addr32[2] = RandomULong(); - else + } else { break; - if (rmask->addr32[1] != 0xffffffff) + } + if (rmask->addr32[1] != 0xffffffff) { rpool->counter.addr32[1] = RandomULong(); - else + } else { break; - if (rmask->addr32[0] != 0xffffffff) + } + if (rmask->addr32[0] != 0xffffffff) { rpool->counter.addr32[0] = RandomULong(); + } break; #endif /* INET6 */ } - PF_POOLMASK(naddr, raddr, rmask, &rpool->counter, af); - PF_ACPY(init_addr, naddr, af); - + PF_POOLMASK(naddr, raddr, rmask, &rpool->counter, + rpool->af); + PF_ACPY(init_addr, naddr, rpool->af); } else { - PF_AINC(&rpool->counter, af); - PF_POOLMASK(naddr, raddr, rmask, &rpool->counter, af); + PF_AINC(&rpool->counter, rpool->af); + PF_POOLMASK(naddr, raddr, rmask, &rpool->counter, + rpool->af); } break; case PF_POOL_SRCHASH: + ASSERT(af == rpool->af); + PF_POOLMASK(naddr, raddr, rmask, saddr, af); pf_hash(saddr, (struct pf_addr *)(void *)&hash, &rpool->key, af); PF_POOLMASK(naddr, raddr, rmask, @@ -3140,102 +3402,115 @@ pf_map_addr(sa_family_t af, struct pf_rule *r, struct pf_addr *saddr, if (rpool->cur->addr.type == PF_ADDR_TABLE) { if (!pfr_pool_get(rpool->cur->addr.p.tbl, &rpool->tblidx, &rpool->counter, - &raddr, &rmask, af)) + &raddr, &rmask, rpool->af)) { goto get_addr; + } } else if (rpool->cur->addr.type == PF_ADDR_DYNIFTL) { if (rpool->cur->addr.p.dyn != NULL && !pfr_pool_get(rpool->cur->addr.p.dyn->pfid_kt, &rpool->tblidx, &rpool->counter, - &raddr, &rmask, af)) + &raddr, &rmask, af)) { goto get_addr; - } else if (pf_match_addr(0, raddr, rmask, &rpool->counter, af)) + } + } else if (pf_match_addr(0, raddr, rmask, &rpool->counter, + rpool->af)) { goto get_addr; + } - try_next: - if ((rpool->cur = TAILQ_NEXT(rpool->cur, entries)) == NULL) +try_next: + if ((rpool->cur = TAILQ_NEXT(rpool->cur, entries)) == NULL) { rpool->cur = TAILQ_FIRST(&rpool->list); + } if (rpool->cur->addr.type == PF_ADDR_TABLE) { rpool->tblidx = -1; if (pfr_pool_get(rpool->cur->addr.p.tbl, &rpool->tblidx, &rpool->counter, - &raddr, &rmask, af)) { - /* table contains no address of type 'af' */ - if (rpool->cur != acur) + &raddr, &rmask, rpool->af)) { + /* table contains no address of type + * 'rpool->af' */ + if (rpool->cur != acur) { goto try_next; - return (1); + } + return 1; } } else if (rpool->cur->addr.type == PF_ADDR_DYNIFTL) { rpool->tblidx = -1; - if (rpool->cur->addr.p.dyn == NULL) - return (1); + if (rpool->cur->addr.p.dyn == NULL) { + return 1; + } if (pfr_pool_get(rpool->cur->addr.p.dyn->pfid_kt, &rpool->tblidx, &rpool->counter, - &raddr, &rmask, af)) { - /* table contains no address of type 'af' */ - if (rpool->cur != acur) + &raddr, &rmask, rpool->af)) { + /* table contains no address of type + * 'rpool->af' */ + if (rpool->cur != acur) { goto try_next; - return (1); + } + return 1; } } else { raddr = &rpool->cur->addr.v.a.addr; rmask = &rpool->cur->addr.v.a.mask; - PF_ACPY(&rpool->counter, raddr, af); + PF_ACPY(&rpool->counter, raddr, rpool->af); } - get_addr: - PF_ACPY(naddr, &rpool->counter, af); - if (init_addr != NULL && PF_AZERO(init_addr, af)) - PF_ACPY(init_addr, naddr, af); - PF_AINC(&rpool->counter, af); +get_addr: + PF_ACPY(naddr, &rpool->counter, rpool->af); + if (init_addr != NULL && PF_AZERO(init_addr, rpool->af)) { + PF_ACPY(init_addr, naddr, rpool->af); + } + PF_AINC(&rpool->counter, rpool->af); break; } - if (*sn != NULL) - PF_ACPY(&(*sn)->raddr, naddr, af); + if (*sn != NULL) { + PF_ACPY(&(*sn)->raddr, naddr, rpool->af); + } if (pf_status.debug >= PF_DEBUG_MISC && (rpool->opts & PF_POOL_TYPEMASK) != PF_POOL_NONE) { printf("pf_map_addr: selected address "); - pf_print_host(naddr, 0, af); + pf_print_host(naddr, 0, rpool->af); printf("\n"); } - return (0); + return 0; } static int pf_get_sport(struct pf_pdesc *pd, struct pfi_kif *kif, struct pf_rule *r, struct pf_addr *saddr, union pf_state_xport *sxport, struct pf_addr *daddr, union pf_state_xport *dxport, struct pf_addr *naddr, - union pf_state_xport *nxport, struct pf_src_node **sn) + union pf_state_xport *nxport, struct pf_src_node **sn + ) { #pragma unused(kif) - struct pf_state_key_cmp key; - struct pf_addr init_addr; + struct pf_state_key_cmp key; + struct pf_addr init_addr; unsigned int cut; sa_family_t af = pd->af; u_int8_t proto = pd->proto; unsigned int low = r->rpool.proxy_port[0]; unsigned int high = r->rpool.proxy_port[1]; - bzero(&init_addr, sizeof (init_addr)); - if (pf_map_addr(af, r, saddr, naddr, &init_addr, sn)) - return (1); + bzero(&init_addr, sizeof(init_addr)); + if (pf_map_addr(af, r, saddr, naddr, &init_addr, sn)) { + return 1; + } if (proto == IPPROTO_ICMP) { low = 1; high = 65535; } - if (!nxport) - return (0); /* No output necessary. */ - + if (!nxport) { + return 0; /* No output necessary. */ + } /*--- Special mapping rules for UDP ---*/ if (proto == IPPROTO_UDP) { - /*--- Never float IKE source port ---*/ if (ntohs(sxport->port) == PF_IKE_PORT) { nxport->port = sxport->port; - return (0); + return 0; } /*--- Apply exterior mapping options ---*/ @@ -3244,22 +3519,29 @@ pf_get_sport(struct pf_pdesc *pd, struct pfi_kif *kif, struct pf_rule *r, TAILQ_FOREACH(s, &state_list, entry_list) { struct pf_state_key *sk = s->state_key; - if (!sk) + if (!sk) { continue; - if (s->nat_rule.ptr != r) + } + if (s->nat_rule.ptr != r) { continue; - if (sk->proto != IPPROTO_UDP || sk->af != af) + } + if (sk->proto != IPPROTO_UDP || + sk->af_lan != af) { continue; - if (sk->lan.xport.port != sxport->port) + } + if (sk->lan.xport.port != sxport->port) { continue; - if (PF_ANEQ(&sk->lan.addr, saddr, af)) + } + if (PF_ANEQ(&sk->lan.addr, saddr, af)) { continue; + } if (r->extmap < PF_EXTMAP_EI && - PF_ANEQ(&sk->ext.addr, daddr, af)) + PF_ANEQ(&sk->ext_lan.addr, daddr, af)) { continue; + } nxport->port = sk->gwy.xport.port; - return (0); + return 0; } } } else if (proto == IPPROTO_TCP) { @@ -3271,59 +3553,71 @@ pf_get_sport(struct pf_pdesc *pd, struct pfi_kif *kif, struct pf_rule *r, */ TAILQ_FOREACH(s, &state_list, entry_list) { struct pf_state_key* sk = s->state_key; - if (!sk) + if (!sk) { continue; - if (s->nat_rule.ptr != r) + } + if (s->nat_rule.ptr != r) { continue; - if (sk->proto != IPPROTO_TCP || sk->af != af) - continue; - if (sk->lan.xport.port != sxport->port) + } + if (sk->proto != IPPROTO_TCP || sk->af_lan != af) { continue; - if (!(PF_AEQ(&sk->lan.addr, saddr, af))) + } + if (sk->lan.xport.port != sxport->port) { continue; + } + if (!(PF_AEQ(&sk->lan.addr, saddr, af))) { + continue; + } nxport->port = sk->gwy.xport.port; - return (0); + return 0; } } do { - key.af = af; + key.af_gwy = af; key.proto = proto; - PF_ACPY(&key.ext.addr, daddr, key.af); - PF_ACPY(&key.gwy.addr, naddr, key.af); + PF_ACPY(&key.ext_gwy.addr, daddr, key.af_gwy); + PF_ACPY(&key.gwy.addr, naddr, key.af_gwy); switch (proto) { - case IPPROTO_UDP: - key.proto_variant = r->extfilter; - break; - default: - key.proto_variant = 0; - break; + case IPPROTO_UDP: + key.proto_variant = r->extfilter; + break; + default: + key.proto_variant = 0; + break; + } + if (dxport) { + key.ext_gwy.xport = *dxport; + } else { + memset(&key.ext_gwy.xport, 0, + sizeof(key.ext_gwy.xport)); } - if (dxport) - key.ext.xport = *dxport; - else - memset(&key.ext.xport, 0, sizeof (key.ext.xport)); /* * port search; start random, step; * similar 2 portloop in in_pcbbind */ if (!(proto == IPPROTO_TCP || proto == IPPROTO_UDP || proto == IPPROTO_ICMP)) { - if (dxport) + if (dxport) { key.gwy.xport = *dxport; - else + } else { memset(&key.gwy.xport, 0, - sizeof (key.ext.xport)); - if (pf_find_state_all(&key, PF_IN, NULL) == NULL) - return (0); + sizeof(key.gwy.xport)); + } + if (pf_find_state_all(&key, PF_IN, NULL) == NULL) { + return 0; + } } else if (low == 0 && high == 0) { key.gwy.xport = *nxport; - if (pf_find_state_all(&key, PF_IN, NULL) == NULL) - return (0); + if (pf_find_state_all(&key, PF_IN, NULL) == NULL + ) { + return 0; + } } else if (low == high) { key.gwy.xport.port = htons(low); - if (pf_find_state_all(&key, PF_IN, NULL) == NULL) { + if (pf_find_state_all(&key, PF_IN, NULL) == NULL + ) { nxport->port = htons(low); - return (0); + return 0; } } else { unsigned int tmp; @@ -3337,18 +3631,18 @@ pf_get_sport(struct pf_pdesc *pd, struct pfi_kif *kif, struct pf_rule *r, /* low <= cut <= high */ for (tmp = cut; tmp <= high; ++(tmp)) { key.gwy.xport.port = htons(tmp); - if (pf_find_state_all(&key, PF_IN, NULL) == - NULL) { + if (pf_find_state_all(&key, PF_IN, NULL) == NULL + ) { nxport->port = htons(tmp); - return (0); + return 0; } } for (tmp = cut - 1; tmp >= low; --(tmp)) { key.gwy.xport.port = htons(tmp); - if (pf_find_state_all(&key, PF_IN, NULL) == - NULL) { + if (pf_find_state_all(&key, PF_IN, NULL) == NULL + ) { nxport->port = htons(tmp); - return (0); + return 0; } } } @@ -3356,43 +3650,45 @@ pf_get_sport(struct pf_pdesc *pd, struct pfi_kif *kif, struct pf_rule *r, switch (r->rpool.opts & PF_POOL_TYPEMASK) { case PF_POOL_RANDOM: case PF_POOL_ROUNDROBIN: - if (pf_map_addr(af, r, saddr, naddr, &init_addr, sn)) - return (1); + if (pf_map_addr(af, r, saddr, naddr, &init_addr, sn)) { + return 1; + } break; case PF_POOL_NONE: case PF_POOL_SRCHASH: case PF_POOL_BITMASK: default: - return (1); + return 1; } } while (!PF_AEQ(&init_addr, naddr, af)); - return (1); /* none available */ + return 1; /* none available */ } static struct pf_rule * -pf_match_translation(struct pf_pdesc *pd, struct mbuf *m, int off, +pf_match_translation(struct pf_pdesc *pd, pbuf_t *pbuf, int off, int direction, struct pfi_kif *kif, struct pf_addr *saddr, union pf_state_xport *sxport, struct pf_addr *daddr, union pf_state_xport *dxport, int rs_num) { - struct pf_rule *r, *rm = NULL; - struct pf_ruleset *ruleset = NULL; - int tag = -1; - unsigned int rtableid = IFSCOPE_NONE; - int asd = 0; + struct pf_rule *r, *rm = NULL; + struct pf_ruleset *ruleset = NULL; + int tag = -1; + unsigned int rtableid = IFSCOPE_NONE; + int asd = 0; r = TAILQ_FIRST(pf_main_ruleset.rules[rs_num].active.ptr); while (r && rm == NULL) { - struct pf_rule_addr *src = NULL, *dst = NULL; - struct pf_addr_wrap *xdst = NULL; - struct pf_addr_wrap *xsrc = NULL; - union pf_rule_xport rdrxport; + struct pf_rule_addr *src = NULL, *dst = NULL; + struct pf_addr_wrap *xdst = NULL; + struct pf_addr_wrap *xsrc = NULL; + union pf_rule_xport rdrxport; if (r->action == PF_BINAT && direction == PF_IN) { src = &r->dst; - if (r->rpool.cur != NULL) + if (r->rpool.cur != NULL) { xdst = &r->rpool.cur->addr; + } } else if (r->action == PF_RDR && direction == PF_OUT) { dst = &r->src; src = &r->dst; @@ -3408,106 +3704,162 @@ pf_match_translation(struct pf_pdesc *pd, struct mbuf *m, int off, } r->evaluations++; - if (pfi_kif_match(r->kif, kif) == r->ifnot) + if (pfi_kif_match(r->kif, kif) == r->ifnot) { r = r->skip[PF_SKIP_IFP].ptr; - else if (r->direction && r->direction != direction) + } else if (r->direction && r->direction != direction) { r = r->skip[PF_SKIP_DIR].ptr; - else if (r->af && r->af != pd->af) + } else if (r->af && r->af != pd->af) { r = r->skip[PF_SKIP_AF].ptr; - else if (r->proto && r->proto != pd->proto) + } else if (r->proto && r->proto != pd->proto) { r = r->skip[PF_SKIP_PROTO].ptr; - else if (xsrc && PF_MISMATCHAW(xsrc, saddr, pd->af, 0, NULL)) + } else if (xsrc && PF_MISMATCHAW(xsrc, saddr, pd->af, 0, NULL)) { r = TAILQ_NEXT(r, entries); - else if (!xsrc && PF_MISMATCHAW(&src->addr, saddr, pd->af, - src->neg, kif)) + } else if (!xsrc && PF_MISMATCHAW(&src->addr, saddr, pd->af, + src->neg, kif)) { r = TAILQ_NEXT(r, entries); - else if (xsrc && (!rdrxport.range.port[0] || + } else if (xsrc && (!rdrxport.range.port[0] || !pf_match_xport(r->proto, r->proto_variant, &rdrxport, - sxport))) + sxport))) { r = TAILQ_NEXT(r, entries); - else if (!xsrc && !pf_match_xport(r->proto, - r->proto_variant, &src->xport, sxport)) + } else if (!xsrc && !pf_match_xport(r->proto, + r->proto_variant, &src->xport, sxport)) { r = r->skip[src == &r->src ? PF_SKIP_SRC_PORT : PF_SKIP_DST_PORT].ptr; - else if (dst != NULL && - PF_MISMATCHAW(&dst->addr, daddr, pd->af, dst->neg, NULL)) + } else if (dst != NULL && + PF_MISMATCHAW(&dst->addr, daddr, pd->af, dst->neg, NULL)) { r = r->skip[PF_SKIP_DST_ADDR].ptr; - else if (xdst != NULL && PF_MISMATCHAW(xdst, daddr, pd->af, - 0, NULL)) + } else if (xdst != NULL && PF_MISMATCHAW(xdst, daddr, pd->af, + 0, NULL)) { r = TAILQ_NEXT(r, entries); - else if (dst && !pf_match_xport(r->proto, r->proto_variant, - &dst->xport, dxport)) + } else if (dst && !pf_match_xport(r->proto, r->proto_variant, + &dst->xport, dxport)) { r = r->skip[PF_SKIP_DST_PORT].ptr; - else if (r->match_tag && !pf_match_tag(m, r, pd->pf_mtag, &tag)) + } else if (r->match_tag && !pf_match_tag(r, pd->pf_mtag, &tag)) { r = TAILQ_NEXT(r, entries); - else if (r->os_fingerprint != PF_OSFP_ANY && (pd->proto != - IPPROTO_TCP || !pf_osfp_match(pf_osfp_fingerprint(pd, m, - off, pd->hdr.tcp), r->os_fingerprint))) + } else if (r->os_fingerprint != PF_OSFP_ANY && (pd->proto != + IPPROTO_TCP || !pf_osfp_match(pf_osfp_fingerprint(pd, pbuf, + off, pd->hdr.tcp), r->os_fingerprint))) { r = TAILQ_NEXT(r, entries); - else { - if (r->tag) + } else { + if (r->tag) { tag = r->tag; - if (PF_RTABLEID_IS_VALID(r->rtableid)) + } + if (PF_RTABLEID_IS_VALID(r->rtableid)) { rtableid = r->rtableid; + } if (r->anchor == NULL) { rm = r; - } else + } else { pf_step_into_anchor(&asd, &ruleset, rs_num, &r, NULL, NULL); + } } - if (r == NULL) + if (r == NULL) { pf_step_out_of_anchor(&asd, &ruleset, rs_num, &r, NULL, NULL); + } + } + if (pf_tag_packet(pbuf, pd->pf_mtag, tag, rtableid, NULL)) { + return NULL; } - if (pf_tag_packet(m, pd->pf_mtag, tag, rtableid, NULL)) - return (NULL); if (rm != NULL && (rm->action == PF_NONAT || - rm->action == PF_NORDR || rm->action == PF_NOBINAT)) - return (NULL); - return (rm); + rm->action == PF_NORDR || rm->action == PF_NOBINAT || + rm->action == PF_NONAT64)) { + return NULL; + } + return rm; } +/* + * Get address translation information for NAT/BINAT/RDR + * pd : pf packet descriptor + * pbuf : pbuf holding the packet + * off : offset to protocol header + * direction : direction of packet + * kif : pf interface info obtained from the packet's recv interface + * sn : source node pointer (output) + * saddr : packet source address + * sxport : packet source port + * daddr : packet destination address + * dxport : packet destination port + * nsxport : translated source port (output) + * + * Translated source & destination address are updated in pd->nsaddr & + * pd->ndaddr + */ static struct pf_rule * -pf_get_translation_aux(struct pf_pdesc *pd, struct mbuf *m, int off, +pf_get_translation_aux(struct pf_pdesc *pd, pbuf_t *pbuf, int off, int direction, struct pfi_kif *kif, struct pf_src_node **sn, struct pf_addr *saddr, union pf_state_xport *sxport, struct pf_addr *daddr, - union pf_state_xport *dxport, struct pf_addr *naddr, - union pf_state_xport *nxport) + union pf_state_xport *dxport, union pf_state_xport *nsxport + ) { - struct pf_rule *r = NULL; + struct pf_rule *r = NULL; + pd->naf = pd->af; if (direction == PF_OUT) { - r = pf_match_translation(pd, m, off, direction, kif, saddr, + r = pf_match_translation(pd, pbuf, off, direction, kif, saddr, sxport, daddr, dxport, PF_RULESET_BINAT); - if (r == NULL) - r = pf_match_translation(pd, m, off, direction, kif, + if (r == NULL) { + r = pf_match_translation(pd, pbuf, off, direction, kif, saddr, sxport, daddr, dxport, PF_RULESET_RDR); - if (r == NULL) - r = pf_match_translation(pd, m, off, direction, kif, + } + if (r == NULL) { + r = pf_match_translation(pd, pbuf, off, direction, kif, saddr, sxport, daddr, dxport, PF_RULESET_NAT); + } } else { - r = pf_match_translation(pd, m, off, direction, kif, saddr, + r = pf_match_translation(pd, pbuf, off, direction, kif, saddr, sxport, daddr, dxport, PF_RULESET_RDR); - if (r == NULL) - r = pf_match_translation(pd, m, off, direction, kif, + if (r == NULL) { + r = pf_match_translation(pd, pbuf, off, direction, kif, saddr, sxport, daddr, dxport, PF_RULESET_BINAT); + } } if (r != NULL) { + struct pf_addr *nsaddr = &pd->naddr; + struct pf_addr *ndaddr = &pd->ndaddr; + + *nsaddr = *saddr; + *ndaddr = *daddr; + switch (r->action) { case PF_NONAT: + case PF_NONAT64: case PF_NOBINAT: case PF_NORDR: - return (NULL); + return NULL; case PF_NAT: + case PF_NAT64: + /* + * we do NAT64 on incoming path and we call ip_input + * which asserts receive interface to be not NULL. + * The below check is to prevent NAT64 action on any + * packet generated by local entity using synthesized + * IPv6 address. + */ + if ((r->action == PF_NAT64) && (direction == PF_OUT)) { + return NULL; + } + if (pf_get_sport(pd, kif, r, saddr, sxport, daddr, - dxport, naddr, nxport, sn)) { + dxport, nsaddr, nsxport, sn + )) { DPFPRINTF(PF_DEBUG_MISC, ("pf: NAT proxy port allocation " "(%u-%u) failed\n", r->rpool.proxy_port[0], r->rpool.proxy_port[1])); - return (NULL); + return NULL; + } + /* + * For NAT64 the destination IPv4 address is derived + * from the last 32 bits of synthesized IPv6 address + */ + if (r->action == PF_NAT64) { + ndaddr->v4addr.s_addr = daddr->addr32[3]; + pd->naf = AF_INET; } break; case PF_BINAT: @@ -3515,15 +3867,17 @@ pf_get_translation_aux(struct pf_pdesc *pd, struct mbuf *m, int off, case PF_OUT: if (r->rpool.cur->addr.type == PF_ADDR_DYNIFTL) { - if (r->rpool.cur->addr.p.dyn == NULL) - return (NULL); + if (r->rpool.cur->addr.p.dyn == NULL) { + return NULL; + } switch (pd->af) { #if INET case AF_INET: if (r->rpool.cur->addr.p.dyn-> - pfid_acnt4 < 1) - return (NULL); - PF_POOLMASK(naddr, + pfid_acnt4 < 1) { + return NULL; + } + PF_POOLMASK(nsaddr, &r->rpool.cur->addr.p.dyn-> pfid_addr4, &r->rpool.cur->addr.p.dyn-> @@ -3534,9 +3888,10 @@ pf_get_translation_aux(struct pf_pdesc *pd, struct mbuf *m, int off, #if INET6 case AF_INET6: if (r->rpool.cur->addr.p.dyn-> - pfid_acnt6 < 1) - return (NULL); - PF_POOLMASK(naddr, + pfid_acnt6 < 1) { + return NULL; + } + PF_POOLMASK(nsaddr, &r->rpool.cur->addr.p.dyn-> pfid_addr6, &r->rpool.cur->addr.p.dyn-> @@ -3546,7 +3901,7 @@ pf_get_translation_aux(struct pf_pdesc *pd, struct mbuf *m, int off, #endif /* INET6 */ } } else { - PF_POOLMASK(naddr, + PF_POOLMASK(nsaddr, &r->rpool.cur->addr.v.a.addr, &r->rpool.cur->addr.v.a.mask, saddr, pd->af); @@ -3554,15 +3909,17 @@ pf_get_translation_aux(struct pf_pdesc *pd, struct mbuf *m, int off, break; case PF_IN: if (r->src.addr.type == PF_ADDR_DYNIFTL) { - if (r->src.addr.p.dyn == NULL) - return (NULL); + if (r->src.addr.p.dyn == NULL) { + return NULL; + } switch (pd->af) { #if INET case AF_INET: if (r->src.addr.p.dyn-> - pfid_acnt4 < 1) - return (NULL); - PF_POOLMASK(naddr, + pfid_acnt4 < 1) { + return NULL; + } + PF_POOLMASK(ndaddr, &r->src.addr.p.dyn-> pfid_addr4, &r->src.addr.p.dyn-> @@ -3573,9 +3930,10 @@ pf_get_translation_aux(struct pf_pdesc *pd, struct mbuf *m, int off, #if INET6 case AF_INET6: if (r->src.addr.p.dyn-> - pfid_acnt6 < 1) - return (NULL); - PF_POOLMASK(naddr, + pfid_acnt6 < 1) { + return NULL; + } + PF_POOLMASK(ndaddr, &r->src.addr.p.dyn-> pfid_addr6, &r->src.addr.p.dyn-> @@ -3584,11 +3942,12 @@ pf_get_translation_aux(struct pf_pdesc *pd, struct mbuf *m, int off, break; #endif /* INET6 */ } - } else - PF_POOLMASK(naddr, + } else { + PF_POOLMASK(ndaddr, &r->src.addr.v.a.addr, &r->src.addr.v.a.mask, daddr, pd->af); + } break; } break; @@ -3596,15 +3955,17 @@ pf_get_translation_aux(struct pf_pdesc *pd, struct mbuf *m, int off, switch (direction) { case PF_OUT: if (r->dst.addr.type == PF_ADDR_DYNIFTL) { - if (r->dst.addr.p.dyn == NULL) - return (NULL); + if (r->dst.addr.p.dyn == NULL) { + return NULL; + } switch (pd->af) { #if INET case AF_INET: if (r->dst.addr.p.dyn-> - pfid_acnt4 < 1) - return (NULL); - PF_POOLMASK(naddr, + pfid_acnt4 < 1) { + return NULL; + } + PF_POOLMASK(nsaddr, &r->dst.addr.p.dyn-> pfid_addr4, &r->dst.addr.p.dyn-> @@ -3615,9 +3976,10 @@ pf_get_translation_aux(struct pf_pdesc *pd, struct mbuf *m, int off, #if INET6 case AF_INET6: if (r->dst.addr.p.dyn-> - pfid_acnt6 < 1) - return (NULL); - PF_POOLMASK(naddr, + pfid_acnt6 < 1) { + return NULL; + } + PF_POOLMASK(nsaddr, &r->dst.addr.p.dyn-> pfid_addr6, &r->dst.addr.p.dyn-> @@ -3627,28 +3989,31 @@ pf_get_translation_aux(struct pf_pdesc *pd, struct mbuf *m, int off, #endif /* INET6 */ } } else { - PF_POOLMASK(naddr, + PF_POOLMASK(nsaddr, &r->dst.addr.v.a.addr, &r->dst.addr.v.a.mask, daddr, pd->af); } - if (nxport && r->dst.xport.range.port[0]) - nxport->port = + if (nsxport && r->dst.xport.range.port[0]) { + nsxport->port = r->dst.xport.range.port[0]; + } break; case PF_IN: if (pf_map_addr(pd->af, r, saddr, - naddr, NULL, sn)) - return (NULL); + ndaddr, NULL, sn)) { + return NULL; + } if ((r->rpool.opts & PF_POOL_TYPEMASK) == - PF_POOL_BITMASK) - PF_POOLMASK(naddr, naddr, + PF_POOL_BITMASK) { + PF_POOLMASK(ndaddr, ndaddr, &r->rpool.cur->addr.v.a.mask, daddr, pd->af); + } - if (nxport && dxport) { + if (nsxport && dxport) { if (r->rpool.proxy_port[1]) { - u_int32_t tmp_nport; + u_int32_t tmp_nport; tmp_nport = ((ntohs(dxport->port) - @@ -3659,12 +4024,13 @@ pf_get_translation_aux(struct pf_pdesc *pd, struct mbuf *m, int off, 1)) + r->rpool.proxy_port[0]; /* wrap around if necessary */ - if (tmp_nport > 65535) + if (tmp_nport > 65535) { tmp_nport -= 65535; - nxport->port = + } + nsxport->port = htons((u_int16_t)tmp_nport); } else if (r->rpool.proxy_port[0]) { - nxport->port = htons(r->rpool. + nsxport->port = htons(r->rpool. proxy_port[0]); } } @@ -3673,50 +4039,53 @@ pf_get_translation_aux(struct pf_pdesc *pd, struct mbuf *m, int off, break; } default: - return (NULL); + return NULL; } } - return (r); + return r; } int pf_socket_lookup(int direction, struct pf_pdesc *pd) { - struct pf_addr *saddr, *daddr; - u_int16_t sport, dport; - struct inpcbinfo *pi; - int inp = 0; + struct pf_addr *saddr, *daddr; + u_int16_t sport, dport; + struct inpcbinfo *pi; + int inp = 0; - if (pd == NULL) - return (-1); + if (pd == NULL) { + return -1; + } pd->lookup.uid = UID_MAX; pd->lookup.gid = GID_MAX; pd->lookup.pid = NO_PID; switch (pd->proto) { case IPPROTO_TCP: - if (pd->hdr.tcp == NULL) - return (-1); + if (pd->hdr.tcp == NULL) { + return -1; + } sport = pd->hdr.tcp->th_sport; dport = pd->hdr.tcp->th_dport; pi = &tcbinfo; break; case IPPROTO_UDP: - if (pd->hdr.udp == NULL) - return (-1); + if (pd->hdr.udp == NULL) { + return -1; + } sport = pd->hdr.udp->uh_sport; dport = pd->hdr.udp->uh_dport; pi = &udbinfo; break; default: - return (-1); + return -1; } if (direction == PF_IN) { saddr = pd->src; daddr = pd->dst; } else { - u_int16_t p; + u_int16_t p; p = sport; sport = dport; @@ -3727,83 +4096,88 @@ pf_socket_lookup(int direction, struct pf_pdesc *pd) switch (pd->af) { #if INET case AF_INET: - inp = in_pcblookup_hash_exists(pi, saddr->v4, sport, daddr->v4, dport, + inp = in_pcblookup_hash_exists(pi, saddr->v4addr, sport, daddr->v4addr, dport, 0, &pd->lookup.uid, &pd->lookup.gid, NULL); #if INET6 if (inp == 0) { struct in6_addr s6, d6; - memset(&s6, 0, sizeof (s6)); + memset(&s6, 0, sizeof(s6)); s6.s6_addr16[5] = htons(0xffff); - memcpy(&s6.s6_addr32[3], &saddr->v4, - sizeof (saddr->v4)); + memcpy(&s6.s6_addr32[3], &saddr->v4addr, + sizeof(saddr->v4addr)); - memset(&d6, 0, sizeof (d6)); + memset(&d6, 0, sizeof(d6)); d6.s6_addr16[5] = htons(0xffff); - memcpy(&d6.s6_addr32[3], &daddr->v4, - sizeof (daddr->v4)); + memcpy(&d6.s6_addr32[3], &daddr->v4addr, + sizeof(daddr->v4addr)); inp = in6_pcblookup_hash_exists(pi, &s6, sport, &d6, dport, 0, &pd->lookup.uid, &pd->lookup.gid, NULL); if (inp == 0) { - inp = in_pcblookup_hash_exists(pi, saddr->v4, sport, - daddr->v4, dport, INPLOOKUP_WILDCARD, &pd->lookup.uid, &pd->lookup.gid, NULL); + inp = in_pcblookup_hash_exists(pi, saddr->v4addr, sport, + daddr->v4addr, dport, INPLOOKUP_WILDCARD, &pd->lookup.uid, &pd->lookup.gid, NULL); if (inp == 0) { inp = in6_pcblookup_hash_exists(pi, &s6, sport, &d6, dport, INPLOOKUP_WILDCARD, &pd->lookup.uid, &pd->lookup.gid, NULL); - if (inp == 0) - return (-1); + if (inp == 0) { + return -1; + } } } } #else if (inp == 0) { - inp = in_pcblookup_hash_exists(pi, saddr->v4, sport, - daddr->v4, dport, INPLOOKUP_WILDCARD, + inp = in_pcblookup_hash_exists(pi, saddr->v4addr, sport, + daddr->v4addr, dport, INPLOOKUP_WILDCARD, &pd->lookup.uid, &pd->lookup.gid, NULL); - if (inp == 0) - return (-1); + if (inp == 0) { + return -1; + } } #endif /* !INET6 */ break; #endif /* INET */ #if INET6 case AF_INET6: - inp = in6_pcblookup_hash_exists(pi, &saddr->v6, sport, &daddr->v6, + inp = in6_pcblookup_hash_exists(pi, &saddr->v6addr, sport, &daddr->v6addr, dport, 0, &pd->lookup.uid, &pd->lookup.gid, NULL); if (inp == 0) { - inp = in6_pcblookup_hash_exists(pi, &saddr->v6, sport, - &daddr->v6, dport, INPLOOKUP_WILDCARD, + inp = in6_pcblookup_hash_exists(pi, &saddr->v6addr, sport, + &daddr->v6addr, dport, INPLOOKUP_WILDCARD, &pd->lookup.uid, &pd->lookup.gid, NULL); - if (inp == 0) - return (-1); + if (inp == 0) { + return -1; + } } break; #endif /* INET6 */ - + default: - return (-1); + return -1; } - return (1); + return 1; } static u_int8_t -pf_get_wscale(struct mbuf *m, int off, u_int16_t th_off, sa_family_t af) +pf_get_wscale(pbuf_t *pbuf, int off, u_int16_t th_off, sa_family_t af) { - int hlen; - u_int8_t hdr[60]; - u_int8_t *opt, optlen; - u_int8_t wscale = 0; - - hlen = th_off << 2; /* hlen <= sizeof (hdr) */ - if (hlen <= (int)sizeof (struct tcphdr)) - return (0); - if (!pf_pull_hdr(m, off, hdr, hlen, NULL, NULL, af)) - return (0); - opt = hdr + sizeof (struct tcphdr); - hlen -= sizeof (struct tcphdr); + int hlen; + u_int8_t hdr[60]; + u_int8_t *opt, optlen; + u_int8_t wscale = 0; + + hlen = th_off << 2; /* hlen <= sizeof (hdr) */ + if (hlen <= (int)sizeof(struct tcphdr)) { + return 0; + } + if (!pf_pull_hdr(pbuf, off, hdr, hlen, NULL, NULL, af)) { + return 0; + } + opt = hdr + sizeof(struct tcphdr); + hlen -= sizeof(struct tcphdr); while (hlen >= 3) { switch (*opt) { case TCPOPT_EOL: @@ -3813,37 +4187,41 @@ pf_get_wscale(struct mbuf *m, int off, u_int16_t th_off, sa_family_t af) break; case TCPOPT_WINDOW: wscale = opt[2]; - if (wscale > TCP_MAX_WINSHIFT) + if (wscale > TCP_MAX_WINSHIFT) { wscale = TCP_MAX_WINSHIFT; + } wscale |= PF_WSCALE_FLAG; - /* FALLTHROUGH */ + /* FALLTHROUGH */ default: optlen = opt[1]; - if (optlen < 2) + if (optlen < 2) { optlen = 2; + } hlen -= optlen; opt += optlen; break; } } - return (wscale); + return wscale; } static u_int16_t -pf_get_mss(struct mbuf *m, int off, u_int16_t th_off, sa_family_t af) +pf_get_mss(pbuf_t *pbuf, int off, u_int16_t th_off, sa_family_t af) { - int hlen; - u_int8_t hdr[60]; - u_int8_t *opt, optlen; - u_int16_t mss = tcp_mssdflt; - - hlen = th_off << 2; /* hlen <= sizeof (hdr) */ - if (hlen <= (int)sizeof (struct tcphdr)) - return (0); - if (!pf_pull_hdr(m, off, hdr, hlen, NULL, NULL, af)) - return (0); - opt = hdr + sizeof (struct tcphdr); - hlen -= sizeof (struct tcphdr); + int hlen; + u_int8_t hdr[60]; + u_int8_t *opt, optlen; + u_int16_t mss = tcp_mssdflt; + + hlen = th_off << 2; /* hlen <= sizeof (hdr) */ + if (hlen <= (int)sizeof(struct tcphdr)) { + return 0; + } + if (!pf_pull_hdr(pbuf, off, hdr, hlen, NULL, NULL, af)) { + return 0; + } + opt = hdr + sizeof(struct tcphdr); + hlen -= sizeof(struct tcphdr); while (hlen >= TCPOLEN_MAXSEG) { switch (*opt) { case TCPOPT_EOL: @@ -3856,98 +4234,101 @@ pf_get_mss(struct mbuf *m, int off, u_int16_t th_off, sa_family_t af) #if BYTE_ORDER != BIG_ENDIAN NTOHS(mss); #endif - /* FALLTHROUGH */ + /* FALLTHROUGH */ default: optlen = opt[1]; - if (optlen < 2) + if (optlen < 2) { optlen = 2; + } hlen -= optlen; opt += optlen; break; } } - return (mss); + return mss; } static u_int16_t pf_calc_mss(struct pf_addr *addr, sa_family_t af, u_int16_t offer) { #if INET - struct sockaddr_in *dst; - struct route ro; + struct sockaddr_in *dst; + struct route ro; #endif /* INET */ #if INET6 - struct sockaddr_in6 *dst6; - struct route_in6 ro6; + struct sockaddr_in6 *dst6; + struct route_in6 ro6; #endif /* INET6 */ - struct rtentry *rt = NULL; - int hlen; - u_int16_t mss = tcp_mssdflt; + struct rtentry *rt = NULL; + int hlen; + u_int16_t mss = tcp_mssdflt; switch (af) { #if INET case AF_INET: - hlen = sizeof (struct ip); - bzero(&ro, sizeof (ro)); + hlen = sizeof(struct ip); + bzero(&ro, sizeof(ro)); dst = (struct sockaddr_in *)(void *)&ro.ro_dst; dst->sin_family = AF_INET; - dst->sin_len = sizeof (*dst); - dst->sin_addr = addr->v4; + dst->sin_len = sizeof(*dst); + dst->sin_addr = addr->v4addr; rtalloc(&ro); rt = ro.ro_rt; break; #endif /* INET */ #if INET6 case AF_INET6: - hlen = sizeof (struct ip6_hdr); - bzero(&ro6, sizeof (ro6)); + hlen = sizeof(struct ip6_hdr); + bzero(&ro6, sizeof(ro6)); dst6 = (struct sockaddr_in6 *)(void *)&ro6.ro_dst; dst6->sin6_family = AF_INET6; - dst6->sin6_len = sizeof (*dst6); - dst6->sin6_addr = addr->v6; + dst6->sin6_len = sizeof(*dst6); + dst6->sin6_addr = addr->v6addr; rtalloc((struct route *)&ro); rt = ro6.ro_rt; break; #endif /* INET6 */ default: panic("pf_calc_mss: not AF_INET or AF_INET6!"); - return (0); + return 0; } if (rt && rt->rt_ifp) { - mss = rt->rt_ifp->if_mtu - hlen - sizeof (struct tcphdr); + /* This is relevant only for PF SYN Proxy */ + int interface_mtu = rt->rt_ifp->if_mtu; + + if (af == AF_INET && + INTF_ADJUST_MTU_FOR_CLAT46(rt->rt_ifp)) { + interface_mtu = IN6_LINKMTU(rt->rt_ifp); + /* Further adjust the size for CLAT46 expansion */ + interface_mtu -= CLAT46_HDR_EXPANSION_OVERHD; + } + mss = interface_mtu - hlen - sizeof(struct tcphdr); mss = max(tcp_mssdflt, mss); rtfree(rt); } mss = min(mss, offer); - mss = max(mss, 64); /* sanity - at least max opt space */ - return (mss); + mss = max(mss, 64); /* sanity - at least max opt space */ + return mss; } static void -pf_set_rt_ifp(struct pf_state *s, struct pf_addr *saddr) +pf_set_rt_ifp(struct pf_state *s, struct pf_addr *saddr, sa_family_t af) { struct pf_rule *r = s->rule.ptr; s->rt_kif = NULL; - if (!r->rt || r->rt == PF_FASTROUTE) + + if (!r->rt || r->rt == PF_FASTROUTE) { return; - switch (s->state_key->af) { -#if INET - case AF_INET: - pf_map_addr(AF_INET, r, saddr, &s->rt_addr, NULL, - &s->nat_src_node); - s->rt_kif = r->rpool.cur->kif; - break; -#endif /* INET */ -#if INET6 - case AF_INET6: - pf_map_addr(AF_INET6, r, saddr, &s->rt_addr, NULL, + } + if ((af == AF_INET) || (af == AF_INET6)) { + pf_map_addr(af, r, saddr, &s->rt_addr, NULL, &s->nat_src_node); s->rt_kif = r->rpool.cur->kif; - break; -#endif /* INET6 */ } + + return; } static void @@ -3957,31 +4338,36 @@ pf_attach_state(struct pf_state_key *sk, struct pf_state *s, int tail) sk->refcnt++; /* list is sorted, if-bound states before floating */ - if (tail) + if (tail) { TAILQ_INSERT_TAIL(&sk->states, s, next); - else + } else { TAILQ_INSERT_HEAD(&sk->states, s, next); + } } static void pf_detach_state(struct pf_state *s, int flags) { - struct pf_state_key *sk = s->state_key; + struct pf_state_key *sk = s->state_key; - if (sk == NULL) + if (sk == NULL) { return; + } s->state_key = NULL; TAILQ_REMOVE(&sk->states, s, next); if (--sk->refcnt == 0) { - if (!(flags & PF_DT_SKIP_EXTGWY)) + if (!(flags & PF_DT_SKIP_EXTGWY)) { RB_REMOVE(pf_state_tree_ext_gwy, &pf_statetbl_ext_gwy, sk); - if (!(flags & PF_DT_SKIP_LANEXT)) + } + if (!(flags & PF_DT_SKIP_LANEXT)) { RB_REMOVE(pf_state_tree_lan_ext, &pf_statetbl_lan_ext, sk); - if (sk->app_state) + } + if (sk->app_state) { pool_put(&pf_app_state_pl, sk->app_state); + } pool_put(&pf_state_key_pl, sk); } } @@ -3989,20 +4375,23 @@ pf_detach_state(struct pf_state *s, int flags) struct pf_state_key * pf_alloc_state_key(struct pf_state *s, struct pf_state_key *psk) { - struct pf_state_key *sk; + struct pf_state_key *sk; - if ((sk = pool_get(&pf_state_key_pl, PR_WAITOK)) == NULL) - return (NULL); - bzero(sk, sizeof (*sk)); + if ((sk = pool_get(&pf_state_key_pl, PR_WAITOK)) == NULL) { + return NULL; + } + bzero(sk, sizeof(*sk)); TAILQ_INIT(&sk->states); pf_attach_state(sk, s, 0); /* initialize state key from psk, if provided */ if (psk != NULL) { - bcopy(&psk->lan, &sk->lan, sizeof (sk->lan)); - bcopy(&psk->gwy, &sk->gwy, sizeof (sk->gwy)); - bcopy(&psk->ext, &sk->ext, sizeof (sk->ext)); - sk->af = psk->af; + bcopy(&psk->lan, &sk->lan, sizeof(sk->lan)); + bcopy(&psk->gwy, &sk->gwy, sizeof(sk->gwy)); + bcopy(&psk->ext_lan, &sk->ext_lan, sizeof(sk->ext_lan)); + bcopy(&psk->ext_gwy, &sk->ext_gwy, sizeof(sk->ext_gwy)); + sk->af_lan = psk->af_lan; + sk->af_gwy = psk->af_gwy; sk->proto = psk->proto; sk->direction = psk->direction; sk->proto_variant = psk->proto_variant; @@ -4012,7 +4401,7 @@ pf_alloc_state_key(struct pf_state *s, struct pf_state_key *psk) /* don't touch tree entries, states and refcnt on sk */ } - return (sk); + return sk; } static u_int32_t @@ -4022,61 +4411,461 @@ pf_tcp_iss(struct pf_pdesc *pd) u_int32_t digest[4]; if (pf_tcp_secret_init == 0) { - read_random(pf_tcp_secret, sizeof (pf_tcp_secret)); + read_frandom(pf_tcp_secret, sizeof(pf_tcp_secret)); MD5Init(&pf_tcp_secret_ctx); MD5Update(&pf_tcp_secret_ctx, pf_tcp_secret, - sizeof (pf_tcp_secret)); + sizeof(pf_tcp_secret)); pf_tcp_secret_init = 1; } ctx = pf_tcp_secret_ctx; - MD5Update(&ctx, (char *)&pd->hdr.tcp->th_sport, sizeof (u_short)); - MD5Update(&ctx, (char *)&pd->hdr.tcp->th_dport, sizeof (u_short)); + MD5Update(&ctx, (char *)&pd->hdr.tcp->th_sport, sizeof(u_short)); + MD5Update(&ctx, (char *)&pd->hdr.tcp->th_dport, sizeof(u_short)); if (pd->af == AF_INET6) { - MD5Update(&ctx, (char *)&pd->src->v6, sizeof (struct in6_addr)); - MD5Update(&ctx, (char *)&pd->dst->v6, sizeof (struct in6_addr)); + MD5Update(&ctx, (char *)&pd->src->v6addr, sizeof(struct in6_addr)); + MD5Update(&ctx, (char *)&pd->dst->v6addr, sizeof(struct in6_addr)); } else { - MD5Update(&ctx, (char *)&pd->src->v4, sizeof (struct in_addr)); - MD5Update(&ctx, (char *)&pd->dst->v4, sizeof (struct in_addr)); + MD5Update(&ctx, (char *)&pd->src->v4addr, sizeof(struct in_addr)); + MD5Update(&ctx, (char *)&pd->dst->v4addr, sizeof(struct in_addr)); } MD5Final((u_char *)digest, &ctx); pf_tcp_iss_off += 4096; - return (digest[0] + random() + pf_tcp_iss_off); + return digest[0] + random() + pf_tcp_iss_off; +} + +/* + * This routine is called to perform address family translation on the + * inner IP header (that may come as payload) of an ICMP(v4addr/6) error + * response. + */ +static int +pf_change_icmp_af(pbuf_t *pbuf, int off, + struct pf_pdesc *pd, struct pf_pdesc *pd2, struct pf_addr *src, + struct pf_addr *dst, sa_family_t af, sa_family_t naf) +{ + struct ip *ip4 = NULL; + struct ip6_hdr *ip6 = NULL; + void *hdr; + int hlen, olen; + + if (af == naf || (af != AF_INET && af != AF_INET6) || + (naf != AF_INET && naf != AF_INET6)) { + return -1; + } + + /* old header */ + olen = pd2->off - off; + /* new header */ + hlen = naf == AF_INET ? sizeof(*ip4) : sizeof(*ip6); + + /* Modify the pbuf to accommodate the new header */ + hdr = pbuf_resize_segment(pbuf, off, olen, hlen); + if (hdr == NULL) { + return -1; + } + + /* translate inner ip/ip6 header */ + switch (naf) { + case AF_INET: + ip4 = hdr; + bzero(ip4, sizeof(*ip4)); + ip4->ip_v = IPVERSION; + ip4->ip_hl = sizeof(*ip4) >> 2; + ip4->ip_len = htons(sizeof(*ip4) + pd2->tot_len - olen); + ip4->ip_id = rfc6864 ? 0 : htons(ip_randomid()); + ip4->ip_off = htons(IP_DF); + ip4->ip_ttl = pd2->ttl; + if (pd2->proto == IPPROTO_ICMPV6) { + ip4->ip_p = IPPROTO_ICMP; + } else { + ip4->ip_p = pd2->proto; + } + ip4->ip_src = src->v4addr; + ip4->ip_dst = dst->v4addr; + ip4->ip_sum = pbuf_inet_cksum(pbuf, 0, 0, ip4->ip_hl << 2); + break; + case AF_INET6: + ip6 = hdr; + bzero(ip6, sizeof(*ip6)); + ip6->ip6_vfc = IPV6_VERSION; + ip6->ip6_plen = htons(pd2->tot_len - olen); + if (pd2->proto == IPPROTO_ICMP) { + ip6->ip6_nxt = IPPROTO_ICMPV6; + } else { + ip6->ip6_nxt = pd2->proto; + } + if (!pd2->ttl || pd2->ttl > IPV6_DEFHLIM) { + ip6->ip6_hlim = IPV6_DEFHLIM; + } else { + ip6->ip6_hlim = pd2->ttl; + } + ip6->ip6_src = src->v6addr; + ip6->ip6_dst = dst->v6addr; + break; + } + + /* adjust payload offset and total packet length */ + pd2->off += hlen - olen; + pd->tot_len += hlen - olen; + + return 0; +} + +#define PTR_IP(field) ((int32_t)offsetof(struct ip, field)) +#define PTR_IP6(field) ((int32_t)offsetof(struct ip6_hdr, field)) + +static int +pf_translate_icmp_af(int af, void *arg) +{ + struct icmp *icmp4; + struct icmp6_hdr *icmp6; + u_int32_t mtu; + int32_t ptr = -1; + u_int8_t type; + u_int8_t code; + + switch (af) { + case AF_INET: + icmp6 = arg; + type = icmp6->icmp6_type; + code = icmp6->icmp6_code; + mtu = ntohl(icmp6->icmp6_mtu); + + switch (type) { + case ICMP6_ECHO_REQUEST: + type = ICMP_ECHO; + break; + case ICMP6_ECHO_REPLY: + type = ICMP_ECHOREPLY; + break; + case ICMP6_DST_UNREACH: + type = ICMP_UNREACH; + switch (code) { + case ICMP6_DST_UNREACH_NOROUTE: + case ICMP6_DST_UNREACH_BEYONDSCOPE: + case ICMP6_DST_UNREACH_ADDR: + code = ICMP_UNREACH_HOST; + break; + case ICMP6_DST_UNREACH_ADMIN: + code = ICMP_UNREACH_HOST_PROHIB; + break; + case ICMP6_DST_UNREACH_NOPORT: + code = ICMP_UNREACH_PORT; + break; + default: + return -1; + } + break; + case ICMP6_PACKET_TOO_BIG: + type = ICMP_UNREACH; + code = ICMP_UNREACH_NEEDFRAG; + mtu -= 20; + break; + case ICMP6_TIME_EXCEEDED: + type = ICMP_TIMXCEED; + break; + case ICMP6_PARAM_PROB: + switch (code) { + case ICMP6_PARAMPROB_HEADER: + type = ICMP_PARAMPROB; + code = ICMP_PARAMPROB_ERRATPTR; + ptr = ntohl(icmp6->icmp6_pptr); + + if (ptr == PTR_IP6(ip6_vfc)) { + ; /* preserve */ + } else if (ptr == PTR_IP6(ip6_vfc) + 1) { + ptr = PTR_IP(ip_tos); + } else if (ptr == PTR_IP6(ip6_plen) || + ptr == PTR_IP6(ip6_plen) + 1) { + ptr = PTR_IP(ip_len); + } else if (ptr == PTR_IP6(ip6_nxt)) { + ptr = PTR_IP(ip_p); + } else if (ptr == PTR_IP6(ip6_hlim)) { + ptr = PTR_IP(ip_ttl); + } else if (ptr >= PTR_IP6(ip6_src) && + ptr < PTR_IP6(ip6_dst)) { + ptr = PTR_IP(ip_src); + } else if (ptr >= PTR_IP6(ip6_dst) && + ptr < (int32_t)sizeof(struct ip6_hdr)) { + ptr = PTR_IP(ip_dst); + } else { + return -1; + } + break; + case ICMP6_PARAMPROB_NEXTHEADER: + type = ICMP_UNREACH; + code = ICMP_UNREACH_PROTOCOL; + break; + default: + return -1; + } + break; + default: + return -1; + } + icmp6->icmp6_type = type; + icmp6->icmp6_code = code; + /* aligns well with a icmpv4 nextmtu */ + icmp6->icmp6_mtu = htonl(mtu); + /* icmpv4 pptr is a one most significant byte */ + if (ptr >= 0) { + icmp6->icmp6_pptr = htonl(ptr << 24); + } + break; + + case AF_INET6: + icmp4 = arg; + type = icmp4->icmp_type; + code = icmp4->icmp_code; + mtu = ntohs(icmp4->icmp_nextmtu); + + switch (type) { + case ICMP_ECHO: + type = ICMP6_ECHO_REQUEST; + break; + case ICMP_ECHOREPLY: + type = ICMP6_ECHO_REPLY; + break; + case ICMP_UNREACH: + type = ICMP6_DST_UNREACH; + switch (code) { + case ICMP_UNREACH_NET: + case ICMP_UNREACH_HOST: + case ICMP_UNREACH_NET_UNKNOWN: + case ICMP_UNREACH_HOST_UNKNOWN: + case ICMP_UNREACH_ISOLATED: + case ICMP_UNREACH_TOSNET: + case ICMP_UNREACH_TOSHOST: + code = ICMP6_DST_UNREACH_NOROUTE; + break; + case ICMP_UNREACH_PORT: + code = ICMP6_DST_UNREACH_NOPORT; + break; + case ICMP_UNREACH_NET_PROHIB: + case ICMP_UNREACH_HOST_PROHIB: + case ICMP_UNREACH_FILTER_PROHIB: + case ICMP_UNREACH_PRECEDENCE_CUTOFF: + code = ICMP6_DST_UNREACH_ADMIN; + break; + case ICMP_UNREACH_PROTOCOL: + type = ICMP6_PARAM_PROB; + code = ICMP6_PARAMPROB_NEXTHEADER; + ptr = offsetof(struct ip6_hdr, ip6_nxt); + break; + case ICMP_UNREACH_NEEDFRAG: + type = ICMP6_PACKET_TOO_BIG; + code = 0; + mtu += 20; + break; + default: + return -1; + } + break; + case ICMP_TIMXCEED: + type = ICMP6_TIME_EXCEEDED; + break; + case ICMP_PARAMPROB: + type = ICMP6_PARAM_PROB; + switch (code) { + case ICMP_PARAMPROB_ERRATPTR: + code = ICMP6_PARAMPROB_HEADER; + break; + case ICMP_PARAMPROB_LENGTH: + code = ICMP6_PARAMPROB_HEADER; + break; + default: + return -1; + } + + ptr = icmp4->icmp_pptr; + if (ptr == 0 || ptr == PTR_IP(ip_tos)) { + ; /* preserve */ + } else if (ptr == PTR_IP(ip_len) || + ptr == PTR_IP(ip_len) + 1) { + ptr = PTR_IP6(ip6_plen); + } else if (ptr == PTR_IP(ip_ttl)) { + ptr = PTR_IP6(ip6_hlim); + } else if (ptr == PTR_IP(ip_p)) { + ptr = PTR_IP6(ip6_nxt); + } else if (ptr >= PTR_IP(ip_src) && + ptr < PTR_IP(ip_dst)) { + ptr = PTR_IP6(ip6_src); + } else if (ptr >= PTR_IP(ip_dst) && + ptr < (int32_t)sizeof(struct ip)) { + ptr = PTR_IP6(ip6_dst); + } else { + return -1; + } + break; + default: + return -1; + } + icmp4->icmp_type = type; + icmp4->icmp_code = code; + icmp4->icmp_nextmtu = htons(mtu); + if (ptr >= 0) { + icmp4->icmp_void = htonl(ptr); + } + break; + } + + return 0; +} + +/* Note: frees pbuf if PF_NAT64 is returned */ +static int +pf_nat64_ipv6(pbuf_t *pbuf, int off, struct pf_pdesc *pd) +{ + struct ip *ip4; + struct mbuf *m; + + /* + * ip_input asserts for rcvif to be not NULL + * That may not be true for two corner cases + * 1. If for some reason a local app sends DNS + * AAAA query to local host + * 2. If IPv6 stack in kernel internally generates a + * message destined for a synthesized IPv6 end-point. + */ + if (pbuf->pb_ifp == NULL) { + return PF_DROP; + } + + ip4 = (struct ip *)pbuf_resize_segment(pbuf, 0, off, sizeof(*ip4)); + if (ip4 == NULL) { + return PF_DROP; + } + + ip4->ip_v = 4; + ip4->ip_hl = 5; + ip4->ip_tos = pd->tos & htonl(0x0ff00000); + ip4->ip_len = htons(sizeof(*ip4) + (pd->tot_len - off)); + ip4->ip_id = 0; + ip4->ip_off = htons(IP_DF); + ip4->ip_ttl = pd->ttl; + ip4->ip_p = pd->proto; + ip4->ip_sum = 0; + ip4->ip_src = pd->naddr.v4addr; + ip4->ip_dst = pd->ndaddr.v4addr; + ip4->ip_sum = pbuf_inet_cksum(pbuf, 0, 0, ip4->ip_hl << 2); + + /* recalculate icmp checksums */ + if (pd->proto == IPPROTO_ICMP) { + struct icmp *icmp; + int hlen = sizeof(*ip4); + + icmp = (struct icmp *)pbuf_contig_segment(pbuf, hlen, + ICMP_MINLEN); + if (icmp == NULL) { + return PF_DROP; + } + + icmp->icmp_cksum = 0; + icmp->icmp_cksum = pbuf_inet_cksum(pbuf, 0, hlen, + ntohs(ip4->ip_len) - hlen); + } + + if ((m = pbuf_to_mbuf(pbuf, TRUE)) != NULL) { + ip_input(m); + } + + return PF_NAT64; +} + +static int +pf_nat64_ipv4(pbuf_t *pbuf, int off, struct pf_pdesc *pd) +{ + struct ip6_hdr *ip6; + struct mbuf *m; + + if (pbuf->pb_ifp == NULL) { + return PF_DROP; + } + + ip6 = (struct ip6_hdr *)pbuf_resize_segment(pbuf, 0, off, sizeof(*ip6)); + if (ip6 == NULL) { + return PF_DROP; + } + + ip6->ip6_vfc = htonl((6 << 28) | (pd->tos << 20)); + ip6->ip6_plen = htons(pd->tot_len - off); + ip6->ip6_nxt = pd->proto; + ip6->ip6_hlim = pd->ttl; + ip6->ip6_src = pd->naddr.v6addr; + ip6->ip6_dst = pd->ndaddr.v6addr; + + /* recalculate icmp6 checksums */ + if (pd->proto == IPPROTO_ICMPV6) { + struct icmp6_hdr *icmp6; + int hlen = sizeof(*ip6); + + icmp6 = (struct icmp6_hdr *)pbuf_contig_segment(pbuf, hlen, + sizeof(*icmp6)); + if (icmp6 == NULL) { + return PF_DROP; + } + + icmp6->icmp6_cksum = 0; + icmp6->icmp6_cksum = pbuf_inet6_cksum(pbuf, + IPPROTO_ICMPV6, hlen, + ntohs(ip6->ip6_plen)); + } else if (pd->proto == IPPROTO_UDP) { + struct udphdr *uh; + int hlen = sizeof(*ip6); + + uh = (struct udphdr *)pbuf_contig_segment(pbuf, hlen, + sizeof(*uh)); + if (uh == NULL) { + return PF_DROP; + } + + if (uh->uh_sum == 0) { + uh->uh_sum = pbuf_inet6_cksum(pbuf, IPPROTO_UDP, + hlen, ntohs(ip6->ip6_plen)); + } + } + + if ((m = pbuf_to_mbuf(pbuf, TRUE)) != NULL) { + ip6_input(m); + } + + return PF_NAT64; } static int pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, - struct pfi_kif *kif, struct mbuf *m, int off, void *h, + struct pfi_kif *kif, pbuf_t *pbuf, int off, void *h, struct pf_pdesc *pd, struct pf_rule **am, struct pf_ruleset **rsm, struct ifqueue *ifq) { #pragma unused(h) - struct pf_rule *nr = NULL; - struct pf_addr *saddr = pd->src, *daddr = pd->dst; - sa_family_t af = pd->af; - struct pf_rule *r, *a = NULL; - struct pf_ruleset *ruleset = NULL; - struct pf_src_node *nsn = NULL; - struct tcphdr *th = pd->hdr.tcp; - u_short reason; - int rewrite = 0, hdrlen = 0; - int tag = -1; - unsigned int rtableid = IFSCOPE_NONE; - int asd = 0; - int match = 0; - int state_icmp = 0; - u_int16_t mss = tcp_mssdflt; - u_int8_t icmptype = 0, icmpcode = 0; - - struct pf_grev1_hdr *grev1 = pd->hdr.grev1; - union pf_state_xport bxport, nxport, sxport, dxport; - struct pf_state_key psk; - - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); + struct pf_rule *nr = NULL; + struct pf_addr *saddr = pd->src, *daddr = pd->dst; + sa_family_t af = pd->af; + struct pf_rule *r, *a = NULL; + struct pf_ruleset *ruleset = NULL; + struct pf_src_node *nsn = NULL; + struct tcphdr *th = pd->hdr.tcp; + struct udphdr *uh = pd->hdr.udp; + u_short reason; + int rewrite = 0, hdrlen = 0; + int tag = -1; + unsigned int rtableid = IFSCOPE_NONE; + int asd = 0; + int match = 0; + int state_icmp = 0; + u_int16_t mss = tcp_mssdflt; + u_int8_t icmptype = 0, icmpcode = 0; + + struct pf_grev1_hdr *grev1 = pd->hdr.grev1; + union pf_state_xport bxport, bdxport, nxport, sxport, dxport; + struct pf_state_key psk; + + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); if (direction == PF_IN && pf_check_congestion(ifq)) { REASON_SET(&reason, PFRES_CONGEST); - return (PF_DROP); + return PF_DROP; } hdrlen = 0; @@ -4088,353 +4877,409 @@ pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, case IPPROTO_TCP: sxport.port = th->th_sport; dxport.port = th->th_dport; - hdrlen = sizeof (*th); + hdrlen = sizeof(*th); break; case IPPROTO_UDP: - sxport.port = pd->hdr.udp->uh_sport; - dxport.port = pd->hdr.udp->uh_dport; - hdrlen = sizeof (*pd->hdr.udp); + sxport.port = uh->uh_sport; + dxport.port = uh->uh_dport; + hdrlen = sizeof(*uh); break; #if INET case IPPROTO_ICMP: - if (pd->af != AF_INET) + if (pd->af != AF_INET) { break; + } sxport.port = dxport.port = pd->hdr.icmp->icmp_id; hdrlen = ICMP_MINLEN; icmptype = pd->hdr.icmp->icmp_type; icmpcode = pd->hdr.icmp->icmp_code; - if (icmptype == ICMP_UNREACH || - icmptype == ICMP_SOURCEQUENCH || - icmptype == ICMP_REDIRECT || - icmptype == ICMP_TIMXCEED || - icmptype == ICMP_PARAMPROB) + if (ICMP_ERRORTYPE(icmptype)) { state_icmp++; + } break; #endif /* INET */ #if INET6 case IPPROTO_ICMPV6: - if (pd->af != AF_INET6) + if (pd->af != AF_INET6) { break; + } sxport.port = dxport.port = pd->hdr.icmp6->icmp6_id; - hdrlen = sizeof (*pd->hdr.icmp6); + hdrlen = sizeof(*pd->hdr.icmp6); icmptype = pd->hdr.icmp6->icmp6_type; icmpcode = pd->hdr.icmp6->icmp6_code; - if (icmptype == ICMP6_DST_UNREACH || - icmptype == ICMP6_PACKET_TOO_BIG || - icmptype == ICMP6_TIME_EXCEEDED || - icmptype == ICMP6_PARAM_PROB) + if (ICMP6_ERRORTYPE(icmptype)) { state_icmp++; + } break; #endif /* INET6 */ case IPPROTO_GRE: if (pd->proto_variant == PF_GRE_PPTP_VARIANT) { sxport.call_id = dxport.call_id = pd->hdr.grev1->call_id; - hdrlen = sizeof (*pd->hdr.grev1); + hdrlen = sizeof(*pd->hdr.grev1); } break; case IPPROTO_ESP: sxport.spi = 0; dxport.spi = pd->hdr.esp->spi; - hdrlen = sizeof (*pd->hdr.esp); + hdrlen = sizeof(*pd->hdr.esp); break; } r = TAILQ_FIRST(pf_main_ruleset.rules[PF_RULESET_FILTER].active.ptr); + bxport = sxport; + bdxport = dxport; + if (direction == PF_OUT) { - bxport = nxport = sxport; - /* check outgoing packet for BINAT/NAT */ - if ((nr = pf_get_translation_aux(pd, m, off, PF_OUT, kif, &nsn, - saddr, &sxport, daddr, &dxport, &pd->naddr, &nxport)) != - NULL) { - PF_ACPY(&pd->baddr, saddr, af); - switch (pd->proto) { - case IPPROTO_TCP: + nxport = sxport; + } else { + nxport = dxport; + } + + /* check packet for BINAT/NAT/RDR */ + if ((nr = pf_get_translation_aux(pd, pbuf, off, direction, kif, &nsn, + saddr, &sxport, daddr, &dxport, &nxport + )) != NULL) { + int ua; + u_int16_t dport; + + if (pd->af != pd->naf) { + ua = 0; + } else { + ua = 1; + } + + PF_ACPY(&pd->baddr, saddr, af); + PF_ACPY(&pd->bdaddr, daddr, af); + + switch (pd->proto) { + case IPPROTO_TCP: + if (pd->af != pd->naf || + PF_ANEQ(saddr, &pd->naddr, pd->af)) { pf_change_ap(direction, pd->mp, saddr, &th->th_sport, pd->ip_sum, &th->th_sum, - &pd->naddr, nxport.port, 0, af); + &pd->naddr, nxport.port, 0, af, + pd->naf, ua); sxport.port = th->th_sport; - rewrite++; - break; - case IPPROTO_UDP: + } + + if (pd->af != pd->naf || + PF_ANEQ(daddr, &pd->ndaddr, pd->af) || + (nr && (nr->action == PF_RDR) && + (th->th_dport != nxport.port))) { + if (nr && nr->action == PF_RDR) { + dport = nxport.port; + } else { + dport = th->th_dport; + } + pf_change_ap(direction, pd->mp, daddr, + &th->th_dport, pd->ip_sum, + &th->th_sum, &pd->ndaddr, + dport, 0, af, pd->naf, ua); + dxport.port = th->th_dport; + } + rewrite++; + break; + + case IPPROTO_UDP: + if (pd->af != pd->naf || + PF_ANEQ(saddr, &pd->naddr, pd->af)) { pf_change_ap(direction, pd->mp, saddr, - &pd->hdr.udp->uh_sport, pd->ip_sum, - &pd->hdr.udp->uh_sum, &pd->naddr, - nxport.port, 1, af); - sxport.port = pd->hdr.udp->uh_sport; - rewrite++; - break; -#if INET - case IPPROTO_ICMP: - if (pd->af == AF_INET) { - pf_change_a(&saddr->v4.s_addr, pd->ip_sum, - pd->naddr.v4.s_addr, 0); - pd->hdr.icmp->icmp_cksum = pf_cksum_fixup( - pd->hdr.icmp->icmp_cksum, sxport.port, - nxport.port, 0); - pd->hdr.icmp->icmp_id = nxport.port; - ++rewrite; + &uh->uh_sport, pd->ip_sum, + &uh->uh_sum, &pd->naddr, + nxport.port, 1, af, pd->naf, ua); + sxport.port = uh->uh_sport; + } + + if (pd->af != pd->naf || + PF_ANEQ(daddr, &pd->ndaddr, pd->af) || + (nr && (nr->action == PF_RDR) && + (uh->uh_dport != nxport.port))) { + if (nr && nr->action == PF_RDR) { + dport = nxport.port; + } else { + dport = uh->uh_dport; } + pf_change_ap(direction, pd->mp, daddr, + &uh->uh_dport, pd->ip_sum, + &uh->uh_sum, &pd->ndaddr, + dport, 0, af, pd->naf, ua); + dxport.port = uh->uh_dport; + } + rewrite++; + break; +#if INET + case IPPROTO_ICMP: + if (pd->af != AF_INET) { break; + } + /* + * TODO: + * pd->af != pd->naf not handled yet here and would be + * needed for NAT46 needed to support XLAT. + * Will cross the bridge when it comes. + */ + if (PF_ANEQ(saddr, &pd->naddr, pd->af)) { + pf_change_a(&saddr->v4addr.s_addr, pd->ip_sum, + pd->naddr.v4addr.s_addr, 0); + pd->hdr.icmp->icmp_cksum = pf_cksum_fixup( + pd->hdr.icmp->icmp_cksum, sxport.port, + nxport.port, 0); + pd->hdr.icmp->icmp_id = nxport.port; + } + + if (PF_ANEQ(daddr, &pd->ndaddr, pd->af)) { + pf_change_a(&daddr->v4addr.s_addr, pd->ip_sum, + pd->ndaddr.v4addr.s_addr, 0); + } + ++rewrite; + break; #endif /* INET */ #if INET6 - case IPPROTO_ICMPV6: - if (pd->af == AF_INET6) { - pf_change_a6(saddr, &pd->hdr.icmp6->icmp6_cksum, - &pd->naddr, 0); - rewrite++; - } + case IPPROTO_ICMPV6: + if (pd->af != AF_INET6) { break; + } + + if (pd->af != pd->naf || + PF_ANEQ(saddr, &pd->naddr, pd->af)) { + pf_change_addr(saddr, + &pd->hdr.icmp6->icmp6_cksum, + &pd->naddr, 0, pd->af, pd->naf); + } + + if (pd->af != pd->naf || + PF_ANEQ(daddr, &pd->ndaddr, pd->af)) { + pf_change_addr(daddr, + &pd->hdr.icmp6->icmp6_cksum, + &pd->ndaddr, 0, pd->af, pd->naf); + } + + if (pd->af != pd->naf) { + if (pf_translate_icmp_af(AF_INET, + pd->hdr.icmp6)) { + return PF_DROP; + } + pd->proto = IPPROTO_ICMP; + } + rewrite++; + break; #endif /* INET */ - case IPPROTO_GRE: - switch (af) { + case IPPROTO_GRE: + if ((direction == PF_IN) && + (pd->proto_variant == PF_GRE_PPTP_VARIANT)) { + grev1->call_id = nxport.call_id; + } + + switch (pd->af) { #if INET - case AF_INET: - pf_change_a(&saddr->v4.s_addr, - pd->ip_sum, pd->naddr.v4.s_addr, 0); - break; -#endif /* INET */ -#if INET6 - case AF_INET6: - PF_ACPY(saddr, &pd->naddr, AF_INET6); - break; -#endif /* INET6 */ + case AF_INET: + if (PF_ANEQ(saddr, &pd->naddr, pd->af)) { + pf_change_a(&saddr->v4addr.s_addr, + pd->ip_sum, + pd->naddr.v4addr.s_addr, 0); + } + if (PF_ANEQ(daddr, &pd->ndaddr, pd->af)) { + pf_change_a(&daddr->v4addr.s_addr, + pd->ip_sum, + pd->ndaddr.v4addr.s_addr, 0); } - ++rewrite; break; - case IPPROTO_ESP: - bxport.spi = 0; - switch (af) { -#if INET - case AF_INET: - pf_change_a(&saddr->v4.s_addr, - pd->ip_sum, pd->naddr.v4.s_addr, 0); - break; #endif /* INET */ #if INET6 - case AF_INET6: + case AF_INET6: + if (PF_ANEQ(saddr, &pd->naddr, pd->af)) { PF_ACPY(saddr, &pd->naddr, AF_INET6); - break; -#endif /* INET6 */ } - break; - default: - switch (af) { -#if INET - case AF_INET: - pf_change_a(&saddr->v4.s_addr, - pd->ip_sum, pd->naddr.v4.s_addr, 0); - break; -#endif /* INET */ -#if INET6 - case AF_INET6: - PF_ACPY(saddr, &pd->naddr, af); - break; -#endif /* INET */ + if (PF_ANEQ(daddr, &pd->ndaddr, pd->af)) { + PF_ACPY(daddr, &pd->ndaddr, AF_INET6); } break; +#endif /* INET6 */ + } + ++rewrite; + break; + case IPPROTO_ESP: + if (direction == PF_OUT) { + bxport.spi = 0; } - if (nr->natpass) - r = NULL; - pd->nat_rule = nr; - } - } else { - bxport.port = nxport.port = dxport.port; - /* check incoming packet for BINAT/RDR */ - if ((nr = pf_get_translation_aux(pd, m, off, PF_IN, kif, &nsn, - saddr, &sxport, daddr, &dxport, &pd->naddr, &nxport)) != - NULL) { - PF_ACPY(&pd->baddr, daddr, af); - switch (pd->proto) { - case IPPROTO_TCP: - pf_change_ap(direction, pd->mp, daddr, - &th->th_dport, pd->ip_sum, &th->th_sum, - &pd->naddr, nxport.port, 0, af); - dxport.port = th->th_dport; - rewrite++; - break; - case IPPROTO_UDP: - pf_change_ap(direction, pd->mp, daddr, - &pd->hdr.udp->uh_dport, pd->ip_sum, - &pd->hdr.udp->uh_sum, &pd->naddr, - nxport.port, 1, af); - dxport.port = pd->hdr.udp->uh_dport; - rewrite++; - break; + switch (pd->af) { #if INET - case IPPROTO_ICMP: - if (pd->af == AF_INET) { - pf_change_a(&daddr->v4.s_addr, pd->ip_sum, - pd->naddr.v4.s_addr, 0); + case AF_INET: + if (PF_ANEQ(saddr, &pd->naddr, pd->af)) { + pf_change_a(&saddr->v4addr.s_addr, + pd->ip_sum, pd->naddr.v4addr.s_addr, 0); + } + if (PF_ANEQ(daddr, &pd->ndaddr, pd->af)) { + pf_change_a(&daddr->v4addr.s_addr, + pd->ip_sum, + pd->ndaddr.v4addr.s_addr, 0); } break; #endif /* INET */ #if INET6 - case IPPROTO_ICMPV6: - if (pd->af == AF_INET6) { - pf_change_a6(daddr, &pd->hdr.icmp6->icmp6_cksum, - &pd->naddr, 0); - rewrite++; + case AF_INET6: + if (PF_ANEQ(saddr, &pd->naddr, pd->af)) { + PF_ACPY(saddr, &pd->naddr, AF_INET6); + } + if (PF_ANEQ(daddr, &pd->ndaddr, pd->af)) { + PF_ACPY(daddr, &pd->ndaddr, AF_INET6); } break; #endif /* INET6 */ - case IPPROTO_GRE: - if (pd->proto_variant == PF_GRE_PPTP_VARIANT) - grev1->call_id = nxport.call_id; - - switch (af) { + } + break; + default: + switch (pd->af) { #if INET - case AF_INET: - pf_change_a(&daddr->v4.s_addr, - pd->ip_sum, pd->naddr.v4.s_addr, 0); - break; -#endif /* INET */ -#if INET6 - case AF_INET6: - PF_ACPY(daddr, &pd->naddr, AF_INET6); - break; -#endif /* INET6 */ + case AF_INET: + if ((pd->naf != AF_INET) || + (PF_ANEQ(saddr, &pd->naddr, pd->af))) { + pf_change_addr(saddr, pd->ip_sum, + &pd->naddr, 0, af, pd->naf); } - ++rewrite; - break; - case IPPROTO_ESP: - switch (af) { -#if INET - case AF_INET: - pf_change_a(&daddr->v4.s_addr, - pd->ip_sum, pd->naddr.v4.s_addr, 0); - break; -#endif /* INET */ -#if INET6 - case AF_INET6: - PF_ACPY(daddr, &pd->naddr, AF_INET6); - break; -#endif /* INET6 */ + + if ((pd->naf != AF_INET) || + (PF_ANEQ(daddr, &pd->ndaddr, pd->af))) { + pf_change_addr(daddr, pd->ip_sum, + &pd->ndaddr, 0, af, pd->naf); } break; - default: - switch (af) { -#if INET - case AF_INET: - pf_change_a(&daddr->v4.s_addr, - pd->ip_sum, pd->naddr.v4.s_addr, 0); - break; #endif /* INET */ #if INET6 - case AF_INET6: - PF_ACPY(daddr, &pd->naddr, af); - break; -#endif /* INET */ + case AF_INET6: + if (PF_ANEQ(saddr, &pd->naddr, pd->af)) { + PF_ACPY(saddr, &pd->naddr, af); + } + if (PF_ANEQ(daddr, &pd->ndaddr, pd->af)) { + PF_ACPY(daddr, &pd->ndaddr, af); } break; +#endif /* INET */ } + break; + } - if (nr->natpass) - r = NULL; - pd->nat_rule = nr; + if (nr->natpass) { + r = NULL; } + pd->nat_rule = nr; + pd->af = pd->naf; + } else { } - if (nr && nr->tag > 0) + if (nr && nr->tag > 0) { tag = nr->tag; + } while (r != NULL) { r->evaluations++; - if (pfi_kif_match(r->kif, kif) == r->ifnot) + if (pfi_kif_match(r->kif, kif) == r->ifnot) { r = r->skip[PF_SKIP_IFP].ptr; - else if (r->direction && r->direction != direction) + } else if (r->direction && r->direction != direction) { r = r->skip[PF_SKIP_DIR].ptr; - else if (r->af && r->af != af) + } else if (r->af && r->af != pd->af) { r = r->skip[PF_SKIP_AF].ptr; - else if (r->proto && r->proto != pd->proto) + } else if (r->proto && r->proto != pd->proto) { r = r->skip[PF_SKIP_PROTO].ptr; - else if (PF_MISMATCHAW(&r->src.addr, saddr, af, - r->src.neg, kif)) + } else if (PF_MISMATCHAW(&r->src.addr, saddr, pd->af, + r->src.neg, kif)) { r = r->skip[PF_SKIP_SRC_ADDR].ptr; + } /* tcp/udp only. port_op always 0 in other cases */ else if (r->proto == pd->proto && (r->proto == IPPROTO_TCP || r->proto == IPPROTO_UDP) && r->src.xport.range.op && !pf_match_port(r->src.xport.range.op, r->src.xport.range.port[0], r->src.xport.range.port[1], - th->th_sport)) + th->th_sport)) { r = r->skip[PF_SKIP_SRC_PORT].ptr; - else if (PF_MISMATCHAW(&r->dst.addr, daddr, af, - r->dst.neg, NULL)) + } else if (PF_MISMATCHAW(&r->dst.addr, daddr, pd->af, + r->dst.neg, NULL)) { r = r->skip[PF_SKIP_DST_ADDR].ptr; + } /* tcp/udp only. port_op always 0 in other cases */ else if (r->proto == pd->proto && (r->proto == IPPROTO_TCP || r->proto == IPPROTO_UDP) && r->dst.xport.range.op && !pf_match_port(r->dst.xport.range.op, r->dst.xport.range.port[0], r->dst.xport.range.port[1], - th->th_dport)) + th->th_dport)) { r = r->skip[PF_SKIP_DST_PORT].ptr; + } /* icmp only. type always 0 in other cases */ - else if (r->type && r->type != icmptype + 1) + else if (r->type && r->type != icmptype + 1) { r = TAILQ_NEXT(r, entries); + } /* icmp only. type always 0 in other cases */ - else if (r->code && r->code != icmpcode + 1) + else if (r->code && r->code != icmpcode + 1) { r = TAILQ_NEXT(r, entries); - else if ((r->rule_flag & PFRULE_TOS) && r->tos && - !(r->tos & pd->tos)) + } else if ((r->rule_flag & PFRULE_TOS) && r->tos && + !(r->tos & pd->tos)) { r = TAILQ_NEXT(r, entries); - else if ((r->rule_flag & PFRULE_DSCP) && r->tos && - !(r->tos & (pd->tos & DSCP_MASK))) + } else if ((r->rule_flag & PFRULE_DSCP) && r->tos && + !(r->tos & (pd->tos & DSCP_MASK))) { r = TAILQ_NEXT(r, entries); - else if ((r->rule_flag & PFRULE_SC) && r->tos && - ((r->tos & SCIDX_MASK) != pd->sc)) + } else if ((r->rule_flag & PFRULE_SC) && r->tos && + ((r->tos & SCIDX_MASK) != pd->sc)) { r = TAILQ_NEXT(r, entries); - else if (r->rule_flag & PFRULE_FRAGMENT) + } else if (r->rule_flag & PFRULE_FRAGMENT) { r = TAILQ_NEXT(r, entries); - else if (pd->proto == IPPROTO_TCP && - (r->flagset & th->th_flags) != r->flags) + } else if (pd->proto == IPPROTO_TCP && + (r->flagset & th->th_flags) != r->flags) { r = TAILQ_NEXT(r, entries); + } /* tcp/udp only. uid.op always 0 in other cases */ - else if (r->uid.op && (pd->lookup.done || (pd->lookup.done = - pf_socket_lookup(direction, pd), 1)) && + else if (r->uid.op && (pd->lookup.done || ((void)(pd->lookup.done = + pf_socket_lookup(direction, pd)), 1)) && !pf_match_uid(r->uid.op, r->uid.uid[0], r->uid.uid[1], - pd->lookup.uid)) + pd->lookup.uid)) { r = TAILQ_NEXT(r, entries); + } /* tcp/udp only. gid.op always 0 in other cases */ - else if (r->gid.op && (pd->lookup.done || (pd->lookup.done = - pf_socket_lookup(direction, pd), 1)) && + else if (r->gid.op && (pd->lookup.done || ((void)(pd->lookup.done = + pf_socket_lookup(direction, pd)), 1)) && !pf_match_gid(r->gid.op, r->gid.gid[0], r->gid.gid[1], - pd->lookup.gid)) + pd->lookup.gid)) { r = TAILQ_NEXT(r, entries); - else if (r->prob && r->prob <= (RandomULong() % (UINT_MAX - 1) + 1)) + } else if (r->prob && r->prob <= (RandomULong() % (UINT_MAX - 1) + 1)) { r = TAILQ_NEXT(r, entries); - else if (r->match_tag && !pf_match_tag(m, r, pd->pf_mtag, &tag)) + } else if (r->match_tag && !pf_match_tag(r, pd->pf_mtag, &tag)) { r = TAILQ_NEXT(r, entries); - else if (r->os_fingerprint != PF_OSFP_ANY && + } else if (r->os_fingerprint != PF_OSFP_ANY && (pd->proto != IPPROTO_TCP || !pf_osfp_match( - pf_osfp_fingerprint(pd, m, off, th), - r->os_fingerprint))) + pf_osfp_fingerprint(pd, pbuf, off, th), + r->os_fingerprint))) { r = TAILQ_NEXT(r, entries); - else { - if (r->tag) + } else { + if (r->tag) { tag = r->tag; - if (PF_RTABLEID_IS_VALID(r->rtableid)) + } + if (PF_RTABLEID_IS_VALID(r->rtableid)) { rtableid = r->rtableid; + } if (r->anchor == NULL) { match = 1; *rm = r; *am = a; *rsm = ruleset; - if ((*rm)->quick) + if ((*rm)->quick) { break; + } r = TAILQ_NEXT(r, entries); - } else + } else { pf_step_into_anchor(&asd, &ruleset, PF_RULESET_FILTER, &r, &a, &match); + } } if (r == NULL && pf_step_out_of_anchor(&asd, &ruleset, - PF_RULESET_FILTER, &r, &a, &match)) + PF_RULESET_FILTER, &r, &a, &match)) { break; + } } r = *rm; a = *am; @@ -4444,19 +5289,19 @@ pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, if (r->log || (nr != NULL && nr->log)) { if (rewrite > 0) { - if (rewrite < off + hdrlen) + if (rewrite < off + hdrlen) { rewrite = off + hdrlen; + } - m = pf_lazy_makewritable(pd, m, rewrite); - if (!m) { + if (pf_lazy_makewritable(pd, pbuf, rewrite) == NULL) { REASON_SET(&reason, PFRES_MEMORY); - return (PF_DROP); + return PF_DROP; } - m_copyback(m, off, hdrlen, pd->hdr.any); + pbuf_copy_back(pbuf, off, hdrlen, pd->hdr.any); } - PFLOG_PACKET(kif, h, m, af, direction, reason, r->log ? r : nr, - a, ruleset, pd); + PFLOG_PACKET(kif, h, pbuf, pd->af, direction, reason, + r->log ? r : nr, a, ruleset, pd); } if ((r->action == PF_DROP) && @@ -4464,14 +5309,16 @@ pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, (r->rule_flag & PFRULE_RETURNICMP) || (r->rule_flag & PFRULE_RETURN))) { /* undo NAT changes, if they have taken place */ - if (nr != NULL) { + /* XXX For NAT64 we are not reverting the changes */ + if (nr != NULL && nr->action != PF_NAT64) { if (direction == PF_OUT) { + pd->af = af; switch (pd->proto) { case IPPROTO_TCP: pf_change_ap(direction, pd->mp, saddr, &th->th_sport, pd->ip_sum, &th->th_sum, &pd->baddr, - bxport.port, 0, af); + bxport.port, 0, af, pd->af, 1); sxport.port = th->th_sport; rewrite++; break; @@ -4479,7 +5326,7 @@ pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, pf_change_ap(direction, pd->mp, saddr, &pd->hdr.udp->uh_sport, pd->ip_sum, &pd->hdr.udp->uh_sum, &pd->baddr, - bxport.port, 1, af); + bxport.port, 1, af, pd->af, 1); sxport.port = pd->hdr.udp->uh_sport; rewrite++; break; @@ -4495,9 +5342,9 @@ pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, switch (af) { #if INET case AF_INET: - pf_change_a(&saddr->v4.s_addr, + pf_change_a(&saddr->v4addr.s_addr, pd->ip_sum, - pd->baddr.v4.s_addr, 0); + pd->baddr.v4addr.s_addr, 0); break; #endif /* INET */ #if INET6 @@ -4513,9 +5360,9 @@ pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, switch (af) { #if INET case AF_INET: - pf_change_a(&saddr->v4.s_addr, + pf_change_a(&saddr->v4addr.s_addr, pd->ip_sum, - pd->baddr.v4.s_addr, 0); + pd->baddr.v4addr.s_addr, 0); break; #endif /* INET */ #if INET6 @@ -4529,9 +5376,9 @@ pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, default: switch (af) { case AF_INET: - pf_change_a(&saddr->v4.s_addr, + pf_change_a(&saddr->v4addr.s_addr, pd->ip_sum, - pd->baddr.v4.s_addr, 0); + pd->baddr.v4addr.s_addr, 0); break; case AF_INET6: PF_ACPY(saddr, &pd->baddr, af); @@ -4543,16 +5390,16 @@ pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, case IPPROTO_TCP: pf_change_ap(direction, pd->mp, daddr, &th->th_dport, pd->ip_sum, - &th->th_sum, &pd->baddr, - bxport.port, 0, af); + &th->th_sum, &pd->bdaddr, + bdxport.port, 0, af, pd->af, 1); dxport.port = th->th_dport; rewrite++; break; case IPPROTO_UDP: pf_change_ap(direction, pd->mp, daddr, &pd->hdr.udp->uh_dport, pd->ip_sum, - &pd->hdr.udp->uh_sum, &pd->baddr, - bxport.port, 1, af); + &pd->hdr.udp->uh_sum, &pd->bdaddr, + bdxport.port, 1, af, pd->af, 1); dxport.port = pd->hdr.udp->uh_dport; rewrite++; break; @@ -4564,20 +5411,22 @@ pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, break; case IPPROTO_GRE: if (pd->proto_variant == - PF_GRE_PPTP_VARIANT) - grev1->call_id = bxport.call_id; + PF_GRE_PPTP_VARIANT) { + grev1->call_id = + bdxport.call_id; + } ++rewrite; switch (af) { #if INET case AF_INET: - pf_change_a(&daddr->v4.s_addr, + pf_change_a(&daddr->v4addr.s_addr, pd->ip_sum, - pd->baddr.v4.s_addr, 0); + pd->bdaddr.v4addr.s_addr, 0); break; #endif /* INET */ #if INET6 case AF_INET6: - PF_ACPY(daddr, &pd->baddr, + PF_ACPY(daddr, &pd->bdaddr, AF_INET6); break; #endif /* INET6 */ @@ -4587,14 +5436,14 @@ pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, switch (af) { #if INET case AF_INET: - pf_change_a(&daddr->v4.s_addr, + pf_change_a(&daddr->v4addr.s_addr, pd->ip_sum, - pd->baddr.v4.s_addr, 0); + pd->bdaddr.v4addr.s_addr, 0); break; #endif /* INET */ #if INET6 case AF_INET6: - PF_ACPY(daddr, &pd->baddr, + PF_ACPY(daddr, &pd->bdaddr, AF_INET6); break; #endif /* INET6 */ @@ -4603,13 +5452,13 @@ pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, default: switch (af) { case AF_INET: - pf_change_a(&daddr->v4.s_addr, + pf_change_a(&daddr->v4addr.s_addr, pd->ip_sum, - pd->baddr.v4.s_addr, 0); + pd->bdaddr.v4addr.s_addr, 0); break; #if INET6 case AF_INET6: - PF_ACPY(daddr, &pd->baddr, af); + PF_ACPY(daddr, &pd->bdaddr, af); break; #endif /* INET6 */ } @@ -4620,128 +5469,191 @@ pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, ((r->rule_flag & PFRULE_RETURNRST) || (r->rule_flag & PFRULE_RETURN)) && !(th->th_flags & TH_RST)) { - u_int32_t ack = ntohl(th->th_seq) + pd->p_len; - int len = 0; - struct ip *h4; + u_int32_t ack = ntohl(th->th_seq) + pd->p_len; + int len = 0; + struct ip *h4; #if INET6 - struct ip6_hdr *h6; + struct ip6_hdr *h6; #endif /* INET6 */ - switch (af) { + switch (pd->af) { case AF_INET: - h4 = mtod(m, struct ip *); + h4 = pbuf->pb_data; len = ntohs(h4->ip_len) - off; break; #if INET6 case AF_INET6: - h6 = mtod(m, struct ip6_hdr *); + h6 = pbuf->pb_data; len = ntohs(h6->ip6_plen) - - (off - sizeof (*h6)); + (off - sizeof(*h6)); break; #endif /* INET6 */ } - if (pf_check_proto_cksum(m, off, len, IPPROTO_TCP, af)) + if (pf_check_proto_cksum(pbuf, off, len, IPPROTO_TCP, + pd->af)) { REASON_SET(&reason, PFRES_PROTCKSUM); - else { - if (th->th_flags & TH_SYN) + } else { + if (th->th_flags & TH_SYN) { ack++; - if (th->th_flags & TH_FIN) + } + if (th->th_flags & TH_FIN) { ack++; - pf_send_tcp(r, af, pd->dst, + } + pf_send_tcp(r, pd->af, pd->dst, pd->src, th->th_dport, th->th_sport, - ntohl(th->th_ack), ack, TH_RST|TH_ACK, 0, 0, + ntohl(th->th_ack), ack, TH_RST | TH_ACK, 0, 0, r->return_ttl, 1, 0, pd->eh, kif->pfik_ifp); } - } else if (pd->proto != IPPROTO_ICMP && af == AF_INET && + } else if (pd->proto != IPPROTO_ICMP && pd->af == AF_INET && pd->proto != IPPROTO_ESP && pd->proto != IPPROTO_AH && - r->return_icmp) - pf_send_icmp(m, r->return_icmp >> 8, - r->return_icmp & 255, af, r); - else if (pd->proto != IPPROTO_ICMPV6 && af == AF_INET6 && + r->return_icmp) { + pf_send_icmp(pbuf, r->return_icmp >> 8, + r->return_icmp & 255, pd->af, r); + } else if (pd->proto != IPPROTO_ICMPV6 && af == AF_INET6 && pd->proto != IPPROTO_ESP && pd->proto != IPPROTO_AH && - r->return_icmp6) - pf_send_icmp(m, r->return_icmp6 >> 8, - r->return_icmp6 & 255, af, r); + r->return_icmp6) { + pf_send_icmp(pbuf, r->return_icmp6 >> 8, + r->return_icmp6 & 255, pd->af, r); + } } - if (r->action == PF_DROP) - return (PF_DROP); + if (r->action == PF_DROP) { + return PF_DROP; + } /* prepare state key, for flowhash and/or the state (if created) */ - bzero(&psk, sizeof (psk)); + bzero(&psk, sizeof(psk)); psk.proto = pd->proto; psk.direction = direction; - psk.af = af; if (pd->proto == IPPROTO_UDP) { if (ntohs(pd->hdr.udp->uh_sport) == PF_IKE_PORT && ntohs(pd->hdr.udp->uh_dport) == PF_IKE_PORT) { psk.proto_variant = PF_EXTFILTER_APD; } else { psk.proto_variant = nr ? nr->extfilter : r->extfilter; - if (psk.proto_variant < PF_EXTFILTER_APD) + if (psk.proto_variant < PF_EXTFILTER_APD) { psk.proto_variant = PF_EXTFILTER_APD; + } } } else if (pd->proto == IPPROTO_GRE) { psk.proto_variant = pd->proto_variant; } if (direction == PF_OUT) { + psk.af_gwy = af; PF_ACPY(&psk.gwy.addr, saddr, af); - PF_ACPY(&psk.ext.addr, daddr, af); + PF_ACPY(&psk.ext_gwy.addr, daddr, af); switch (pd->proto) { - case IPPROTO_UDP: - psk.gwy.xport = sxport; - psk.ext.xport = dxport; - break; case IPPROTO_ESP: psk.gwy.xport.spi = 0; - psk.ext.xport.spi = pd->hdr.esp->spi; + psk.ext_gwy.xport.spi = pd->hdr.esp->spi; break; case IPPROTO_ICMP: #if INET6 case IPPROTO_ICMPV6: #endif + /* + * NAT64 requires protocol translation between ICMPv4 + * and ICMPv6. TCP and UDP do not require protocol + * translation. To avoid adding complexity just to + * handle ICMP(v4addr/v6addr), we always lookup for + * proto = IPPROTO_ICMP on both LAN and WAN side + */ + psk.proto = IPPROTO_ICMP; psk.gwy.xport.port = nxport.port; - psk.ext.xport.spi = 0; + psk.ext_gwy.xport.spi = 0; break; default: psk.gwy.xport = sxport; - psk.ext.xport = dxport; + psk.ext_gwy.xport = dxport; break; } + psk.af_lan = af; if (nr != NULL) { PF_ACPY(&psk.lan.addr, &pd->baddr, af); psk.lan.xport = bxport; + PF_ACPY(&psk.ext_lan.addr, &pd->bdaddr, af); + psk.ext_lan.xport = bdxport; } else { PF_ACPY(&psk.lan.addr, &psk.gwy.addr, af); psk.lan.xport = psk.gwy.xport; + PF_ACPY(&psk.ext_lan.addr, &psk.ext_gwy.addr, af); + psk.ext_lan.xport = psk.ext_gwy.xport; } } else { - PF_ACPY(&psk.lan.addr, daddr, af); - PF_ACPY(&psk.ext.addr, saddr, af); + psk.af_lan = af; + if (nr && nr->action == PF_NAT64) { + PF_ACPY(&psk.lan.addr, &pd->baddr, af); + PF_ACPY(&psk.ext_lan.addr, &pd->bdaddr, af); + } else { + PF_ACPY(&psk.lan.addr, daddr, af); + PF_ACPY(&psk.ext_lan.addr, saddr, af); + } switch (pd->proto) { case IPPROTO_ICMP: #if INET6 case IPPROTO_ICMPV6: #endif - psk.lan.xport = nxport; - psk.ext.xport.spi = 0; + /* + * NAT64 requires protocol translation between ICMPv4 + * and ICMPv6. TCP and UDP do not require protocol + * translation. To avoid adding complexity just to + * handle ICMP(v4addr/v6addr), we always lookup for + * proto = IPPROTO_ICMP on both LAN and WAN side + */ + psk.proto = IPPROTO_ICMP; + if (nr && nr->action == PF_NAT64) { + psk.lan.xport = bxport; + psk.ext_lan.xport = bxport; + } else { + psk.lan.xport = nxport; + psk.ext_lan.xport.spi = 0; + } break; case IPPROTO_ESP: - psk.ext.xport.spi = 0; + psk.ext_lan.xport.spi = 0; psk.lan.xport.spi = pd->hdr.esp->spi; break; default: - psk.lan.xport = dxport; - psk.ext.xport = sxport; + if (nr != NULL) { + if (nr->action == PF_NAT64) { + psk.lan.xport = bxport; + psk.ext_lan.xport = bdxport; + } else { + psk.lan.xport = dxport; + psk.ext_lan.xport = sxport; + } + } else { + psk.lan.xport = dxport; + psk.ext_lan.xport = sxport; + } break; } + psk.af_gwy = pd->naf; if (nr != NULL) { - PF_ACPY(&psk.gwy.addr, &pd->baddr, af); - psk.gwy.xport = bxport; + if (nr->action == PF_NAT64) { + PF_ACPY(&psk.gwy.addr, &pd->naddr, pd->naf); + PF_ACPY(&psk.ext_gwy.addr, &pd->ndaddr, + pd->naf); + if ((pd->proto == IPPROTO_ICMPV6) || + (pd->proto == IPPROTO_ICMP)) { + psk.gwy.xport = nxport; + psk.ext_gwy.xport = nxport; + } else { + psk.gwy.xport = sxport; + psk.ext_gwy.xport = dxport; + } + } else { + PF_ACPY(&psk.gwy.addr, &pd->bdaddr, af); + psk.gwy.xport = bdxport; + PF_ACPY(&psk.ext_gwy.addr, saddr, af); + psk.ext_gwy.xport = sxport; + } } else { PF_ACPY(&psk.gwy.addr, &psk.lan.addr, af); psk.gwy.xport = psk.lan.xport; + PF_ACPY(&psk.ext_gwy.addr, &psk.ext_lan.addr, af); + psk.ext_gwy.xport = psk.ext_lan.xport; } } if (pd->pktflags & PKTF_FLOW_ID) { @@ -4758,35 +5670,36 @@ pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, pd->pktflags &= ~PKTF_FLOW_ADV; } - if (pf_tag_packet(m, pd->pf_mtag, tag, rtableid, pd)) { + if (pf_tag_packet(pbuf, pd->pf_mtag, tag, rtableid, pd)) { REASON_SET(&reason, PFRES_MEMORY); - return (PF_DROP); + return PF_DROP; } if (!state_icmp && (r->keep_state || nr != NULL || (pd->flags & PFDESC_TCP_NORM))) { /* create new state */ - struct pf_state *s = NULL; + struct pf_state *s = NULL; struct pf_state_key *sk = NULL; struct pf_src_node *sn = NULL; struct pf_ike_hdr ike; if (pd->proto == IPPROTO_UDP) { - struct udphdr *uh = pd->hdr.udp; - size_t plen = m->m_pkthdr.len - off - sizeof (*uh); + size_t plen = pbuf->pb_packet_len - off - sizeof(*uh); if (ntohs(uh->uh_sport) == PF_IKE_PORT && ntohs(uh->uh_dport) == PF_IKE_PORT && plen >= PF_IKE_PACKET_MINSIZE) { - if (plen > PF_IKE_PACKET_MINSIZE) + if (plen > PF_IKE_PACKET_MINSIZE) { plen = PF_IKE_PACKET_MINSIZE; - m_copydata(m, off + sizeof (*uh), plen, &ike); + } + pbuf_copy_data(pbuf, off + sizeof(*uh), plen, + &ike); } } if (nr != NULL && pd->proto == IPPROTO_ESP && direction == PF_OUT) { - struct pf_state_key_cmp sk0; + struct pf_state_key_cmp sk0; struct pf_state *s0; /* @@ -4796,11 +5709,11 @@ pf_test_rule(struct pf_rule **rm, struct pf_state **sm, int direction, * different internal address. Only one 'blocking' * partial state is allowed for each external address. */ - memset(&sk0, 0, sizeof (sk0)); - sk0.af = pd->af; + memset(&sk0, 0, sizeof(sk0)); + sk0.af_gwy = pd->af; sk0.proto = IPPROTO_ESP; - PF_ACPY(&sk0.gwy.addr, saddr, sk0.af); - PF_ACPY(&sk0.ext.addr, daddr, sk0.af); + PF_ACPY(&sk0.gwy.addr, saddr, sk0.af_gwy); + PF_ACPY(&sk0.ext_gwy.addr, daddr, sk0.af_gwy); s0 = pf_find_state(kif, &sk0, PF_IN); if (s0 && PF_ANEQ(&s0->state_key->lan.addr, @@ -4850,14 +5763,15 @@ cleanup: pool_put(&pf_src_tree_pl, nsn); } if (sk != NULL) { - if (sk->app_state) + if (sk->app_state) { pool_put(&pf_app_state_pl, sk->app_state); + } pool_put(&pf_state_key_pl, sk); } - return (PF_DROP); + return PF_DROP; } - bzero(s, sizeof (*s)); + bzero(s, sizeof(*s)); TAILQ_INIT(&s->unlink_hooks); s->rule.ptr = r; s->nat_rule.ptr = nr; @@ -4865,26 +5779,29 @@ cleanup: STATE_INC_COUNTERS(s); s->allow_opts = r->allow_opts; s->log = r->log & PF_LOG_ALL; - if (nr != NULL) + if (nr != NULL) { s->log |= nr->log & PF_LOG_ALL; + } switch (pd->proto) { case IPPROTO_TCP: s->src.seqlo = ntohl(th->th_seq); s->src.seqhi = s->src.seqlo + pd->p_len + 1; - if ((th->th_flags & (TH_SYN|TH_ACK)) == + if ((th->th_flags & (TH_SYN | TH_ACK)) == TH_SYN && r->keep_state == PF_STATE_MODULATE) { /* Generate sequence number modulator */ if ((s->src.seqdiff = pf_tcp_iss(pd) - - s->src.seqlo) == 0) + s->src.seqlo) == 0) { s->src.seqdiff = 1; + } pf_change_a(&th->th_seq, &th->th_sum, htonl(s->src.seqlo + s->src.seqdiff), 0); - rewrite = off + sizeof (*th); - } else + rewrite = off + sizeof(*th); + } else { s->src.seqdiff = 0; + } if (th->th_flags & TH_SYN) { s->src.seqhi++; - s->src.wscale = pf_get_wscale(m, off, + s->src.wscale = pf_get_wscale(pbuf, off, th->th_off, af); } s->src.max_win = MAX(ntohs(th->th_win), 1); @@ -4895,8 +5812,9 @@ cleanup: s->src.max_win = (win - 1) >> (s->src.wscale & PF_WSCALE_MASK); } - if (th->th_flags & TH_FIN) + if (th->th_flags & TH_FIN) { s->src.seqhi++; + } s->dst.seqhi = 1; s->dst.max_win = 1; s->src.state = TCPS_SYN_SENT; @@ -4946,16 +5864,16 @@ cleanup: } if (pd->proto == IPPROTO_TCP) { if ((pd->flags & PFDESC_TCP_NORM) && - pf_normalize_tcp_init(m, off, pd, th, &s->src, + pf_normalize_tcp_init(pbuf, off, pd, th, &s->src, &s->dst)) { REASON_SET(&reason, PFRES_MEMORY); pf_src_tree_remove_state(s); STATE_DEC_COUNTERS(s); pool_put(&pf_state_pl, s); - return (PF_DROP); + return PF_DROP; } if ((pd->flags & PFDESC_TCP_NORM) && s->src.scrub && - pf_normalize_tcp_stateful(m, off, pd, &reason, + pf_normalize_tcp_stateful(pbuf, off, pd, &reason, th, s, &s->src, &s->dst, &rewrite)) { /* This really shouldn't happen!!! */ DPFPRINTF(PF_DEBUG_URGENT, @@ -4965,25 +5883,30 @@ cleanup: pf_src_tree_remove_state(s); STATE_DEC_COUNTERS(s); pool_put(&pf_state_pl, s); - return (PF_DROP); + return PF_DROP; } } /* allocate state key and import values from psk */ if ((sk = pf_alloc_state_key(s, &psk)) == NULL) { REASON_SET(&reason, PFRES_MEMORY); + /* + * XXXSCW: This will leak the freshly-allocated + * state structure 's'. Although it should + * eventually be aged-out and removed. + */ goto cleanup; } - pf_set_rt_ifp(s, saddr); /* needs s->state_key set */ + pf_set_rt_ifp(s, saddr, af); /* needs s->state_key set */ - m = pd->mp; + pbuf = pd->mp; // XXXSCW: Why? if (sk->app_state == 0) { switch (pd->proto) { case IPPROTO_TCP: { u_int16_t dport = (direction == PF_OUT) ? - sk->ext.xport.port : sk->gwy.xport.port; + sk->ext_gwy.xport.port : sk->gwy.xport.port; if (nr != NULL && ntohs(dport) == PF_PPTP_PORT) { @@ -4997,7 +5920,7 @@ cleanup: goto cleanup; } - bzero(as, sizeof (*as)); + bzero(as, sizeof(*as)); as->handler = pf_pptp_handler; as->compare_lan_ext = 0; as->compare_ext_gwy = 0; @@ -5010,8 +5933,6 @@ cleanup: } case IPPROTO_UDP: { - struct udphdr *uh = pd->hdr.udp; - if (nr != NULL && ntohs(uh->uh_sport) == PF_IKE_PORT && ntohs(uh->uh_dport) == PF_IKE_PORT) { @@ -5025,7 +5946,7 @@ cleanup: goto cleanup; } - bzero(as, sizeof (*as)); + bzero(as, sizeof(*as)); as->compare_lan_ext = pf_ike_compare; as->compare_ext_gwy = pf_ike_compare; as->u.ike.cookie = ike.initiator_cookie; @@ -5040,49 +5961,52 @@ cleanup: } if (pf_insert_state(BOUND_IFACE(r, kif), s)) { - if (pd->proto == IPPROTO_TCP) + if (pd->proto == IPPROTO_TCP) { pf_normalize_tcp_cleanup(s); + } REASON_SET(&reason, PFRES_STATEINS); pf_src_tree_remove_state(s); STATE_DEC_COUNTERS(s); pool_put(&pf_state_pl, s); - return (PF_DROP); - } else + return PF_DROP; + } else { *sm = s; + } if (tag > 0) { pf_tag_ref(tag); s->tag = tag; } if (pd->proto == IPPROTO_TCP && - (th->th_flags & (TH_SYN|TH_ACK)) == TH_SYN && + (th->th_flags & (TH_SYN | TH_ACK)) == TH_SYN && r->keep_state == PF_STATE_SYNPROXY) { + int ua = (sk->af_lan == sk->af_gwy) ? 1 : 0; s->src.state = PF_TCPS_PROXY_SRC; if (nr != NULL) { if (direction == PF_OUT) { pf_change_ap(direction, pd->mp, saddr, &th->th_sport, pd->ip_sum, &th->th_sum, &pd->baddr, - bxport.port, 0, af); + bxport.port, 0, af, pd->af, ua); sxport.port = th->th_sport; } else { pf_change_ap(direction, pd->mp, daddr, &th->th_dport, pd->ip_sum, &th->th_sum, &pd->baddr, - bxport.port, 0, af); + bxport.port, 0, af, pd->af, ua); sxport.port = th->th_dport; } } s->src.seqhi = htonl(random()); /* Find mss option */ - mss = pf_get_mss(m, off, th->th_off, af); + mss = pf_get_mss(pbuf, off, th->th_off, af); mss = pf_calc_mss(saddr, af, mss); mss = pf_calc_mss(daddr, af, mss); s->src.mss = mss; pf_send_tcp(r, af, daddr, saddr, th->th_dport, th->th_sport, s->src.seqhi, ntohl(th->th_seq) + 1, - TH_SYN|TH_ACK, 0, s->src.mss, 0, 1, 0, NULL, NULL); + TH_SYN | TH_ACK, 0, s->src.mss, 0, 1, 0, NULL, NULL); REASON_SET(&reason, PFRES_SYNPROXY); - return (PF_SYNPROXY_DROP); + return PF_SYNPROXY_DROP; } if (sk->app_state && sk->app_state->handler) { @@ -5105,91 +6029,120 @@ cleanup: pd, kif); if (pd->lmw < 0) { REASON_SET(&reason, PFRES_MEMORY); - return (PF_DROP); + return PF_DROP; } - m = pd->mp; + pbuf = pd->mp; // XXXSCW: Why? } } } /* copy back packet headers if we performed NAT operations */ if (rewrite) { - if (rewrite < off + hdrlen) + if (rewrite < off + hdrlen) { rewrite = off + hdrlen; + } - m = pf_lazy_makewritable(pd, pd->mp, rewrite); - if (!m) { + if (pf_lazy_makewritable(pd, pd->mp, rewrite) == NULL) { REASON_SET(&reason, PFRES_MEMORY); - return (PF_DROP); + return PF_DROP; + } + + pbuf_copy_back(pbuf, off, hdrlen, pd->hdr.any); + if (af == AF_INET6 && pd->naf == AF_INET) { + return pf_nat64_ipv6(pbuf, off, pd); + } else if (af == AF_INET && pd->naf == AF_INET6) { + return pf_nat64_ipv4(pbuf, off, pd); } + } + + return PF_PASS; +} + +boolean_t is_nlc_enabled_glb = FALSE; + +static inline boolean_t +pf_is_dummynet_enabled(void) +{ +#if DUMMYNET + if (__probable(!PF_IS_ENABLED)) { + return FALSE; + } - m_copyback(m, off, hdrlen, pd->hdr.any); + if (__probable(!DUMMYNET_LOADED)) { + return FALSE; } - return (PF_PASS); + if (__probable(TAILQ_EMPTY(pf_main_ruleset. + rules[PF_RULESET_DUMMYNET].active.ptr))) { + return FALSE; + } + + return TRUE; +#else + return FALSE; +#endif /* DUMMYNET */ } #if DUMMYNET /* - * When pf_test_dummynet() returns PF_PASS, the rule matching parameter "rm" + * When pf_test_dummynet() returns PF_PASS, the rule matching parameter "rm" * remains unchanged, meaning the packet did not match a dummynet rule. - * when the packet does match a dummynet rule, pf_test_dummynet() returns - * PF_PASS and zero out the mbuf rule as the packet is effectively siphoned + * when the packet does match a dummynet rule, pf_test_dummynet() returns + * PF_PASS and zero out the mbuf rule as the packet is effectively siphoned * out by dummynet. */ static int -pf_test_dummynet(struct pf_rule **rm, int direction, struct pfi_kif *kif, - struct mbuf **m0, struct pf_pdesc *pd, struct ip_fw_args *fwa) +pf_test_dummynet(struct pf_rule **rm, int direction, struct pfi_kif *kif, + pbuf_t **pbuf0, struct pf_pdesc *pd, struct ip_fw_args *fwa) { - struct mbuf *m = *m0; - struct pf_rule *am = NULL; - struct pf_ruleset *rsm = NULL; - struct pf_addr *saddr = pd->src, *daddr = pd->dst; - sa_family_t af = pd->af; - struct pf_rule *r, *a = NULL; - struct pf_ruleset *ruleset = NULL; - struct tcphdr *th = pd->hdr.tcp; - u_short reason; - int hdrlen = 0; - int tag = -1; - unsigned int rtableid = IFSCOPE_NONE; - int asd = 0; - int match = 0; - u_int8_t icmptype = 0, icmpcode = 0; - struct ip_fw_args dnflow; - struct pf_rule *prev_matching_rule = fwa ? fwa->fwa_pf_rule : NULL; - int found_prev_rule = (prev_matching_rule) ? 0 : 1; - - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); - - if (!DUMMYNET_LOADED) - return (PF_PASS); - - if (TAILQ_EMPTY(pf_main_ruleset.rules[PF_RULESET_DUMMYNET].active.ptr)) - return (PF_PASS); - + pbuf_t *pbuf = *pbuf0; + struct pf_rule *am = NULL; + struct pf_ruleset *rsm = NULL; + struct pf_addr *saddr = pd->src, *daddr = pd->dst; + sa_family_t af = pd->af; + struct pf_rule *r, *a = NULL; + struct pf_ruleset *ruleset = NULL; + struct tcphdr *th = pd->hdr.tcp; + u_short reason; + int hdrlen = 0; + int tag = -1; + unsigned int rtableid = IFSCOPE_NONE; + int asd = 0; + int match = 0; + u_int8_t icmptype = 0, icmpcode = 0; + struct ip_fw_args dnflow; + struct pf_rule *prev_matching_rule = fwa ? fwa->fwa_pf_rule : NULL; + int found_prev_rule = (prev_matching_rule) ? 0 : 1; + + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); + + if (!pf_is_dummynet_enabled()) { + return PF_PASS; + } + bzero(&dnflow, sizeof(dnflow)); hdrlen = 0; /* Fragments don't gave protocol headers */ - if (!(pd->flags & PFDESC_IP_FRAG)) + if (!(pd->flags & PFDESC_IP_FRAG)) { switch (pd->proto) { case IPPROTO_TCP: dnflow.fwa_id.flags = pd->hdr.tcp->th_flags; dnflow.fwa_id.dst_port = ntohs(pd->hdr.tcp->th_dport); dnflow.fwa_id.src_port = ntohs(pd->hdr.tcp->th_sport); - hdrlen = sizeof (*th); + hdrlen = sizeof(*th); break; case IPPROTO_UDP: dnflow.fwa_id.dst_port = ntohs(pd->hdr.udp->uh_dport); dnflow.fwa_id.src_port = ntohs(pd->hdr.udp->uh_sport); - hdrlen = sizeof (*pd->hdr.udp); + hdrlen = sizeof(*pd->hdr.udp); break; #if INET case IPPROTO_ICMP: - if (af != AF_INET) + if (af != AF_INET) { break; + } hdrlen = ICMP_MINLEN; icmptype = pd->hdr.icmp->icmp_type; icmpcode = pd->hdr.icmp->icmp_code; @@ -5197,49 +6150,54 @@ pf_test_dummynet(struct pf_rule **rm, int direction, struct pfi_kif *kif, #endif /* INET */ #if INET6 case IPPROTO_ICMPV6: - if (af != AF_INET6) + if (af != AF_INET6) { break; - hdrlen = sizeof (*pd->hdr.icmp6); + } + hdrlen = sizeof(*pd->hdr.icmp6); icmptype = pd->hdr.icmp6->icmp6_type; icmpcode = pd->hdr.icmp6->icmp6_code; break; #endif /* INET6 */ case IPPROTO_GRE: - if (pd->proto_variant == PF_GRE_PPTP_VARIANT) - hdrlen = sizeof (*pd->hdr.grev1); + if (pd->proto_variant == PF_GRE_PPTP_VARIANT) { + hdrlen = sizeof(*pd->hdr.grev1); + } break; case IPPROTO_ESP: - hdrlen = sizeof (*pd->hdr.esp); + hdrlen = sizeof(*pd->hdr.esp); break; } + } r = TAILQ_FIRST(pf_main_ruleset.rules[PF_RULESET_DUMMYNET].active.ptr); while (r != NULL) { r->evaluations++; - if (pfi_kif_match(r->kif, kif) == r->ifnot) + if (pfi_kif_match(r->kif, kif) == r->ifnot) { r = r->skip[PF_SKIP_IFP].ptr; - else if (r->direction && r->direction != direction) + } else if (r->direction && r->direction != direction) { r = r->skip[PF_SKIP_DIR].ptr; - else if (r->af && r->af != af) + } else if (r->af && r->af != af) { r = r->skip[PF_SKIP_AF].ptr; - else if (r->proto && r->proto != pd->proto) + } else if (r->proto && r->proto != pd->proto) { r = r->skip[PF_SKIP_PROTO].ptr; - else if (PF_MISMATCHAW(&r->src.addr, saddr, af, - r->src.neg, kif)) + } else if (PF_MISMATCHAW(&r->src.addr, saddr, af, + r->src.neg, kif)) { r = r->skip[PF_SKIP_SRC_ADDR].ptr; + } /* tcp/udp only. port_op always 0 in other cases */ - else if (r->proto == pd->proto && + else if (r->proto == pd->proto && (r->proto == IPPROTO_TCP || r->proto == IPPROTO_UDP) && ((pd->flags & PFDESC_IP_FRAG) || ((r->src.xport.range.op && !pf_match_port(r->src.xport.range.op, r->src.xport.range.port[0], r->src.xport.range.port[1], - th->th_sport))))) + th->th_sport))))) { r = r->skip[PF_SKIP_SRC_PORT].ptr; - else if (PF_MISMATCHAW(&r->dst.addr, daddr, af, - r->dst.neg, NULL)) + } else if (PF_MISMATCHAW(&r->dst.addr, daddr, af, + r->dst.neg, NULL)) { r = r->skip[PF_SKIP_DST_ADDR].ptr; + } /* tcp/udp only. port_op always 0 in other cases */ else if (r->proto == pd->proto && (r->proto == IPPROTO_TCP || r->proto == IPPROTO_UDP) && @@ -5247,46 +6205,51 @@ pf_test_dummynet(struct pf_rule **rm, int direction, struct pfi_kif *kif, ((pd->flags & PFDESC_IP_FRAG) || !pf_match_port(r->dst.xport.range.op, r->dst.xport.range.port[0], r->dst.xport.range.port[1], - th->th_dport))) + th->th_dport))) { r = r->skip[PF_SKIP_DST_PORT].ptr; + } /* icmp only. type always 0 in other cases */ - else if (r->type && - ((pd->flags & PFDESC_IP_FRAG) || - r->type != icmptype + 1)) + else if (r->type && + ((pd->flags & PFDESC_IP_FRAG) || + r->type != icmptype + 1)) { r = TAILQ_NEXT(r, entries); + } /* icmp only. type always 0 in other cases */ - else if (r->code && - ((pd->flags & PFDESC_IP_FRAG) || - r->code != icmpcode + 1)) + else if (r->code && + ((pd->flags & PFDESC_IP_FRAG) || + r->code != icmpcode + 1)) { r = TAILQ_NEXT(r, entries); - else if (r->tos && !(r->tos == pd->tos)) + } else if (r->tos && !(r->tos == pd->tos)) { r = TAILQ_NEXT(r, entries); - else if (r->rule_flag & PFRULE_FRAGMENT) + } else if (r->rule_flag & PFRULE_FRAGMENT) { r = TAILQ_NEXT(r, entries); - else if (pd->proto == IPPROTO_TCP && + } else if (pd->proto == IPPROTO_TCP && ((pd->flags & PFDESC_IP_FRAG) || - (r->flagset & th->th_flags) != r->flags)) + (r->flagset & th->th_flags) != r->flags)) { r = TAILQ_NEXT(r, entries); - else if (r->prob && r->prob <= (RandomULong() % (UINT_MAX - 1) + 1)) + } else if (r->prob && r->prob <= (RandomULong() % (UINT_MAX - 1) + 1)) { r = TAILQ_NEXT(r, entries); - else if (r->match_tag && !pf_match_tag(m, r, pd->pf_mtag, &tag)) + } else if (r->match_tag && !pf_match_tag(r, pd->pf_mtag, &tag)) { r = TAILQ_NEXT(r, entries); - else { - /* - * Need to go past the previous dummynet matching rule + } else { + /* + * Need to go past the previous dummynet matching rule */ if (r->anchor == NULL) { if (found_prev_rule) { - if (r->tag) + if (r->tag) { tag = r->tag; - if (PF_RTABLEID_IS_VALID(r->rtableid)) + } + if (PF_RTABLEID_IS_VALID(r->rtableid)) { rtableid = r->rtableid; + } match = 1; *rm = r; am = a; rsm = ruleset; - if ((*rm)->quick) + if ((*rm)->quick) { break; + } } else if (r == prev_matching_rule) { found_prev_rule = 1; } @@ -5297,66 +6260,69 @@ pf_test_dummynet(struct pf_rule **rm, int direction, struct pfi_kif *kif, } } if (r == NULL && pf_step_out_of_anchor(&asd, &ruleset, - PF_RULESET_DUMMYNET, &r, &a, &match)) + PF_RULESET_DUMMYNET, &r, &a, &match)) { break; + } } r = *rm; a = am; ruleset = rsm; - if (!match) - return (PF_PASS); + if (!match) { + return PF_PASS; + } REASON_SET(&reason, PFRES_DUMMYNET); if (r->log) { - PFLOG_PACKET(kif, h, m, af, direction, reason, r, + PFLOG_PACKET(kif, h, pbuf, af, direction, reason, r, a, ruleset, pd); } if (r->action == PF_NODUMMYNET) { int dirndx = (direction == PF_OUT); - + r->packets[dirndx]++; r->bytes[dirndx] += pd->tot_len; - return (PF_PASS); + return PF_PASS; } - if (pf_tag_packet(m, pd->pf_mtag, tag, rtableid, pd)) { + if (pf_tag_packet(pbuf, pd->pf_mtag, tag, rtableid, pd)) { REASON_SET(&reason, PFRES_MEMORY); - return (PF_DROP); + return PF_DROP; } if (r->dnpipe && ip_dn_io_ptr != NULL) { + struct mbuf *m; int dirndx = (direction == PF_OUT); - + r->packets[dirndx]++; r->bytes[dirndx] += pd->tot_len; - + dnflow.fwa_cookie = r->dnpipe; dnflow.fwa_pf_rule = r; dnflow.fwa_id.proto = pd->proto; dnflow.fwa_flags = r->dntype; switch (af) { - case AF_INET: - dnflow.fwa_id.addr_type = 4; - dnflow.fwa_id.src_ip = ntohl(saddr->v4.s_addr); - dnflow.fwa_id.dst_ip = ntohl(daddr->v4.s_addr); - break; - case AF_INET6: - dnflow.fwa_id.addr_type = 6; - dnflow.fwa_id.src_ip6 = saddr->v6; - dnflow.fwa_id.dst_ip6 = saddr->v6; - break; - } + case AF_INET: + dnflow.fwa_id.addr_type = 4; + dnflow.fwa_id.src_ip = ntohl(saddr->v4addr.s_addr); + dnflow.fwa_id.dst_ip = ntohl(daddr->v4addr.s_addr); + break; + case AF_INET6: + dnflow.fwa_id.addr_type = 6; + dnflow.fwa_id.src_ip6 = saddr->v6addr; + dnflow.fwa_id.dst_ip6 = saddr->v6addr; + break; + } if (fwa != NULL) { dnflow.fwa_oif = fwa->fwa_oif; dnflow.fwa_oflags = fwa->fwa_oflags; /* - * Note that fwa_ro, fwa_dst and fwa_ipoa are - * actually in a union so the following does work + * Note that fwa_ro, fwa_dst and fwa_ipoa are + * actually in a union so the following does work * for both IPv4 and IPv6 */ dnflow.fwa_ro = fwa->fwa_ro; @@ -5365,113 +6331,117 @@ pf_test_dummynet(struct pf_rule **rm, int direction, struct pfi_kif *kif, dnflow.fwa_ro6_pmtu = fwa->fwa_ro6_pmtu; dnflow.fwa_origifp = fwa->fwa_origifp; dnflow.fwa_mtu = fwa->fwa_mtu; - dnflow.fwa_alwaysfrag = fwa->fwa_alwaysfrag; dnflow.fwa_unfragpartlen = fwa->fwa_unfragpartlen; dnflow.fwa_exthdrs = fwa->fwa_exthdrs; } - + if (af == AF_INET) { - struct ip *iphdr = mtod(m, struct ip *); + struct ip *iphdr = pbuf->pb_data; NTOHS(iphdr->ip_len); NTOHS(iphdr->ip_off); } - /* - * Don't need to unlock pf_lock as NET_THREAD_HELD_PF + /* + * Don't need to unlock pf_lock as NET_THREAD_HELD_PF * allows for recursive behavior */ - ip_dn_io_ptr(m, - dnflow.fwa_cookie, - af == AF_INET ? - direction == PF_IN ? DN_TO_IP_IN : DN_TO_IP_OUT : - direction == PF_IN ? DN_TO_IP6_IN : DN_TO_IP6_OUT, - &dnflow, DN_CLIENT_PF); - + m = pbuf_to_mbuf(pbuf, TRUE); + if (m != NULL) { + ip_dn_io_ptr(m, + dnflow.fwa_cookie, (af == AF_INET) ? + ((direction == PF_IN) ? DN_TO_IP_IN : DN_TO_IP_OUT) : + ((direction == PF_IN) ? DN_TO_IP6_IN : DN_TO_IP6_OUT), + &dnflow, DN_CLIENT_PF); + } + /* - * The packet is siphoned out by dummynet so return a NULL - * mbuf so the caller can still return success. + * The packet is siphoned out by dummynet so return a NULL + * pbuf so the caller can still return success. */ - *m0 = NULL; - - return (PF_PASS); + *pbuf0 = NULL; + + return PF_PASS; } - return (PF_PASS); + return PF_PASS; } #endif /* DUMMYNET */ static int pf_test_fragment(struct pf_rule **rm, int direction, struct pfi_kif *kif, - struct mbuf *m, void *h, struct pf_pdesc *pd, struct pf_rule **am, + pbuf_t *pbuf, void *h, struct pf_pdesc *pd, struct pf_rule **am, struct pf_ruleset **rsm) { #pragma unused(h) - struct pf_rule *r, *a = NULL; - struct pf_ruleset *ruleset = NULL; - sa_family_t af = pd->af; - u_short reason; - int tag = -1; - int asd = 0; - int match = 0; + struct pf_rule *r, *a = NULL; + struct pf_ruleset *ruleset = NULL; + sa_family_t af = pd->af; + u_short reason; + int tag = -1; + int asd = 0; + int match = 0; r = TAILQ_FIRST(pf_main_ruleset.rules[PF_RULESET_FILTER].active.ptr); while (r != NULL) { r->evaluations++; - if (pfi_kif_match(r->kif, kif) == r->ifnot) + if (pfi_kif_match(r->kif, kif) == r->ifnot) { r = r->skip[PF_SKIP_IFP].ptr; - else if (r->direction && r->direction != direction) + } else if (r->direction && r->direction != direction) { r = r->skip[PF_SKIP_DIR].ptr; - else if (r->af && r->af != af) + } else if (r->af && r->af != af) { r = r->skip[PF_SKIP_AF].ptr; - else if (r->proto && r->proto != pd->proto) + } else if (r->proto && r->proto != pd->proto) { r = r->skip[PF_SKIP_PROTO].ptr; - else if (PF_MISMATCHAW(&r->src.addr, pd->src, af, - r->src.neg, kif)) + } else if (PF_MISMATCHAW(&r->src.addr, pd->src, af, + r->src.neg, kif)) { r = r->skip[PF_SKIP_SRC_ADDR].ptr; - else if (PF_MISMATCHAW(&r->dst.addr, pd->dst, af, - r->dst.neg, NULL)) + } else if (PF_MISMATCHAW(&r->dst.addr, pd->dst, af, + r->dst.neg, NULL)) { r = r->skip[PF_SKIP_DST_ADDR].ptr; - else if ((r->rule_flag & PFRULE_TOS) && r->tos && - !(r->tos & pd->tos)) + } else if ((r->rule_flag & PFRULE_TOS) && r->tos && + !(r->tos & pd->tos)) { r = TAILQ_NEXT(r, entries); - else if ((r->rule_flag & PFRULE_DSCP) && r->tos && - !(r->tos & (pd->tos & DSCP_MASK))) + } else if ((r->rule_flag & PFRULE_DSCP) && r->tos && + !(r->tos & (pd->tos & DSCP_MASK))) { r = TAILQ_NEXT(r, entries); - else if ((r->rule_flag & PFRULE_SC) && r->tos && - ((r->tos & SCIDX_MASK) != pd->sc)) + } else if ((r->rule_flag & PFRULE_SC) && r->tos && + ((r->tos & SCIDX_MASK) != pd->sc)) { r = TAILQ_NEXT(r, entries); - else if (r->os_fingerprint != PF_OSFP_ANY) + } else if (r->os_fingerprint != PF_OSFP_ANY) { r = TAILQ_NEXT(r, entries); - else if (pd->proto == IPPROTO_UDP && - (r->src.xport.range.op || r->dst.xport.range.op)) + } else if (pd->proto == IPPROTO_UDP && + (r->src.xport.range.op || r->dst.xport.range.op)) { r = TAILQ_NEXT(r, entries); - else if (pd->proto == IPPROTO_TCP && + } else if (pd->proto == IPPROTO_TCP && (r->src.xport.range.op || r->dst.xport.range.op || - r->flagset)) + r->flagset)) { r = TAILQ_NEXT(r, entries); - else if ((pd->proto == IPPROTO_ICMP || + } else if ((pd->proto == IPPROTO_ICMP || pd->proto == IPPROTO_ICMPV6) && - (r->type || r->code)) + (r->type || r->code)) { r = TAILQ_NEXT(r, entries); - else if (r->prob && r->prob <= (RandomULong() % (UINT_MAX - 1) + 1)) + } else if (r->prob && r->prob <= (RandomULong() % (UINT_MAX - 1) + 1)) { r = TAILQ_NEXT(r, entries); - else if (r->match_tag && !pf_match_tag(m, r, pd->pf_mtag, &tag)) + } else if (r->match_tag && !pf_match_tag(r, pd->pf_mtag, &tag)) { r = TAILQ_NEXT(r, entries); - else { + } else { if (r->anchor == NULL) { match = 1; *rm = r; *am = a; *rsm = ruleset; - if ((*rm)->quick) + if ((*rm)->quick) { break; + } r = TAILQ_NEXT(r, entries); - } else + } else { pf_step_into_anchor(&asd, &ruleset, PF_RULESET_FILTER, &r, &a, &match); + } } if (r == NULL && pf_step_out_of_anchor(&asd, &ruleset, - PF_RULESET_FILTER, &r, &a, &match)) + PF_RULESET_FILTER, &r, &a, &match)) { break; + } } r = *rm; a = *am; @@ -5479,19 +6449,21 @@ pf_test_fragment(struct pf_rule **rm, int direction, struct pfi_kif *kif, REASON_SET(&reason, PFRES_MATCH); - if (r->log) - PFLOG_PACKET(kif, h, m, af, direction, reason, r, a, ruleset, + if (r->log) { + PFLOG_PACKET(kif, h, pbuf, af, direction, reason, r, a, ruleset, pd); + } - if (r->action != PF_PASS) - return (PF_DROP); + if (r->action != PF_PASS) { + return PF_DROP; + } - if (pf_tag_packet(m, pd->pf_mtag, tag, -1, NULL)) { + if (pf_tag_packet(pbuf, pd->pf_mtag, tag, -1, NULL)) { REASON_SET(&reason, PFRES_MEMORY); - return (PF_DROP); + return PF_DROP; } - return (PF_PASS); + return PF_PASS; } static void @@ -5502,7 +6474,7 @@ pf_pptp_handler(struct pf_state *s, int direction, int off, struct tcphdr *th; struct pf_pptp_state *pptps; struct pf_pptp_ctrl_msg cm; - size_t plen; + size_t plen, tlen; struct pf_state *gs; u_int16_t ct; u_int16_t *pac_call_id; @@ -5511,7 +6483,7 @@ pf_pptp_handler(struct pf_state *s, int direction, int off, u_int8_t *pac_state; u_int8_t *pns_state; enum { PF_PPTP_PASS, PF_PPTP_INSERT_GRE, PF_PPTP_REMOVE_GRE } op; - struct mbuf *m; + pbuf_t *pbuf; struct pf_state_key *sk; struct pf_state_key *gsk; struct pf_app_state *gas; @@ -5520,30 +6492,62 @@ pf_pptp_handler(struct pf_state *s, int direction, int off, pptps = &sk->app_state->u.pptp; gs = pptps->grev1_state; - if (gs) + if (gs) { gs->expire = pf_time_second(); + } - m = pd->mp; - plen = min(sizeof (cm), m->m_pkthdr.len - off); - if (plen < PF_PPTP_CTRL_MSG_MINSIZE) + pbuf = pd->mp; + plen = min(sizeof(cm), pbuf->pb_packet_len - off); + if (plen < PF_PPTP_CTRL_MSG_MINSIZE) { return; + } + tlen = plen - PF_PPTP_CTRL_MSG_MINSIZE; + pbuf_copy_data(pbuf, off, plen, &cm); - m_copydata(m, off, plen, &cm); - - if (ntohl(cm.hdr.magic) != PF_PPTP_MAGIC_NUMBER) + if (ntohl(cm.hdr.magic) != PF_PPTP_MAGIC_NUMBER) { + return; + } + if (ntohs(cm.hdr.type) != 1) { return; - if (ntohs(cm.hdr.type) != 1) + } + +#define TYPE_LEN_CHECK(_type, _name) \ + case PF_PPTP_CTRL_TYPE_##_type: \ + if (tlen < sizeof(struct pf_pptp_ctrl_##_name)) \ + return; \ + break; + + switch (cm.ctrl.type) { + TYPE_LEN_CHECK(START_REQ, start_req); + TYPE_LEN_CHECK(START_RPY, start_rpy); + TYPE_LEN_CHECK(STOP_REQ, stop_req); + TYPE_LEN_CHECK(STOP_RPY, stop_rpy); + TYPE_LEN_CHECK(ECHO_REQ, echo_req); + TYPE_LEN_CHECK(ECHO_RPY, echo_rpy); + TYPE_LEN_CHECK(CALL_OUT_REQ, call_out_req); + TYPE_LEN_CHECK(CALL_OUT_RPY, call_out_rpy); + TYPE_LEN_CHECK(CALL_IN_1ST, call_in_1st); + TYPE_LEN_CHECK(CALL_IN_2ND, call_in_2nd); + TYPE_LEN_CHECK(CALL_IN_3RD, call_in_3rd); + TYPE_LEN_CHECK(CALL_CLR, call_clr); + TYPE_LEN_CHECK(CALL_DISC, call_disc); + TYPE_LEN_CHECK(ERROR, error); + TYPE_LEN_CHECK(SET_LINKINFO, set_linkinfo); + default: return; + } +#undef TYPE_LEN_CHECK if (!gs) { gs = pool_get(&pf_state_pl, PR_WAITOK); - if (!gs) + if (!gs) { return; + } - memcpy(gs, s, sizeof (*gs)); + memcpy(gs, s, sizeof(*gs)); - memset(&gs->entry_id, 0, sizeof (gs->entry_id)); - memset(&gs->entry_list, 0, sizeof (gs->entry_list)); + memset(&gs->entry_id, 0, sizeof(gs->entry_id)); + memset(&gs->entry_list, 0, sizeof(gs->entry_list)); TAILQ_INIT(&gs->unlink_hooks); gs->rt_kif = NULL; @@ -5569,19 +6573,22 @@ pf_pptp_handler(struct pf_state *s, int direction, int off, return; } - memcpy(&gsk->lan, &sk->lan, sizeof (gsk->lan)); - memcpy(&gsk->gwy, &sk->gwy, sizeof (gsk->gwy)); - memcpy(&gsk->ext, &sk->ext, sizeof (gsk->ext)); - gsk->af = sk->af; + memcpy(&gsk->lan, &sk->lan, sizeof(gsk->lan)); + memcpy(&gsk->gwy, &sk->gwy, sizeof(gsk->gwy)); + memcpy(&gsk->ext_lan, &sk->ext_lan, sizeof(gsk->ext_lan)); + memcpy(&gsk->ext_gwy, &sk->ext_gwy, sizeof(gsk->ext_gwy)); + gsk->af_lan = sk->af_lan; + gsk->af_gwy = sk->af_gwy; gsk->proto = IPPROTO_GRE; gsk->proto_variant = PF_GRE_PPTP_VARIANT; gsk->app_state = gas; gsk->lan.xport.call_id = 0; gsk->gwy.xport.call_id = 0; - gsk->ext.xport.call_id = 0; + gsk->ext_lan.xport.call_id = 0; + gsk->ext_gwy.xport.call_id = 0; gsk->flowsrc = FLOWSRC_PF; gsk->flowhash = pf_calc_state_key_flowhash(gsk); - memset(gas, 0, sizeof (*gas)); + memset(gas, 0, sizeof(*gas)); gas->u.grev1.pptp_state = s; STATE_INC_COUNTERS(gs); pptps->grev1_state = gs; @@ -5593,7 +6600,7 @@ pf_pptp_handler(struct pf_state *s, int direction, int off, switch (sk->direction) { case PF_IN: - pns_call_id = &gsk->ext.xport.call_id; + pns_call_id = &gsk->ext_lan.xport.call_id; pns_state = &gs->dst.state; pac_call_id = &gsk->lan.xport.call_id; pac_state = &gs->src.state; @@ -5602,7 +6609,7 @@ pf_pptp_handler(struct pf_state *s, int direction, int off, case PF_OUT: pns_call_id = &gsk->lan.xport.call_id; pns_state = &gs->src.state; - pac_call_id = &gsk->ext.xport.call_id; + pac_call_id = &gsk->ext_lan.xport.call_id; pac_state = &gs->dst.state; break; @@ -5621,17 +6628,19 @@ pf_pptp_handler(struct pf_state *s, int direction, int off, case PF_PPTP_CTRL_TYPE_CALL_OUT_REQ: *pns_call_id = cm.msg.call_out_req.call_id; *pns_state = PFGRE1S_INITIATING; - if (s->nat_rule.ptr && pns_call_id == &gsk->lan.xport.call_id) + if (s->nat_rule.ptr && pns_call_id == &gsk->lan.xport.call_id) { spoof_call_id = &cm.msg.call_out_req.call_id; + } break; case PF_PPTP_CTRL_TYPE_CALL_OUT_RPY: *pac_call_id = cm.msg.call_out_rpy.call_id; - if (s->nat_rule.ptr) + if (s->nat_rule.ptr) { spoof_call_id = (pac_call_id == &gsk->lan.xport.call_id) ? &cm.msg.call_out_rpy.call_id : &cm.msg.call_out_rpy.peer_call_id; + } if (gs->timeout == PFTM_UNLINKED) { *pac_state = PFGRE1S_INITIATING; op = PF_PPTP_INSERT_GRE; @@ -5641,48 +6650,56 @@ pf_pptp_handler(struct pf_state *s, int direction, int off, case PF_PPTP_CTRL_TYPE_CALL_IN_1ST: *pns_call_id = cm.msg.call_in_1st.call_id; *pns_state = PFGRE1S_INITIATING; - if (s->nat_rule.ptr && pns_call_id == &gsk->lan.xport.call_id) + if (s->nat_rule.ptr && pns_call_id == &gsk->lan.xport.call_id) { spoof_call_id = &cm.msg.call_in_1st.call_id; + } break; case PF_PPTP_CTRL_TYPE_CALL_IN_2ND: *pac_call_id = cm.msg.call_in_2nd.call_id; *pac_state = PFGRE1S_INITIATING; - if (s->nat_rule.ptr) + if (s->nat_rule.ptr) { spoof_call_id = (pac_call_id == &gsk->lan.xport.call_id) ? &cm.msg.call_in_2nd.call_id : &cm.msg.call_in_2nd.peer_call_id; + } break; case PF_PPTP_CTRL_TYPE_CALL_IN_3RD: - if (s->nat_rule.ptr && pns_call_id == &gsk->lan.xport.call_id) + if (s->nat_rule.ptr && pns_call_id == &gsk->lan.xport.call_id) { spoof_call_id = &cm.msg.call_in_3rd.call_id; + } if (cm.msg.call_in_3rd.call_id != *pns_call_id) { break; } - if (gs->timeout == PFTM_UNLINKED) + if (gs->timeout == PFTM_UNLINKED) { op = PF_PPTP_INSERT_GRE; + } break; case PF_PPTP_CTRL_TYPE_CALL_CLR: - if (cm.msg.call_clr.call_id != *pns_call_id) + if (cm.msg.call_clr.call_id != *pns_call_id) { op = PF_PPTP_REMOVE_GRE; + } break; case PF_PPTP_CTRL_TYPE_CALL_DISC: - if (cm.msg.call_clr.call_id != *pac_call_id) + if (cm.msg.call_clr.call_id != *pac_call_id) { op = PF_PPTP_REMOVE_GRE; + } break; case PF_PPTP_CTRL_TYPE_ERROR: - if (s->nat_rule.ptr && pns_call_id == &gsk->lan.xport.call_id) + if (s->nat_rule.ptr && pns_call_id == &gsk->lan.xport.call_id) { spoof_call_id = &cm.msg.error.peer_call_id; + } break; case PF_PPTP_CTRL_TYPE_SET_LINKINFO: - if (s->nat_rule.ptr && pac_call_id == &gsk->lan.xport.call_id) + if (s->nat_rule.ptr && pac_call_id == &gsk->lan.xport.call_id) { spoof_call_id = &cm.msg.set_linkinfo.peer_call_id; + } break; default: @@ -5697,13 +6714,13 @@ pf_pptp_handler(struct pf_state *s, int direction, int off, int n = 0; struct pf_state_key_cmp key; - key.af = gsk->af; + key.af_gwy = gsk->af_gwy; key.proto = IPPROTO_GRE; key.proto_variant = PF_GRE_PPTP_VARIANT; - PF_ACPY(&key.gwy.addr, &gsk->gwy.addr, key.af); - PF_ACPY(&key.ext.addr, &gsk->ext.addr, key.af); + PF_ACPY(&key.gwy.addr, &gsk->gwy.addr, key.af_gwy); + PF_ACPY(&key.ext_gwy.addr, &gsk->ext_gwy.addr, key.af_gwy); key.gwy.xport.call_id = gsk->gwy.xport.call_id; - key.ext.xport.call_id = gsk->ext.xport.call_id; + key.ext_gwy.xport.call_id = gsk->ext_gwy.xport.call_id; do { call_id = htonl(random()); } while (!call_id); @@ -5711,7 +6728,9 @@ pf_pptp_handler(struct pf_state *s, int direction, int off, while (pf_find_state_all(&key, PF_IN, 0)) { call_id = ntohs(call_id); --call_id; - if (--call_id == 0) call_id = 0xffff; + if (--call_id == 0) { + call_id = 0xffff; + } call_id = htons(call_id); key.gwy.xport.call_id = call_id; @@ -5742,14 +6761,13 @@ pf_pptp_handler(struct pf_state *s, int direction, int off, gsk->lan.xport.call_id, gsk->gwy.xport.call_id, 0); } - m = pf_lazy_makewritable(pd, m, off + plen); - if (!m) { + if (pf_lazy_makewritable(pd, pbuf, off + plen) == NULL) { pptps->grev1_state = NULL; STATE_DEC_COUNTERS(gs); pool_put(&pf_state_pl, gs); return; } - m_copyback(m, off, plen, &cm); + pbuf_copy_back(pbuf, off, plen, &cm); } switch (op) { @@ -5758,7 +6776,8 @@ pf_pptp_handler(struct pf_state *s, int direction, int off, gs->src.state = gs->dst.state = PFGRE1S_NO_TRAFFIC; gsk->lan.xport.call_id = 0; gsk->gwy.xport.call_id = 0; - gsk->ext.xport.call_id = 0; + gsk->ext_lan.xport.call_id = 0; + gsk->ext_gwy.xport.call_id = 0; gs->id = gs->creatorid = 0; break; @@ -5774,9 +6793,8 @@ pf_pptp_handler(struct pf_state *s, int direction, int off, ++gs->nat_src_node->states; VERIFY(gs->nat_src_node->states != 0); } - pf_set_rt_ifp(gs, &sk->lan.addr); + pf_set_rt_ifp(gs, &sk->lan.addr, sk->af_lan); if (pf_insert_state(BOUND_IFACE(s->rule.ptr, kif), gs)) { - /* * * FIX ME: insertion can fail when multiple PNS @@ -5789,7 +6807,7 @@ pf_pptp_handler(struct pf_state *s, int direction, int off, * that fixing this is a low priority. */ pptps->grev1_state = NULL; - pd->lmw = -1; /* Force PF_DROP on PFRES_MEMORY */ + pd->lmw = -1; /* Force PF_DROP on PFRES_MEMORY */ pf_src_tree_remove_state(gs); STATE_DEC_COUNTERS(gs); pool_put(&pf_state_pl, gs); @@ -5812,8 +6830,9 @@ pf_pptp_unlink(struct pf_state *s) if (grev1s) { struct pf_app_state *gas = grev1s->state_key->app_state; - if (grev1s->timeout < PFTM_MAX) + if (grev1s->timeout < PFTM_MAX) { grev1s->timeout = PFTM_PURGE; + } gas->u.grev1.pptp_state = NULL; as->u.pptp.grev1_state = NULL; } @@ -5837,42 +6856,87 @@ static int pf_ike_compare(struct pf_app_state *a, struct pf_app_state *b) { int64_t d = a->u.ike.cookie - b->u.ike.cookie; - return ((d > 0) ? 1 : ((d < 0) ? -1 : 0)); + return (d > 0) ? 1 : ((d < 0) ? -1 : 0); +} + +static int +pf_do_nat64(struct pf_state_key *sk, struct pf_pdesc *pd, pbuf_t *pbuf, + int off) +{ + if (pd->af == AF_INET) { + if (pd->af != sk->af_lan) { + pd->ndaddr = sk->lan.addr; + pd->naddr = sk->ext_lan.addr; + } else { + pd->naddr = sk->gwy.addr; + pd->ndaddr = sk->ext_gwy.addr; + } + return pf_nat64_ipv4(pbuf, off, pd); + } else if (pd->af == AF_INET6) { + if (pd->af != sk->af_lan) { + pd->ndaddr = sk->lan.addr; + pd->naddr = sk->ext_lan.addr; + } else { + pd->naddr = sk->gwy.addr; + pd->ndaddr = sk->ext_gwy.addr; + } + return pf_nat64_ipv6(pbuf, off, pd); + } + return PF_DROP; } static int pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, - struct mbuf *m, int off, void *h, struct pf_pdesc *pd, + pbuf_t *pbuf, int off, void *h, struct pf_pdesc *pd, u_short *reason) { #pragma unused(h) - struct pf_state_key_cmp key; - struct tcphdr *th = pd->hdr.tcp; - u_int16_t win = ntohs(th->th_win); - u_int32_t ack, end, seq, orig_seq; - u_int8_t sws, dws; - int ackskew; - int copyback = 0; - struct pf_state_peer *src, *dst; + struct pf_state_key_cmp key; + struct tcphdr *th = pd->hdr.tcp; + u_int16_t win = ntohs(th->th_win); + u_int32_t ack, end, seq, orig_seq; + u_int8_t sws, dws; + int ackskew; + int copyback = 0; + struct pf_state_peer *src, *dst; + struct pf_state_key *sk; key.app_state = 0; - key.af = pd->af; key.proto = IPPROTO_TCP; - if (direction == PF_IN) { - PF_ACPY(&key.ext.addr, pd->src, key.af); - PF_ACPY(&key.gwy.addr, pd->dst, key.af); - key.ext.xport.port = th->th_sport; - key.gwy.xport.port = th->th_dport; - } else { - PF_ACPY(&key.lan.addr, pd->src, key.af); - PF_ACPY(&key.ext.addr, pd->dst, key.af); - key.lan.xport.port = th->th_sport; - key.ext.xport.port = th->th_dport; - } + key.af_lan = key.af_gwy = pd->af; + + /* + * For NAT64 the first time rule search and state creation + * is done on the incoming side only. + * Once the state gets created, NAT64's LAN side (ipv6) will + * not be able to find the state in ext-gwy tree as that normally + * is intended to be looked up for incoming traffic from the + * WAN side. + * Therefore to handle NAT64 case we init keys here for both + * lan-ext as well as ext-gwy trees. + * In the state lookup we attempt a lookup on both trees if + * first one does not return any result and return a match if + * the match state's was created by NAT64 rule. + */ + PF_ACPY(&key.ext_gwy.addr, pd->src, key.af_gwy); + PF_ACPY(&key.gwy.addr, pd->dst, key.af_gwy); + key.ext_gwy.xport.port = th->th_sport; + key.gwy.xport.port = th->th_dport; + + PF_ACPY(&key.lan.addr, pd->src, key.af_lan); + PF_ACPY(&key.ext_lan.addr, pd->dst, key.af_lan); + key.lan.xport.port = th->th_sport; + key.ext_lan.xport.port = th->th_dport; STATE_LOOKUP(); - if (direction == (*state)->state_key->direction) { + sk = (*state)->state_key; + /* + * In case of NAT64 the translation is first applied on the LAN + * side. Therefore for stack's address family comparison + * we use sk->af_lan. + */ + if ((direction == sk->direction) && (pd->af == sk->af_lan)) { src = &(*state)->src; dst = &(*state)->dst; } else { @@ -5880,96 +6944,98 @@ pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, dst = &(*state)->src; } - if ((*state)->src.state == PF_TCPS_PROXY_SRC) { - if (direction != (*state)->state_key->direction) { + if (src->state == PF_TCPS_PROXY_SRC) { + if (direction != sk->direction) { REASON_SET(reason, PFRES_SYNPROXY); - return (PF_SYNPROXY_DROP); + return PF_SYNPROXY_DROP; } if (th->th_flags & TH_SYN) { - if (ntohl(th->th_seq) != (*state)->src.seqlo) { + if (ntohl(th->th_seq) != src->seqlo) { REASON_SET(reason, PFRES_SYNPROXY); - return (PF_DROP); + return PF_DROP; } pf_send_tcp((*state)->rule.ptr, pd->af, pd->dst, pd->src, th->th_dport, th->th_sport, - (*state)->src.seqhi, ntohl(th->th_seq) + 1, - TH_SYN|TH_ACK, 0, (*state)->src.mss, 0, 1, + src->seqhi, ntohl(th->th_seq) + 1, + TH_SYN | TH_ACK, 0, src->mss, 0, 1, 0, NULL, NULL); REASON_SET(reason, PFRES_SYNPROXY); - return (PF_SYNPROXY_DROP); + return PF_SYNPROXY_DROP; } else if (!(th->th_flags & TH_ACK) || - (ntohl(th->th_ack) != (*state)->src.seqhi + 1) || - (ntohl(th->th_seq) != (*state)->src.seqlo + 1)) { + (ntohl(th->th_ack) != src->seqhi + 1) || + (ntohl(th->th_seq) != src->seqlo + 1)) { REASON_SET(reason, PFRES_SYNPROXY); - return (PF_DROP); + return PF_DROP; } else if ((*state)->src_node != NULL && pf_src_connlimit(state)) { REASON_SET(reason, PFRES_SRCLIMIT); - return (PF_DROP); - } else - (*state)->src.state = PF_TCPS_PROXY_DST; + return PF_DROP; + } else { + src->state = PF_TCPS_PROXY_DST; + } } - if ((*state)->src.state == PF_TCPS_PROXY_DST) { + if (src->state == PF_TCPS_PROXY_DST) { struct pf_state_host *psrc, *pdst; if (direction == PF_OUT) { - psrc = &(*state)->state_key->gwy; - pdst = &(*state)->state_key->ext; + psrc = &sk->gwy; + pdst = &sk->ext_gwy; } else { - psrc = &(*state)->state_key->ext; - pdst = &(*state)->state_key->lan; + psrc = &sk->ext_lan; + pdst = &sk->lan; } - if (direction == (*state)->state_key->direction) { - if (((th->th_flags & (TH_SYN|TH_ACK)) != TH_ACK) || - (ntohl(th->th_ack) != (*state)->src.seqhi + 1) || - (ntohl(th->th_seq) != (*state)->src.seqlo + 1)) { + if (direction == sk->direction) { + if (((th->th_flags & (TH_SYN | TH_ACK)) != TH_ACK) || + (ntohl(th->th_ack) != src->seqhi + 1) || + (ntohl(th->th_seq) != src->seqlo + 1)) { REASON_SET(reason, PFRES_SYNPROXY); - return (PF_DROP); + return PF_DROP; + } + src->max_win = MAX(ntohs(th->th_win), 1); + if (dst->seqhi == 1) { + dst->seqhi = htonl(random()); } - (*state)->src.max_win = MAX(ntohs(th->th_win), 1); - if ((*state)->dst.seqhi == 1) - (*state)->dst.seqhi = htonl(random()); pf_send_tcp((*state)->rule.ptr, pd->af, &psrc->addr, &pdst->addr, psrc->xport.port, pdst->xport.port, - (*state)->dst.seqhi, 0, TH_SYN, 0, - (*state)->src.mss, 0, 0, (*state)->tag, NULL, NULL); + dst->seqhi, 0, TH_SYN, 0, + src->mss, 0, 0, (*state)->tag, NULL, NULL); REASON_SET(reason, PFRES_SYNPROXY); - return (PF_SYNPROXY_DROP); - } else if (((th->th_flags & (TH_SYN|TH_ACK)) != - (TH_SYN|TH_ACK)) || - (ntohl(th->th_ack) != (*state)->dst.seqhi + 1)) { + return PF_SYNPROXY_DROP; + } else if (((th->th_flags & (TH_SYN | TH_ACK)) != + (TH_SYN | TH_ACK)) || + (ntohl(th->th_ack) != dst->seqhi + 1)) { REASON_SET(reason, PFRES_SYNPROXY); - return (PF_DROP); + return PF_DROP; } else { - (*state)->dst.max_win = MAX(ntohs(th->th_win), 1); - (*state)->dst.seqlo = ntohl(th->th_seq); + dst->max_win = MAX(ntohs(th->th_win), 1); + dst->seqlo = ntohl(th->th_seq); pf_send_tcp((*state)->rule.ptr, pd->af, pd->dst, pd->src, th->th_dport, th->th_sport, ntohl(th->th_ack), ntohl(th->th_seq) + 1, - TH_ACK, (*state)->src.max_win, 0, 0, 0, + TH_ACK, src->max_win, 0, 0, 0, (*state)->tag, NULL, NULL); pf_send_tcp((*state)->rule.ptr, pd->af, &psrc->addr, &pdst->addr, psrc->xport.port, pdst->xport.port, - (*state)->src.seqhi + 1, (*state)->src.seqlo + 1, - TH_ACK, (*state)->dst.max_win, 0, 0, 1, + src->seqhi + 1, src->seqlo + 1, + TH_ACK, dst->max_win, 0, 0, 1, 0, NULL, NULL); - (*state)->src.seqdiff = (*state)->dst.seqhi - - (*state)->src.seqlo; - (*state)->dst.seqdiff = (*state)->src.seqhi - - (*state)->dst.seqlo; - (*state)->src.seqhi = (*state)->src.seqlo + - (*state)->dst.max_win; - (*state)->dst.seqhi = (*state)->dst.seqlo + - (*state)->src.max_win; - (*state)->src.wscale = (*state)->dst.wscale = 0; - (*state)->src.state = (*state)->dst.state = + src->seqdiff = dst->seqhi - + src->seqlo; + dst->seqdiff = src->seqhi - + dst->seqlo; + src->seqhi = src->seqlo + + dst->max_win; + dst->seqhi = dst->seqlo + + src->max_win; + src->wscale = dst->wscale = 0; + src->state = dst->state = TCPS_ESTABLISHED; REASON_SET(reason, PFRES_SYNPROXY); - return (PF_SYNPROXY_DROP); + return PF_SYNPROXY_DROP; } } - if (((th->th_flags & (TH_SYN|TH_ACK)) == TH_SYN) && + if (((th->th_flags & (TH_SYN | TH_ACK)) == TH_SYN) && dst->state >= TCPS_FIN_WAIT_2 && src->state >= TCPS_FIN_WAIT_2) { if (pf_status.debug >= PF_DEBUG_MISC) { @@ -5979,10 +7045,10 @@ pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, printf("\n"); } /* XXX make sure it's the same direction ?? */ - (*state)->src.state = (*state)->dst.state = TCPS_CLOSED; + src->state = dst->state = TCPS_CLOSED; pf_unlink_state(*state); *state = NULL; - return (PF_DROP); + return PF_DROP; } if ((th->th_flags & TH_SYN) == 0) { @@ -5990,9 +7056,9 @@ pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, (src->wscale & PF_WSCALE_MASK) : TCP_MAX_WINSHIFT; dws = (dst->wscale & PF_WSCALE_FLAG) ? (dst->wscale & PF_WSCALE_MASK) : TCP_MAX_WINSHIFT; - } - else + } else { sws = dws = 0; + } /* * Sequence tracking algorithm from Guido van Rooij's paper: @@ -6006,22 +7072,23 @@ pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, if ((pd->flags & PFDESC_TCP_NORM || dst->scrub) && src->scrub == NULL) { - if (pf_normalize_tcp_init(m, off, pd, th, src, dst)) { + if (pf_normalize_tcp_init(pbuf, off, pd, th, src, dst)) { REASON_SET(reason, PFRES_MEMORY); - return (PF_DROP); + return PF_DROP; } } /* Deferred generation of sequence number modulator */ if (dst->seqdiff && !src->seqdiff) { /* use random iss for the TCP server */ - while ((src->seqdiff = random() - seq) == 0) + while ((src->seqdiff = random() - seq) == 0) { ; + } ack = ntohl(th->th_ack) - dst->seqdiff; pf_change_a(&th->th_seq, &th->th_sum, htonl(seq + src->seqdiff), 0); pf_change_a(&th->th_ack, &th->th_sum, htonl(ack), 0); - copyback = off + sizeof (*th); + copyback = off + sizeof(*th); } else { ack = ntohl(th->th_ack); } @@ -6030,8 +7097,8 @@ pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, if (th->th_flags & TH_SYN) { end++; if (dst->wscale & PF_WSCALE_FLAG) { - src->wscale = pf_get_wscale(m, off, th->th_off, - pd->af); + src->wscale = pf_get_wscale(pbuf, off, + th->th_off, pd->af); if (src->wscale & PF_WSCALE_FLAG) { /* * Remove scale factor from initial @@ -6061,12 +7128,14 @@ pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, } } } - if (th->th_flags & TH_FIN) + if (th->th_flags & TH_FIN) { end++; + } src->seqlo = seq; - if (src->state < TCPS_SYN_SENT) + if (src->state < TCPS_SYN_SENT) { src->state = TCPS_SYN_SENT; + } /* * May need to slide the window (seqhi may have been set by @@ -6075,11 +7144,12 @@ pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, */ if (src->seqhi == 1 || SEQ_GEQ(end + MAX(1, (u_int32_t)dst->max_win << dws), - src->seqhi)) + src->seqhi)) { src->seqhi = end + MAX(1, (u_int32_t)dst->max_win << dws); - if (win > src->max_win) + } + if (win > src->max_win) { src->max_win = win; - + } } else { ack = ntohl(th->th_ack) - dst->seqdiff; if (src->seqdiff) { @@ -6087,20 +7157,22 @@ pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, pf_change_a(&th->th_seq, &th->th_sum, htonl(seq + src->seqdiff), 0); pf_change_a(&th->th_ack, &th->th_sum, htonl(ack), 0); - copyback = off+ sizeof (*th); + copyback = off + sizeof(*th); } end = seq + pd->p_len; - if (th->th_flags & TH_SYN) + if (th->th_flags & TH_SYN) { end++; - if (th->th_flags & TH_FIN) + } + if (th->th_flags & TH_FIN) { end++; + } } if ((th->th_flags & TH_ACK) == 0) { /* Let it pass through the ack skew check */ ack = dst->seqlo; } else if ((ack == 0 && - (th->th_flags & (TH_ACK|TH_RST)) == (TH_ACK|TH_RST)) || + (th->th_flags & (TH_ACK | TH_RST)) == (TH_ACK | TH_RST)) || /* broken tcp stacks do not set ack */ (dst->state < TCPS_SYN_SENT)) { /* @@ -6131,18 +7203,18 @@ pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, * spoof into a TCP connection won't bother blindly sending SACK * options anyway. */ - if (dst->seqdiff && (th->th_off << 2) > (int)sizeof (struct tcphdr)) { - copyback = pf_modulate_sack(m, off, pd, th, dst); + if (dst->seqdiff && (th->th_off << 2) > (int)sizeof(struct tcphdr)) { + copyback = pf_modulate_sack(pbuf, off, pd, th, dst); if (copyback == -1) { REASON_SET(reason, PFRES_MEMORY); - return (PF_DROP); + return PF_DROP; } - m = pd->mp; + pbuf = pd->mp; // XXXSCW: Why? } -#define MAXACKWINDOW (0xffff + 1500) /* 1500 is an arbitrary fudge factor */ +#define MAXACKWINDOW (0xffff + 1500) /* 1500 is an arbitrary fudge factor */ if (SEQ_GEQ(src->seqhi, end) && /* Last octet inside other's window space */ SEQ_GEQ(seq, src->seqlo - ((u_int32_t)dst->max_win << dws)) && @@ -6154,33 +7226,41 @@ pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, ((th->th_flags & TH_RST) == 0 || orig_seq == src->seqlo || (orig_seq == src->seqlo + 1) || (orig_seq + 1 == src->seqlo) || (pd->flags & PFDESC_IP_REAS) == 0)) { - /* Require an exact/+1 sequence match on resets when possible */ + /* Require an exact/+1 sequence match on resets when possible */ if (dst->scrub || src->scrub) { - if (pf_normalize_tcp_stateful(m, off, pd, reason, th, - *state, src, dst, ©back)) - return (PF_DROP); + if (pf_normalize_tcp_stateful(pbuf, off, pd, reason, th, + *state, src, dst, ©back)) { + return PF_DROP; + } - m = pd->mp; + pbuf = pd->mp; // XXXSCW: Why? } /* update max window */ - if (src->max_win < win) + if (src->max_win < win) { src->max_win = win; + } /* synchronize sequencing */ - if (SEQ_GT(end, src->seqlo)) + if (SEQ_GT(end, src->seqlo)) { src->seqlo = end; + } /* slide the window of what the other end can send */ - if (SEQ_GEQ(ack + ((u_int32_t)win << sws), dst->seqhi)) + if (SEQ_GEQ(ack + ((u_int32_t)win << sws), dst->seqhi)) { dst->seqhi = ack + MAX(((u_int32_t)win << sws), 1); + } /* update states */ - if (th->th_flags & TH_SYN) - if (src->state < TCPS_SYN_SENT) + if (th->th_flags & TH_SYN) { + if (src->state < TCPS_SYN_SENT) { src->state = TCPS_SYN_SENT; - if (th->th_flags & TH_FIN) - if (src->state < TCPS_CLOSING) + } + } + if (th->th_flags & TH_FIN) { + if (src->state < TCPS_CLOSING) { src->state = TCPS_CLOSING; + } + } if (th->th_flags & TH_ACK) { if (dst->state == TCPS_SYN_SENT) { dst->state = TCPS_ESTABLISHED; @@ -6188,39 +7268,41 @@ pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, (*state)->src_node != NULL && pf_src_connlimit(state)) { REASON_SET(reason, PFRES_SRCLIMIT); - return (PF_DROP); + return PF_DROP; } - } else if (dst->state == TCPS_CLOSING) + } else if (dst->state == TCPS_CLOSING) { dst->state = TCPS_FIN_WAIT_2; + } } - if (th->th_flags & TH_RST) + if (th->th_flags & TH_RST) { src->state = dst->state = TCPS_TIME_WAIT; + } /* update expire time */ (*state)->expire = pf_time_second(); if (src->state >= TCPS_FIN_WAIT_2 && - dst->state >= TCPS_FIN_WAIT_2) + dst->state >= TCPS_FIN_WAIT_2) { (*state)->timeout = PFTM_TCP_CLOSED; - else if (src->state >= TCPS_CLOSING && - dst->state >= TCPS_CLOSING) + } else if (src->state >= TCPS_CLOSING && + dst->state >= TCPS_CLOSING) { (*state)->timeout = PFTM_TCP_FIN_WAIT; - else if (src->state < TCPS_ESTABLISHED || - dst->state < TCPS_ESTABLISHED) + } else if (src->state < TCPS_ESTABLISHED || + dst->state < TCPS_ESTABLISHED) { (*state)->timeout = PFTM_TCP_OPENING; - else if (src->state >= TCPS_CLOSING || - dst->state >= TCPS_CLOSING) + } else if (src->state >= TCPS_CLOSING || + dst->state >= TCPS_CLOSING) { (*state)->timeout = PFTM_TCP_CLOSING; - else + } else { (*state)->timeout = PFTM_TCP_ESTABLISHED; + } /* Fall through to PASS packet */ - } else if ((dst->state < TCPS_SYN_SENT || dst->state >= TCPS_FIN_WAIT_2 || src->state >= TCPS_FIN_WAIT_2) && SEQ_GEQ(src->seqhi + MAXACKWINDOW, end) && /* Within a window forward of the originating packet */ SEQ_GEQ(seq, src->seqlo - MAXACKWINDOW)) { - /* Within a window backward of the originating packet */ + /* Within a window backward of the originating packet */ /* * This currently handles three situations: @@ -6252,51 +7334,58 @@ pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, pd->p_len, ackskew, (*state)->packets[0], (*state)->packets[1], direction == PF_IN ? "in" : "out", - direction == (*state)->state_key->direction ? + direction == sk->direction ? "fwd" : "rev"); } if (dst->scrub || src->scrub) { - if (pf_normalize_tcp_stateful(m, off, pd, reason, th, - *state, src, dst, ©back)) - return (PF_DROP); - m = pd->mp; + if (pf_normalize_tcp_stateful(pbuf, off, pd, reason, th, + *state, src, dst, ©back)) { + return PF_DROP; + } + pbuf = pd->mp; // XXXSCW: Why? } /* update max window */ - if (src->max_win < win) + if (src->max_win < win) { src->max_win = win; + } /* synchronize sequencing */ - if (SEQ_GT(end, src->seqlo)) + if (SEQ_GT(end, src->seqlo)) { src->seqlo = end; + } /* slide the window of what the other end can send */ - if (SEQ_GEQ(ack + ((u_int32_t)win << sws), dst->seqhi)) + if (SEQ_GEQ(ack + ((u_int32_t)win << sws), dst->seqhi)) { dst->seqhi = ack + MAX(((u_int32_t)win << sws), 1); + } /* * Cannot set dst->seqhi here since this could be a shotgunned * SYN and not an already established connection. */ - if (th->th_flags & TH_FIN) - if (src->state < TCPS_CLOSING) + if (th->th_flags & TH_FIN) { + if (src->state < TCPS_CLOSING) { src->state = TCPS_CLOSING; - if (th->th_flags & TH_RST) + } + } + if (th->th_flags & TH_RST) { src->state = dst->state = TCPS_TIME_WAIT; + } /* Fall through to PASS packet */ - } else { - if ((*state)->dst.state == TCPS_SYN_SENT && - (*state)->src.state == TCPS_SYN_SENT) { + if (dst->state == TCPS_SYN_SENT && + src->state == TCPS_SYN_SENT) { /* Send RST for state mismatches during handshake */ - if (!(th->th_flags & TH_RST)) + if (!(th->th_flags & TH_RST)) { pf_send_tcp((*state)->rule.ptr, pd->af, pd->dst, pd->src, th->th_dport, th->th_sport, ntohl(th->th_ack), 0, TH_RST, 0, 0, (*state)->rule.ptr->return_ttl, 1, 0, pd->eh, kif->pfik_ifp); + } src->seqlo = 0; src->seqhi = 1; src->max_win = 1; @@ -6310,7 +7399,7 @@ pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, (unsigned int)sws, (unsigned int)dws, (*state)->packets[0], (*state)->packets[1], direction == PF_IN ? "in" : "out", - direction == (*state)->state_key->direction ? + direction == sk->direction ? "fwd" : "rev"); printf("pf: State failure on: %c %c %c %c | %c %c\n", SEQ_GEQ(src->seqhi, end) ? ' ' : '1', @@ -6323,93 +7412,139 @@ pf_test_state_tcp(struct pf_state **state, int direction, struct pfi_kif *kif, SEQ_GEQ(seq, src->seqlo - MAXACKWINDOW) ?' ' :'6'); } REASON_SET(reason, PFRES_BADSTATE); - return (PF_DROP); + return PF_DROP; } /* Any packets which have gotten here are to be passed */ - if ((*state)->state_key->app_state && - (*state)->state_key->app_state->handler) { - (*state)->state_key->app_state->handler(*state, direction, + if (sk->app_state && + sk->app_state->handler) { + sk->app_state->handler(*state, direction, off + (th->th_off << 2), pd, kif); if (pd->lmw < 0) { REASON_SET(reason, PFRES_MEMORY); - return (PF_DROP); + return PF_DROP; } - m = pd->mp; + pbuf = pd->mp; // XXXSCW: Why? } /* translate source/destination address, if necessary */ - if (STATE_TRANSLATE((*state)->state_key)) { - if (direction == PF_OUT) + if (STATE_TRANSLATE(sk)) { + pd->naf = (pd->af == sk->af_lan) ? sk->af_gwy : sk->af_lan; + + if (direction == PF_OUT) { pf_change_ap(direction, pd->mp, pd->src, &th->th_sport, - pd->ip_sum, &th->th_sum, - &(*state)->state_key->gwy.addr, - (*state)->state_key->gwy.xport.port, 0, pd->af); - else - pf_change_ap(direction, pd->mp, pd->dst, &th->th_dport, - pd->ip_sum, &th->th_sum, - &(*state)->state_key->lan.addr, - (*state)->state_key->lan.xport.port, 0, pd->af); - copyback = off + sizeof (*th); + pd->ip_sum, &th->th_sum, &sk->gwy.addr, + sk->gwy.xport.port, 0, pd->af, pd->naf, 1); + } else { + if (pd->af != pd->naf) { + if (pd->af == sk->af_gwy) { + pf_change_ap(direction, pd->mp, pd->dst, + &th->th_dport, pd->ip_sum, + &th->th_sum, &sk->lan.addr, + sk->lan.xport.port, 0, + pd->af, pd->naf, 0); + + pf_change_ap(direction, pd->mp, pd->src, + &th->th_sport, pd->ip_sum, + &th->th_sum, &sk->ext_lan.addr, + th->th_sport, 0, pd->af, + pd->naf, 0); + } else { + pf_change_ap(direction, pd->mp, pd->dst, + &th->th_dport, pd->ip_sum, + &th->th_sum, &sk->ext_gwy.addr, + th->th_dport, 0, pd->af, + pd->naf, 0); + + pf_change_ap(direction, pd->mp, pd->src, + &th->th_sport, pd->ip_sum, + &th->th_sum, &sk->gwy.addr, + sk->gwy.xport.port, 0, pd->af, + pd->naf, 0); + } + } else { + pf_change_ap(direction, pd->mp, pd->dst, + &th->th_dport, pd->ip_sum, + &th->th_sum, &sk->lan.addr, + sk->lan.xport.port, 0, pd->af, + pd->naf, 1); + } + } + + copyback = off + sizeof(*th); } if (copyback) { - m = pf_lazy_makewritable(pd, m, copyback); - if (!m) { + if (pf_lazy_makewritable(pd, pbuf, copyback) == NULL) { REASON_SET(reason, PFRES_MEMORY); - return (PF_DROP); + return PF_DROP; } /* Copyback sequence modulation or stateful scrub changes */ - m_copyback(m, off, sizeof (*th), th); - } + pbuf_copy_back(pbuf, off, sizeof(*th), th); - return (PF_PASS); + if (sk->af_lan != sk->af_gwy) { + return pf_do_nat64(sk, pd, pbuf, off); + } + } + return PF_PASS; } static int pf_test_state_udp(struct pf_state **state, int direction, struct pfi_kif *kif, - struct mbuf *m, int off, void *h, struct pf_pdesc *pd, u_short *reason) + pbuf_t *pbuf, int off, void *h, struct pf_pdesc *pd, u_short *reason) { #pragma unused(h) - struct pf_state_peer *src, *dst; - struct pf_state_key_cmp key; - struct udphdr *uh = pd->hdr.udp; + struct pf_state_peer *src, *dst; + struct pf_state_key_cmp key; + struct pf_state_key *sk; + struct udphdr *uh = pd->hdr.udp; struct pf_app_state as; - int dx, action, extfilter; + int action, extfilter; key.app_state = 0; key.proto_variant = PF_EXTFILTER_APD; - key.af = pd->af; key.proto = IPPROTO_UDP; - if (direction == PF_IN) { - PF_ACPY(&key.ext.addr, pd->src, key.af); - PF_ACPY(&key.gwy.addr, pd->dst, key.af); - key.ext.xport.port = uh->uh_sport; - key.gwy.xport.port = uh->uh_dport; - dx = PF_IN; - } else { - PF_ACPY(&key.lan.addr, pd->src, key.af); - PF_ACPY(&key.ext.addr, pd->dst, key.af); - key.lan.xport.port = uh->uh_sport; - key.ext.xport.port = uh->uh_dport; - dx = PF_OUT; - } + key.af_lan = key.af_gwy = pd->af; + + /* + * For NAT64 the first time rule search and state creation + * is done on the incoming side only. + * Once the state gets created, NAT64's LAN side (ipv6) will + * not be able to find the state in ext-gwy tree as that normally + * is intended to be looked up for incoming traffic from the + * WAN side. + * Therefore to handle NAT64 case we init keys here for both + * lan-ext as well as ext-gwy trees. + * In the state lookup we attempt a lookup on both trees if + * first one does not return any result and return a match if + * the match state's was created by NAT64 rule. + */ + PF_ACPY(&key.ext_gwy.addr, pd->src, key.af_gwy); + PF_ACPY(&key.gwy.addr, pd->dst, key.af_gwy); + key.ext_gwy.xport.port = uh->uh_sport; + key.gwy.xport.port = uh->uh_dport; + + PF_ACPY(&key.lan.addr, pd->src, key.af_lan); + PF_ACPY(&key.ext_lan.addr, pd->dst, key.af_lan); + key.lan.xport.port = uh->uh_sport; + key.ext_lan.xport.port = uh->uh_dport; if (ntohs(uh->uh_sport) == PF_IKE_PORT && ntohs(uh->uh_dport) == PF_IKE_PORT) { struct pf_ike_hdr ike; - size_t plen = m->m_pkthdr.len - off - sizeof (*uh); + size_t plen = pbuf->pb_packet_len - off - sizeof(*uh); if (plen < PF_IKE_PACKET_MINSIZE) { DPFPRINTF(PF_DEBUG_MISC, ("pf: IKE message too small.\n")); - return (PF_DROP); + return PF_DROP; } - if (plen > sizeof (ike)) - plen = sizeof (ike); - m_copydata(m, off + sizeof (*uh), plen, &ike); + if (plen > sizeof(ike)) { + plen = sizeof(ike); + } + pbuf_copy_data(pbuf, off + sizeof(*uh), plen, &ike); if (ike.initiator_cookie) { key.app_state = &as; @@ -6429,16 +7564,16 @@ pf_test_state_udp(struct pf_state **state, int direction, struct pfi_kif *kif, } } - *state = pf_find_state(kif, &key, dx); + *state = pf_find_state(kif, &key, direction); if (!key.app_state && *state == 0) { key.proto_variant = PF_EXTFILTER_AD; - *state = pf_find_state(kif, &key, dx); + *state = pf_find_state(kif, &key, direction); } if (!key.app_state && *state == 0) { key.proto_variant = PF_EXTFILTER_EI; - *state = pf_find_state(kif, &key, dx); + *state = pf_find_state(kif, &key, direction); } /* similar to STATE_LOOKUP() */ @@ -6451,10 +7586,18 @@ pf_test_state_udp(struct pf_state **state, int direction, struct pfi_kif *kif, } } - if (pf_state_lookup_aux(state, kif, direction, &action)) - return (action); + if (pf_state_lookup_aux(state, kif, direction, &action)) { + return action; + } - if (direction == (*state)->state_key->direction) { + sk = (*state)->state_key; + + /* + * In case of NAT64 the translation is first applied on the LAN + * side. Therefore for stack's address family comparison + * we use sk->af_lan. + */ + if ((direction == sk->direction) && (pd->af == sk->af_lan)) { src = &(*state)->src; dst = &(*state)->dst; } else { @@ -6463,75 +7606,123 @@ pf_test_state_udp(struct pf_state **state, int direction, struct pfi_kif *kif, } /* update states */ - if (src->state < PFUDPS_SINGLE) + if (src->state < PFUDPS_SINGLE) { src->state = PFUDPS_SINGLE; - if (dst->state == PFUDPS_SINGLE) + } + if (dst->state == PFUDPS_SINGLE) { dst->state = PFUDPS_MULTIPLE; + } /* update expire time */ (*state)->expire = pf_time_second(); - if (src->state == PFUDPS_MULTIPLE && dst->state == PFUDPS_MULTIPLE) + if (src->state == PFUDPS_MULTIPLE && dst->state == PFUDPS_MULTIPLE) { (*state)->timeout = PFTM_UDP_MULTIPLE; - else + } else { (*state)->timeout = PFTM_UDP_SINGLE; + } - extfilter = (*state)->state_key->proto_variant; + extfilter = sk->proto_variant; if (extfilter > PF_EXTFILTER_APD) { - (*state)->state_key->ext.xport.port = key.ext.xport.port; - if (extfilter > PF_EXTFILTER_AD) - PF_ACPY(&(*state)->state_key->ext.addr, - &key.ext.addr, key.af); + if (direction == PF_OUT) { + sk->ext_lan.xport.port = key.ext_lan.xport.port; + if (extfilter > PF_EXTFILTER_AD) { + PF_ACPY(&sk->ext_lan.addr, &key.ext_lan.addr, + key.af_lan); + } + } else { + sk->ext_gwy.xport.port = key.ext_gwy.xport.port; + if (extfilter > PF_EXTFILTER_AD) { + PF_ACPY(&sk->ext_gwy.addr, &key.ext_gwy.addr, + key.af_gwy); + } + } } - if ((*state)->state_key->app_state && - (*state)->state_key->app_state->handler) { - (*state)->state_key->app_state->handler(*state, direction, - off + uh->uh_ulen, pd, kif); + if (sk->app_state && sk->app_state->handler) { + sk->app_state->handler(*state, direction, off + uh->uh_ulen, + pd, kif); if (pd->lmw < 0) { REASON_SET(reason, PFRES_MEMORY); - return (PF_DROP); + return PF_DROP; } - m = pd->mp; + pbuf = pd->mp; // XXXSCW: Why? } /* translate source/destination address, if necessary */ - if (STATE_TRANSLATE((*state)->state_key)) { - m = pf_lazy_makewritable(pd, m, off + sizeof (*uh)); - if (!m) { + if (STATE_TRANSLATE(sk)) { + if (pf_lazy_makewritable(pd, pbuf, off + sizeof(*uh)) == NULL) { REASON_SET(reason, PFRES_MEMORY); - return (PF_DROP); + return PF_DROP; } - if (direction == PF_OUT) + pd->naf = (pd->af == sk->af_lan) ? sk->af_gwy : sk->af_lan; + + if (direction == PF_OUT) { pf_change_ap(direction, pd->mp, pd->src, &uh->uh_sport, - pd->ip_sum, &uh->uh_sum, - &(*state)->state_key->gwy.addr, - (*state)->state_key->gwy.xport.port, 1, pd->af); - else - pf_change_ap(direction, pd->mp, pd->dst, &uh->uh_dport, - pd->ip_sum, &uh->uh_sum, - &(*state)->state_key->lan.addr, - (*state)->state_key->lan.xport.port, 1, pd->af); - m_copyback(m, off, sizeof (*uh), uh); - } - - return (PF_PASS); + pd->ip_sum, &uh->uh_sum, &sk->gwy.addr, + sk->gwy.xport.port, 1, pd->af, pd->naf, 1); + } else { + if (pd->af != pd->naf) { + if (pd->af == sk->af_gwy) { + pf_change_ap(direction, pd->mp, pd->dst, + &uh->uh_dport, pd->ip_sum, + &uh->uh_sum, &sk->lan.addr, + sk->lan.xport.port, 1, + pd->af, pd->naf, 0); + + pf_change_ap(direction, pd->mp, pd->src, + &uh->uh_sport, pd->ip_sum, + &uh->uh_sum, &sk->ext_lan.addr, + uh->uh_sport, 1, pd->af, + pd->naf, 0); + } else { + pf_change_ap(direction, pd->mp, pd->dst, + &uh->uh_dport, pd->ip_sum, + &uh->uh_sum, &sk->ext_gwy.addr, + uh->uh_dport, 1, pd->af, + pd->naf, 0); + + pf_change_ap(direction, pd->mp, pd->src, + &uh->uh_sport, pd->ip_sum, + &uh->uh_sum, &sk->gwy.addr, + sk->gwy.xport.port, 1, pd->af, + pd->naf, 0); + } + } else { + pf_change_ap(direction, pd->mp, pd->dst, + &uh->uh_dport, pd->ip_sum, + &uh->uh_sum, &sk->lan.addr, + sk->lan.xport.port, 1, + pd->af, pd->naf, 1); + } + } + + pbuf_copy_back(pbuf, off, sizeof(*uh), uh); + if (sk->af_lan != sk->af_gwy) { + return pf_do_nat64(sk, pd, pbuf, off); + } + } + return PF_PASS; } static int pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, - struct mbuf *m, int off, void *h, struct pf_pdesc *pd, u_short *reason) + pbuf_t *pbuf, int off, void *h, struct pf_pdesc *pd, u_short *reason) { #pragma unused(h) - struct pf_addr *saddr = pd->src, *daddr = pd->dst; - u_int16_t icmpid = 0, *icmpsum; - u_int8_t icmptype; - int state_icmp = 0; + struct pf_addr *saddr = pd->src, *daddr = pd->dst; + struct in_addr srcv4_inaddr = saddr->v4addr; + u_int16_t icmpid = 0, *icmpsum = NULL; + u_int8_t icmptype = 0; + int state_icmp = 0; struct pf_state_key_cmp key; + struct pf_state_key *sk; struct pf_app_state as; key.app_state = 0; + pd->off = off; + switch (pd->proto) { #if INET case IPPROTO_ICMP: @@ -6539,12 +7730,9 @@ pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, icmpid = pd->hdr.icmp->icmp_id; icmpsum = &pd->hdr.icmp->icmp_cksum; - if (icmptype == ICMP_UNREACH || - icmptype == ICMP_SOURCEQUENCH || - icmptype == ICMP_REDIRECT || - icmptype == ICMP_TIMXCEED || - icmptype == ICMP_PARAMPROB) + if (ICMP_ERRORTYPE(icmptype)) { state_icmp++; + } break; #endif /* INET */ #if INET6 @@ -6553,60 +7741,66 @@ pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, icmpid = pd->hdr.icmp6->icmp6_id; icmpsum = &pd->hdr.icmp6->icmp6_cksum; - if (icmptype == ICMP6_DST_UNREACH || - icmptype == ICMP6_PACKET_TOO_BIG || - icmptype == ICMP6_TIME_EXCEEDED || - icmptype == ICMP6_PARAM_PROB) + if (ICMP6_ERRORTYPE(icmptype)) { state_icmp++; + } break; #endif /* INET6 */ } if (!state_icmp) { - /* * ICMP query/reply message not related to a TCP/UDP packet. * Search for an ICMP state. */ - key.af = pd->af; - key.proto = pd->proto; - if (direction == PF_IN) { - PF_ACPY(&key.ext.addr, pd->src, key.af); - PF_ACPY(&key.gwy.addr, pd->dst, key.af); - key.ext.xport.port = 0; - key.gwy.xport.port = icmpid; - } else { - PF_ACPY(&key.lan.addr, pd->src, key.af); - PF_ACPY(&key.ext.addr, pd->dst, key.af); - key.lan.xport.port = icmpid; - key.ext.xport.port = 0; - } + /* + * NAT64 requires protocol translation between ICMPv4 + * and ICMPv6. TCP and UDP do not require protocol + * translation. To avoid adding complexity just to + * handle ICMP(v4addr/v6addr), we always lookup for + * proto = IPPROTO_ICMP on both LAN and WAN side + */ + key.proto = IPPROTO_ICMP; + key.af_lan = key.af_gwy = pd->af; + + PF_ACPY(&key.ext_gwy.addr, pd->src, key.af_gwy); + PF_ACPY(&key.gwy.addr, pd->dst, key.af_gwy); + key.ext_gwy.xport.port = 0; + key.gwy.xport.port = icmpid; + + PF_ACPY(&key.lan.addr, pd->src, key.af_lan); + PF_ACPY(&key.ext_lan.addr, pd->dst, key.af_lan); + key.lan.xport.port = icmpid; + key.ext_lan.xport.port = 0; STATE_LOOKUP(); + sk = (*state)->state_key; (*state)->expire = pf_time_second(); (*state)->timeout = PFTM_ICMP_ERROR_REPLY; /* translate source/destination address, if necessary */ - if (STATE_TRANSLATE((*state)->state_key)) { + if (STATE_TRANSLATE(sk)) { + pd->naf = (pd->af == sk->af_lan) ? + sk->af_gwy : sk->af_lan; if (direction == PF_OUT) { switch (pd->af) { #if INET case AF_INET: - pf_change_a(&saddr->v4.s_addr, + pf_change_a(&saddr->v4addr.s_addr, pd->ip_sum, - (*state)->state_key->gwy.addr.v4.s_addr, 0); + sk->gwy.addr.v4addr.s_addr, 0); pd->hdr.icmp->icmp_cksum = pf_cksum_fixup( - pd->hdr.icmp->icmp_cksum, icmpid, - (*state)->state_key->gwy.xport.port, 0); + pd->hdr.icmp->icmp_cksum, icmpid, + sk->gwy.xport.port, 0); pd->hdr.icmp->icmp_id = - (*state)->state_key->gwy.xport.port; - m = pf_lazy_makewritable(pd, m, - off + ICMP_MINLEN); - if (!m) - return (PF_DROP); - m_copyback(m, off, ICMP_MINLEN, + sk->gwy.xport.port; + if (pf_lazy_makewritable(pd, pbuf, + off + ICMP_MINLEN) == NULL) { + return PF_DROP; + } + pbuf_copy_back(pbuf, off, ICMP_MINLEN, pd->hdr.icmp); break; #endif /* INET */ @@ -6614,13 +7808,14 @@ pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, case AF_INET6: pf_change_a6(saddr, &pd->hdr.icmp6->icmp6_cksum, - &(*state)->state_key->gwy.addr, 0); - m = pf_lazy_makewritable(pd, m, - off + sizeof (struct icmp6_hdr)); - if (!m) - return (PF_DROP); - m_copyback(m, off, - sizeof (struct icmp6_hdr), + &sk->gwy.addr, 0); + if (pf_lazy_makewritable(pd, pbuf, + off + sizeof(struct icmp6_hdr)) == + NULL) { + return PF_DROP; + } + pbuf_copy_back(pbuf, off, + sizeof(struct icmp6_hdr), pd->hdr.icmp6); break; #endif /* INET6 */ @@ -6629,61 +7824,89 @@ pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, switch (pd->af) { #if INET case AF_INET: - pf_change_a(&daddr->v4.s_addr, - pd->ip_sum, - (*state)->state_key->lan.addr.v4.s_addr, 0); - pd->hdr.icmp->icmp_cksum = - pf_cksum_fixup( - pd->hdr.icmp->icmp_cksum, icmpid, - (*state)->state_key->lan.xport.port, 0); - pd->hdr.icmp->icmp_id = - (*state)->state_key->lan.xport.port; - m = pf_lazy_makewritable(pd, m, - off + ICMP_MINLEN); - if (!m) - return (PF_DROP); - m_copyback(m, off, ICMP_MINLEN, + if (pd->naf != AF_INET) { + if (pf_translate_icmp_af( + AF_INET6, pd->hdr.icmp)) { + return PF_DROP; + } + + pd->proto = IPPROTO_ICMPV6; + } else { + pf_change_a(&daddr->v4addr.s_addr, + pd->ip_sum, + sk->lan.addr.v4addr.s_addr, 0); + + pd->hdr.icmp->icmp_cksum = + pf_cksum_fixup( + pd->hdr.icmp->icmp_cksum, + icmpid, sk->lan.xport.port, 0); + + pd->hdr.icmp->icmp_id = + sk->lan.xport.port; + } + + if (pf_lazy_makewritable(pd, pbuf, + off + ICMP_MINLEN) == NULL) { + return PF_DROP; + } + pbuf_copy_back(pbuf, off, ICMP_MINLEN, pd->hdr.icmp); + if (sk->af_lan != sk->af_gwy) { + return pf_do_nat64(sk, pd, + pbuf, off); + } break; #endif /* INET */ #if INET6 case AF_INET6: - pf_change_a6(daddr, - &pd->hdr.icmp6->icmp6_cksum, - &(*state)->state_key->lan.addr, 0); - m = pf_lazy_makewritable(pd, m, - off + sizeof (struct icmp6_hdr)); - if (!m) - return (PF_DROP); - m_copyback(m, off, - sizeof (struct icmp6_hdr), + if (pd->naf != AF_INET6) { + if (pf_translate_icmp_af( + AF_INET, pd->hdr.icmp6)) { + return PF_DROP; + } + + pd->proto = IPPROTO_ICMP; + } else { + pf_change_a6(daddr, + &pd->hdr.icmp6->icmp6_cksum, + &sk->lan.addr, 0); + } + if (pf_lazy_makewritable(pd, pbuf, + off + sizeof(struct icmp6_hdr)) == + NULL) { + return PF_DROP; + } + pbuf_copy_back(pbuf, off, + sizeof(struct icmp6_hdr), pd->hdr.icmp6); + if (sk->af_lan != sk->af_gwy) { + return pf_do_nat64(sk, pd, + pbuf, off); + } break; #endif /* INET6 */ } } } - return (PF_PASS); - + return PF_PASS; } else { /* * ICMP error message in response to a TCP/UDP packet. * Extract the inner TCP/UDP header and search for that state. */ - - struct pf_pdesc pd2; + struct pf_pdesc pd2; /* For inner (original) header */ #if INET - struct ip h2; + struct ip h2; #endif /* INET */ #if INET6 - struct ip6_hdr h2_6; - int terminal = 0; + struct ip6_hdr h2_6; + int terminal = 0; #endif /* INET6 */ - int ipoff2 = 0; - int off2 = 0; + int ipoff2 = 0; + int off2 = 0; - memset(&pd2, 0, sizeof (pd2)); + memset(&pd2, 0, sizeof(pd2)); pd2.af = pd->af; switch (pd->af) { @@ -6692,12 +7915,12 @@ pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, /* offset of h2 in mbuf chain */ ipoff2 = off + ICMP_MINLEN; - if (!pf_pull_hdr(m, ipoff2, &h2, sizeof (h2), + if (!pf_pull_hdr(pbuf, ipoff2, &h2, sizeof(h2), NULL, reason, pd2.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: ICMP error message too short " "(ip)\n")); - return (PF_DROP); + return PF_DROP; } /* * ICMP error messages don't refer to non-first @@ -6705,11 +7928,13 @@ pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, */ if (h2.ip_off & htons(IP_OFFMASK)) { REASON_SET(reason, PFRES_FRAG); - return (PF_DROP); + return PF_DROP; } /* offset of protocol header that follows h2 */ off2 = ipoff2 + (h2.ip_hl << 2); + /* TODO */ + pd2.off = ipoff2 + (h2.ip_hl << 2); pd2.proto = h2.ip_p; pd2.src = (struct pf_addr *)&h2.ip_src; @@ -6719,20 +7944,20 @@ pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, #endif /* INET */ #if INET6 case AF_INET6: - ipoff2 = off + sizeof (struct icmp6_hdr); + ipoff2 = off + sizeof(struct icmp6_hdr); - if (!pf_pull_hdr(m, ipoff2, &h2_6, sizeof (h2_6), + if (!pf_pull_hdr(pbuf, ipoff2, &h2_6, sizeof(h2_6), NULL, reason, pd2.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: ICMP error message too short " "(ip6)\n")); - return (PF_DROP); + return PF_DROP; } pd2.proto = h2_6.ip6_nxt; - pd2.src = (struct pf_addr *)&h2_6.ip6_src; - pd2.dst = (struct pf_addr *)&h2_6.ip6_dst; + pd2.src = (struct pf_addr *)(uintptr_t)&h2_6.ip6_src; + pd2.dst = (struct pf_addr *)(uintptr_t)&h2_6.ip6_dst; pd2.ip_sum = NULL; - off2 = ipoff2 + sizeof (h2_6); + off2 = ipoff2 + sizeof(h2_6); do { switch (pd2.proto) { case IPPROTO_FRAGMENT: @@ -6741,7 +7966,7 @@ pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, * non-first fragments */ REASON_SET(reason, PFRES_FRAG); - return (PF_DROP); + return PF_DROP; case IPPROTO_AH: case IPPROTO_HOPOPTS: case IPPROTO_ROUTING: @@ -6749,17 +7974,18 @@ pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, /* get next header and header length */ struct ip6_ext opt6; - if (!pf_pull_hdr(m, off2, &opt6, - sizeof (opt6), NULL, reason, + if (!pf_pull_hdr(pbuf, off2, &opt6, + sizeof(opt6), NULL, reason, pd2.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: ICMPv6 short opt\n")); - return (PF_DROP); + return PF_DROP; } - if (pd2.proto == IPPROTO_AH) + if (pd2.proto == IPPROTO_AH) { off2 += (opt6.ip6e_len + 2) * 4; - else + } else { off2 += (opt6.ip6e_len + 1) * 8; + } pd2.proto = opt6.ip6e_nxt; /* goto the next header */ break; @@ -6769,48 +7995,52 @@ pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, break; } } while (!terminal); + /* TODO */ + pd2.off = ipoff2; break; #endif /* INET6 */ } switch (pd2.proto) { case IPPROTO_TCP: { - struct tcphdr th; - u_int32_t seq; - struct pf_state_peer *src, *dst; - u_int8_t dws; - int copyback = 0; + struct tcphdr th; + u_int32_t seq; + struct pf_state_peer *src, *dst; + u_int8_t dws; + int copyback = 0; /* * Only the first 8 bytes of the TCP header can be * expected. Don't access any TCP header fields after * th_seq, an ackskew test is not possible. */ - if (!pf_pull_hdr(m, off2, &th, 8, NULL, reason, + if (!pf_pull_hdr(pbuf, off2, &th, 8, NULL, reason, pd2.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: ICMP error message too short " "(tcp)\n")); - return (PF_DROP); + return PF_DROP; } - key.af = pd2.af; key.proto = IPPROTO_TCP; - if (direction == PF_IN) { - PF_ACPY(&key.ext.addr, pd2.dst, key.af); - PF_ACPY(&key.gwy.addr, pd2.src, key.af); - key.ext.xport.port = th.th_dport; - key.gwy.xport.port = th.th_sport; - } else { - PF_ACPY(&key.lan.addr, pd2.dst, key.af); - PF_ACPY(&key.ext.addr, pd2.src, key.af); - key.lan.xport.port = th.th_dport; - key.ext.xport.port = th.th_sport; - } + key.af_gwy = pd2.af; + PF_ACPY(&key.ext_gwy.addr, pd2.dst, key.af_gwy); + PF_ACPY(&key.gwy.addr, pd2.src, key.af_gwy); + key.ext_gwy.xport.port = th.th_dport; + key.gwy.xport.port = th.th_sport; + + key.af_lan = pd2.af; + PF_ACPY(&key.lan.addr, pd2.dst, key.af_lan); + PF_ACPY(&key.ext_lan.addr, pd2.src, key.af_lan); + key.lan.xport.port = th.th_dport; + key.ext_lan.xport.port = th.th_sport; STATE_LOOKUP(); - if (direction == (*state)->state_key->direction) { + sk = (*state)->state_key; + if ((direction == sk->direction) && + ((sk->af_lan == sk->af_gwy) || + (pd2.af == sk->af_lan))) { src = &(*state)->dst; dst = &(*state)->src; } else { @@ -6818,10 +8048,11 @@ pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, dst = &(*state)->dst; } - if (src->wscale && (dst->wscale & PF_WSCALE_FLAG)) + if (src->wscale && (dst->wscale & PF_WSCALE_FLAG)) { dws = dst->wscale & PF_WSCALE_MASK; - else + } else { dws = TCP_MAX_WINSHIFT; + } /* Demodulate sequence number */ seq = ntohl(th.th_seq) - src->seqdiff; @@ -6845,20 +8076,100 @@ pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, printf(" seq=%u\n", seq); } REASON_SET(reason, PFRES_BADSTATE); - return (PF_DROP); + return PF_DROP; } - if (STATE_TRANSLATE((*state)->state_key)) { + pd->naf = pd2.naf = (pd2.af == sk->af_lan) ? + sk->af_gwy : sk->af_lan; + + if (STATE_TRANSLATE(sk)) { + /* NAT64 case */ + if (sk->af_lan != sk->af_gwy) { + struct pf_state_host *saddr2, *daddr2; + + if (pd2.naf == sk->af_lan) { + saddr2 = &sk->lan; + daddr2 = &sk->ext_lan; + } else { + saddr2 = &sk->ext_gwy; + daddr2 = &sk->gwy; + } + + /* translate ICMP message types and codes */ + if (pf_translate_icmp_af(pd->naf, + pd->hdr.icmp)) { + return PF_DROP; + } + + if (pf_lazy_makewritable(pd, pbuf, + off2 + 8) == NULL) { + return PF_DROP; + } + + pbuf_copy_back(pbuf, pd->off, + sizeof(struct icmp6_hdr), + pd->hdr.icmp6); + + /* + * translate inner ip header within the + * ICMP message + */ + if (pf_change_icmp_af(pbuf, ipoff2, pd, + &pd2, &saddr2->addr, &daddr2->addr, + pd->af, pd->naf)) { + return PF_DROP; + } + + if (pd->naf == AF_INET) { + pd->proto = IPPROTO_ICMP; + } else { + pd->proto = IPPROTO_ICMPV6; + } + + /* + * translate inner tcp header within + * the ICMP message + */ + pf_change_ap(direction, NULL, pd2.src, + &th.th_sport, pd2.ip_sum, + &th.th_sum, &daddr2->addr, + saddr2->xport.port, 0, pd2.af, + pd2.naf, 0); + + pf_change_ap(direction, NULL, pd2.dst, + &th.th_dport, pd2.ip_sum, + &th.th_sum, &saddr2->addr, + daddr2->xport.port, 0, pd2.af, + pd2.naf, 0); + + pbuf_copy_back(pbuf, pd2.off, 8, &th); + + /* translate outer ip header */ + PF_ACPY(&pd->naddr, &daddr2->addr, + pd->naf); + PF_ACPY(&pd->ndaddr, &saddr2->addr, + pd->naf); + if (pd->af == AF_INET) { + memcpy(&pd->naddr.addr32[3], + &srcv4_inaddr, + sizeof(pd->naddr.addr32[3])); + return pf_nat64_ipv4(pbuf, off, + pd); + } else { + return pf_nat64_ipv6(pbuf, off, + pd); + } + } if (direction == PF_IN) { pf_change_icmp(pd2.src, &th.th_sport, - daddr, &(*state)->state_key->lan.addr, - (*state)->state_key->lan.xport.port, NULL, + daddr, &sk->lan.addr, + sk->lan.xport.port, NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, pd2.af); } else { pf_change_icmp(pd2.dst, &th.th_dport, - saddr, &(*state)->state_key->gwy.addr, - (*state)->state_key->gwy.xport.port, NULL, + saddr, &sk->gwy.addr, + sk->gwy.xport.port, NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, pd2.af); } @@ -6866,79 +8177,79 @@ pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, } if (copyback) { - m = pf_lazy_makewritable(pd, m, off2 + 8); - if (!m) - return (PF_DROP); + if (pf_lazy_makewritable(pd, pbuf, off2 + 8) == + NULL) { + return PF_DROP; + } switch (pd2.af) { #if INET case AF_INET: - m_copyback(m, off, ICMP_MINLEN, + pbuf_copy_back(pbuf, off, ICMP_MINLEN, pd->hdr.icmp); - m_copyback(m, ipoff2, sizeof (h2), + pbuf_copy_back(pbuf, ipoff2, sizeof(h2), &h2); break; #endif /* INET */ #if INET6 case AF_INET6: - m_copyback(m, off, - sizeof (struct icmp6_hdr), + pbuf_copy_back(pbuf, off, + sizeof(struct icmp6_hdr), pd->hdr.icmp6); - m_copyback(m, ipoff2, sizeof (h2_6), - &h2_6); + pbuf_copy_back(pbuf, ipoff2, + sizeof(h2_6), &h2_6); break; #endif /* INET6 */ } - m_copyback(m, off2, 8, &th); + pbuf_copy_back(pbuf, off2, 8, &th); } - return (PF_PASS); - break; + return PF_PASS; } case IPPROTO_UDP: { - struct udphdr uh; + struct udphdr uh; int dx, action; - if (!pf_pull_hdr(m, off2, &uh, sizeof (uh), + if (!pf_pull_hdr(pbuf, off2, &uh, sizeof(uh), NULL, reason, pd2.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: ICMP error message too short " "(udp)\n")); - return (PF_DROP); + return PF_DROP; } - key.af = pd2.af; - key.proto = IPPROTO_UDP; - if (direction == PF_IN) { - PF_ACPY(&key.ext.addr, pd2.dst, key.af); - PF_ACPY(&key.gwy.addr, pd2.src, key.af); - key.ext.xport.port = uh.uh_dport; - key.gwy.xport.port = uh.uh_sport; - dx = PF_IN; - } else { - PF_ACPY(&key.lan.addr, pd2.dst, key.af); - PF_ACPY(&key.ext.addr, pd2.src, key.af); - key.lan.xport.port = uh.uh_dport; - key.ext.xport.port = uh.uh_sport; - dx = PF_OUT; - } + key.af_gwy = pd2.af; + PF_ACPY(&key.ext_gwy.addr, pd2.dst, key.af_gwy); + PF_ACPY(&key.gwy.addr, pd2.src, key.af_gwy); + key.ext_gwy.xport.port = uh.uh_dport; + key.gwy.xport.port = uh.uh_sport; + + key.af_lan = pd2.af; + PF_ACPY(&key.lan.addr, pd2.dst, key.af_lan); + PF_ACPY(&key.ext_lan.addr, pd2.src, key.af_lan); + key.lan.xport.port = uh.uh_dport; + key.ext_lan.xport.port = uh.uh_sport; + key.proto = IPPROTO_UDP; key.proto_variant = PF_EXTFILTER_APD; + dx = direction; if (ntohs(uh.uh_sport) == PF_IKE_PORT && ntohs(uh.uh_dport) == PF_IKE_PORT) { struct pf_ike_hdr ike; - size_t plen = - m->m_pkthdr.len - off2 - sizeof (uh); + size_t plen = pbuf->pb_packet_len - off2 - + sizeof(uh); if (direction == PF_IN && plen < 8 /* PF_IKE_PACKET_MINSIZE */) { DPFPRINTF(PF_DEBUG_MISC, ("pf: " "ICMP error, embedded IKE message " "too small.\n")); - return (PF_DROP); + return PF_DROP; } - if (plen > sizeof (ike)) - plen = sizeof (ike); - m_copydata(m, off + sizeof (uh), plen, &ike); + if (plen > sizeof(ike)) { + plen = sizeof(ike); + } + pbuf_copy_data(pbuf, off + sizeof(uh), plen, + &ike); key.app_state = &as; as.compare_lan_ext = pf_ike_compare; @@ -6974,219 +8285,317 @@ pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif, } } - if (pf_state_lookup_aux(state, kif, direction, &action)) - return (action); + if (pf_state_lookup_aux(state, kif, direction, &action)) { + return action; + } + + sk = (*state)->state_key; + pd->naf = pd2.naf = (pd2.af == sk->af_lan) ? + sk->af_gwy : sk->af_lan; + + if (STATE_TRANSLATE(sk)) { + /* NAT64 case */ + if (sk->af_lan != sk->af_gwy) { + struct pf_state_host *saddr2, *daddr2; + + if (pd2.naf == sk->af_lan) { + saddr2 = &sk->lan; + daddr2 = &sk->ext_lan; + } else { + saddr2 = &sk->ext_gwy; + daddr2 = &sk->gwy; + } + + /* translate ICMP message */ + if (pf_translate_icmp_af(pd->naf, + pd->hdr.icmp)) { + return PF_DROP; + } + if (pf_lazy_makewritable(pd, pbuf, + off2 + 8) == NULL) { + return PF_DROP; + } + + pbuf_copy_back(pbuf, pd->off, + sizeof(struct icmp6_hdr), + pd->hdr.icmp6); + + /* + * translate inner ip header within the + * ICMP message + */ + if (pf_change_icmp_af(pbuf, ipoff2, pd, + &pd2, &saddr2->addr, &daddr2->addr, + pd->af, pd->naf)) { + return PF_DROP; + } + + if (pd->naf == AF_INET) { + pd->proto = IPPROTO_ICMP; + } else { + pd->proto = IPPROTO_ICMPV6; + } - if (STATE_TRANSLATE((*state)->state_key)) { + /* + * translate inner udp header within + * the ICMP message + */ + pf_change_ap(direction, NULL, pd2.src, + &uh.uh_sport, pd2.ip_sum, + &uh.uh_sum, &daddr2->addr, + saddr2->xport.port, 0, pd2.af, + pd2.naf, 0); + + pf_change_ap(direction, NULL, pd2.dst, + &uh.uh_dport, pd2.ip_sum, + &uh.uh_sum, &saddr2->addr, + daddr2->xport.port, 0, pd2.af, + pd2.naf, 0); + + pbuf_copy_back(pbuf, pd2.off, + sizeof(uh), &uh); + + /* translate outer ip header */ + PF_ACPY(&pd->naddr, &daddr2->addr, + pd->naf); + PF_ACPY(&pd->ndaddr, &saddr2->addr, + pd->naf); + if (pd->af == AF_INET) { + memcpy(&pd->naddr.addr32[3], + &srcv4_inaddr, + sizeof(pd->naddr.addr32[3])); + return pf_nat64_ipv4(pbuf, off, + pd); + } else { + return pf_nat64_ipv6(pbuf, off, + pd); + } + } if (direction == PF_IN) { pf_change_icmp(pd2.src, &uh.uh_sport, - daddr, &(*state)->state_key->lan.addr, - (*state)->state_key->lan.xport.port, &uh.uh_sum, + daddr, &sk->lan.addr, + sk->lan.xport.port, &uh.uh_sum, pd2.ip_sum, icmpsum, pd->ip_sum, 1, pd2.af); } else { pf_change_icmp(pd2.dst, &uh.uh_dport, - saddr, &(*state)->state_key->gwy.addr, - (*state)->state_key->gwy.xport.port, &uh.uh_sum, + saddr, &sk->gwy.addr, + sk->gwy.xport.port, &uh.uh_sum, pd2.ip_sum, icmpsum, pd->ip_sum, 1, pd2.af); } - m = pf_lazy_makewritable(pd, m, - off2 + sizeof (uh)); - if (!m) - return (PF_DROP); + if (pf_lazy_makewritable(pd, pbuf, + off2 + sizeof(uh)) == NULL) { + return PF_DROP; + } switch (pd2.af) { #if INET case AF_INET: - m_copyback(m, off, ICMP_MINLEN, + pbuf_copy_back(pbuf, off, ICMP_MINLEN, pd->hdr.icmp); - m_copyback(m, ipoff2, sizeof (h2), &h2); + pbuf_copy_back(pbuf, ipoff2, + sizeof(h2), &h2); break; #endif /* INET */ #if INET6 case AF_INET6: - m_copyback(m, off, - sizeof (struct icmp6_hdr), + pbuf_copy_back(pbuf, off, + sizeof(struct icmp6_hdr), pd->hdr.icmp6); - m_copyback(m, ipoff2, sizeof (h2_6), - &h2_6); + pbuf_copy_back(pbuf, ipoff2, + sizeof(h2_6), &h2_6); break; #endif /* INET6 */ } - m_copyback(m, off2, sizeof (uh), &uh); + pbuf_copy_back(pbuf, off2, sizeof(uh), &uh); } - return (PF_PASS); - break; + return PF_PASS; } #if INET case IPPROTO_ICMP: { - struct icmp iih; + struct icmp iih; - if (!pf_pull_hdr(m, off2, &iih, ICMP_MINLEN, + if (!pf_pull_hdr(pbuf, off2, &iih, ICMP_MINLEN, NULL, reason, pd2.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: ICMP error message too short i" "(icmp)\n")); - return (PF_DROP); + return PF_DROP; } - key.af = pd2.af; key.proto = IPPROTO_ICMP; - if (direction == PF_IN) { - PF_ACPY(&key.ext.addr, pd2.dst, key.af); - PF_ACPY(&key.gwy.addr, pd2.src, key.af); - key.ext.xport.port = 0; + if (direction == PF_IN) { + key.af_gwy = pd2.af; + PF_ACPY(&key.ext_gwy.addr, pd2.dst, key.af_gwy); + PF_ACPY(&key.gwy.addr, pd2.src, key.af_gwy); + key.ext_gwy.xport.port = 0; key.gwy.xport.port = iih.icmp_id; } else { - PF_ACPY(&key.lan.addr, pd2.dst, key.af); - PF_ACPY(&key.ext.addr, pd2.src, key.af); + key.af_lan = pd2.af; + PF_ACPY(&key.lan.addr, pd2.dst, key.af_lan); + PF_ACPY(&key.ext_lan.addr, pd2.src, key.af_lan); key.lan.xport.port = iih.icmp_id; - key.ext.xport.port = 0; + key.ext_lan.xport.port = 0; } STATE_LOOKUP(); - if (STATE_TRANSLATE((*state)->state_key)) { + sk = (*state)->state_key; + if (STATE_TRANSLATE(sk)) { if (direction == PF_IN) { pf_change_icmp(pd2.src, &iih.icmp_id, - daddr, &(*state)->state_key->lan.addr, - (*state)->state_key->lan.xport.port, NULL, + daddr, &sk->lan.addr, + sk->lan.xport.port, NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, AF_INET); } else { pf_change_icmp(pd2.dst, &iih.icmp_id, - saddr, &(*state)->state_key->gwy.addr, - (*state)->state_key->gwy.xport.port, NULL, + saddr, &sk->gwy.addr, + sk->gwy.xport.port, NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, AF_INET); } - m = pf_lazy_makewritable(pd, m, off2 + ICMP_MINLEN); - if (!m) - return (PF_DROP); - m_copyback(m, off, ICMP_MINLEN, pd->hdr.icmp); - m_copyback(m, ipoff2, sizeof (h2), &h2); - m_copyback(m, off2, ICMP_MINLEN, &iih); + if (pf_lazy_makewritable(pd, pbuf, + off2 + ICMP_MINLEN) == NULL) { + return PF_DROP; + } + pbuf_copy_back(pbuf, off, ICMP_MINLEN, + pd->hdr.icmp); + pbuf_copy_back(pbuf, ipoff2, sizeof(h2), &h2); + pbuf_copy_back(pbuf, off2, ICMP_MINLEN, &iih); } - return (PF_PASS); - break; + return PF_PASS; } #endif /* INET */ #if INET6 case IPPROTO_ICMPV6: { - struct icmp6_hdr iih; + struct icmp6_hdr iih; - if (!pf_pull_hdr(m, off2, &iih, - sizeof (struct icmp6_hdr), NULL, reason, pd2.af)) { + if (!pf_pull_hdr(pbuf, off2, &iih, + sizeof(struct icmp6_hdr), NULL, reason, pd2.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: ICMP error message too short " "(icmp6)\n")); - return (PF_DROP); + return PF_DROP; } - key.af = pd2.af; key.proto = IPPROTO_ICMPV6; - if (direction == PF_IN) { - PF_ACPY(&key.ext.addr, pd2.dst, key.af); - PF_ACPY(&key.gwy.addr, pd2.src, key.af); - key.ext.xport.port = 0; + if (direction == PF_IN) { + key.af_gwy = pd2.af; + PF_ACPY(&key.ext_gwy.addr, pd2.dst, key.af_gwy); + PF_ACPY(&key.gwy.addr, pd2.src, key.af_gwy); + key.ext_gwy.xport.port = 0; key.gwy.xport.port = iih.icmp6_id; } else { - PF_ACPY(&key.lan.addr, pd2.dst, key.af); - PF_ACPY(&key.ext.addr, pd2.src, key.af); + key.af_lan = pd2.af; + PF_ACPY(&key.lan.addr, pd2.dst, key.af_lan); + PF_ACPY(&key.ext_lan.addr, pd2.src, key.af_lan); key.lan.xport.port = iih.icmp6_id; - key.ext.xport.port = 0; + key.ext_lan.xport.port = 0; } STATE_LOOKUP(); - if (STATE_TRANSLATE((*state)->state_key)) { + sk = (*state)->state_key; + if (STATE_TRANSLATE(sk)) { if (direction == PF_IN) { pf_change_icmp(pd2.src, &iih.icmp6_id, - daddr, &(*state)->state_key->lan.addr, - (*state)->state_key->lan.xport.port, NULL, + daddr, &sk->lan.addr, + sk->lan.xport.port, NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, AF_INET6); } else { pf_change_icmp(pd2.dst, &iih.icmp6_id, - saddr, &(*state)->state_key->gwy.addr, - (*state)->state_key->gwy.xport.port, NULL, + saddr, &sk->gwy.addr, + sk->gwy.xport.port, NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, AF_INET6); } - m = pf_lazy_makewritable(pd, m, off2 + - sizeof (struct icmp6_hdr)); - if (!m) - return (PF_DROP); - m_copyback(m, off, sizeof (struct icmp6_hdr), - pd->hdr.icmp6); - m_copyback(m, ipoff2, sizeof (h2_6), &h2_6); - m_copyback(m, off2, sizeof (struct icmp6_hdr), - &iih); - } - - return (PF_PASS); - break; + if (pf_lazy_makewritable(pd, pbuf, off2 + + sizeof(struct icmp6_hdr)) == NULL) { + return PF_DROP; + } + pbuf_copy_back(pbuf, off, + sizeof(struct icmp6_hdr), pd->hdr.icmp6); + pbuf_copy_back(pbuf, ipoff2, sizeof(h2_6), + &h2_6); + pbuf_copy_back(pbuf, off2, + sizeof(struct icmp6_hdr), &iih); + } + + return PF_PASS; } #endif /* INET6 */ default: { - key.af = pd2.af; key.proto = pd2.proto; - if (direction == PF_IN) { - PF_ACPY(&key.ext.addr, pd2.dst, key.af); - PF_ACPY(&key.gwy.addr, pd2.src, key.af); - key.ext.xport.port = 0; + if (direction == PF_IN) { + key.af_gwy = pd2.af; + PF_ACPY(&key.ext_gwy.addr, pd2.dst, key.af_gwy); + PF_ACPY(&key.gwy.addr, pd2.src, key.af_gwy); + key.ext_gwy.xport.port = 0; key.gwy.xport.port = 0; } else { - PF_ACPY(&key.lan.addr, pd2.dst, key.af); - PF_ACPY(&key.ext.addr, pd2.src, key.af); + key.af_lan = pd2.af; + PF_ACPY(&key.lan.addr, pd2.dst, key.af_lan); + PF_ACPY(&key.ext_lan.addr, pd2.src, key.af_lan); key.lan.xport.port = 0; - key.ext.xport.port = 0; + key.ext_lan.xport.port = 0; } STATE_LOOKUP(); - if (STATE_TRANSLATE((*state)->state_key)) { + sk = (*state)->state_key; + if (STATE_TRANSLATE(sk)) { if (direction == PF_IN) { - pf_change_icmp(pd2.src, NULL, - daddr, &(*state)->state_key->lan.addr, - 0, NULL, + pf_change_icmp(pd2.src, NULL, daddr, + &sk->lan.addr, 0, NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, pd2.af); } else { - pf_change_icmp(pd2.dst, NULL, - saddr, &(*state)->state_key->gwy.addr, - 0, NULL, + pf_change_icmp(pd2.dst, NULL, saddr, + &sk->gwy.addr, 0, NULL, pd2.ip_sum, icmpsum, pd->ip_sum, 0, pd2.af); } switch (pd2.af) { #if INET case AF_INET: - m = pf_lazy_makewritable(pd, m, - ipoff2 + sizeof (h2)); - if (!m) - return (PF_DROP); + if (pf_lazy_makewritable(pd, pbuf, + ipoff2 + sizeof(h2)) == NULL) { + return PF_DROP; + } + /* + * + * Xnu was missing the following... + */ + pbuf_copy_back(pbuf, off, ICMP_MINLEN, + pd->hdr.icmp); + pbuf_copy_back(pbuf, ipoff2, + sizeof(h2), &h2); + break; + /* + * + */ #endif /* INET */ #if INET6 case AF_INET6: - m = pf_lazy_makewritable(pd, m, - ipoff2 + sizeof (h2_6)); - if (!m) - return (PF_DROP); - m_copyback(m, off, - sizeof (struct icmp6_hdr), + if (pf_lazy_makewritable(pd, pbuf, + ipoff2 + sizeof(h2_6)) == NULL) { + return PF_DROP; + } + pbuf_copy_back(pbuf, off, + sizeof(struct icmp6_hdr), pd->hdr.icmp6); - m_copyback(m, ipoff2, sizeof (h2_6), - &h2_6); + pbuf_copy_back(pbuf, ipoff2, + sizeof(h2_6), &h2_6); break; #endif /* INET6 */ } } - return (PF_PASS); - break; + return PF_PASS; } } } @@ -7200,20 +8609,20 @@ pf_test_state_grev1(struct pf_state **state, int direction, struct pf_state_peer *dst; struct pf_state_key_cmp key; struct pf_grev1_hdr *grev1 = pd->hdr.grev1; - struct mbuf *m; key.app_state = 0; - key.af = pd->af; key.proto = IPPROTO_GRE; key.proto_variant = PF_GRE_PPTP_VARIANT; - if (direction == PF_IN) { - PF_ACPY(&key.ext.addr, pd->src, key.af); - PF_ACPY(&key.gwy.addr, pd->dst, key.af); + if (direction == PF_IN) { + key.af_gwy = pd->af; + PF_ACPY(&key.ext_gwy.addr, pd->src, key.af_gwy); + PF_ACPY(&key.gwy.addr, pd->dst, key.af_gwy); key.gwy.xport.call_id = grev1->call_id; } else { - PF_ACPY(&key.lan.addr, pd->src, key.af); - PF_ACPY(&key.ext.addr, pd->dst, key.af); - key.ext.xport.call_id = grev1->call_id; + key.af_lan = pd->af; + PF_ACPY(&key.lan.addr, pd->src, key.af_lan); + PF_ACPY(&key.ext_lan.addr, pd->dst, key.af_lan); + key.ext_lan.xport.call_id = grev1->call_id; } STATE_LOOKUP(); @@ -7227,24 +8636,27 @@ pf_test_state_grev1(struct pf_state **state, int direction, } /* update states */ - if (src->state < PFGRE1S_INITIATING) + if (src->state < PFGRE1S_INITIATING) { src->state = PFGRE1S_INITIATING; + } /* update expire time */ (*state)->expire = pf_time_second(); if (src->state >= PFGRE1S_INITIATING && dst->state >= PFGRE1S_INITIATING) { - if ((*state)->timeout != PFTM_TCP_ESTABLISHED) + if ((*state)->timeout != PFTM_TCP_ESTABLISHED) { (*state)->timeout = PFTM_GREv1_ESTABLISHED; + } src->state = PFGRE1S_ESTABLISHED; dst->state = PFGRE1S_ESTABLISHED; } else { (*state)->timeout = PFTM_GREv1_INITIATING; } - if ((*state)->state_key->app_state) + if ((*state)->state_key->app_state) { (*state)->state_key->app_state->u.grev1.pptp_state->expire = pf_time_second(); + } /* translate source/destination address, if necessary */ if (STATE_GRE_TRANSLATE((*state)->state_key)) { @@ -7252,9 +8664,9 @@ pf_test_state_grev1(struct pf_state **state, int direction, switch (pd->af) { #if INET case AF_INET: - pf_change_a(&pd->src->v4.s_addr, + pf_change_a(&pd->src->v4addr.s_addr, pd->ip_sum, - (*state)->state_key->gwy.addr.v4.s_addr, 0); + (*state)->state_key->gwy.addr.v4addr.s_addr, 0); break; #endif /* INET */ #if INET6 @@ -7270,9 +8682,9 @@ pf_test_state_grev1(struct pf_state **state, int direction, switch (pd->af) { #if INET case AF_INET: - pf_change_a(&pd->dst->v4.s_addr, + pf_change_a(&pd->dst->v4addr.s_addr, pd->ip_sum, - (*state)->state_key->lan.addr.v4.s_addr, 0); + (*state)->state_key->lan.addr.v4addr.s_addr, 0); break; #endif /* INET */ #if INET6 @@ -7284,13 +8696,14 @@ pf_test_state_grev1(struct pf_state **state, int direction, } } - m = pf_lazy_makewritable(pd, pd->mp, off + sizeof (*grev1)); - if (!m) - return (PF_DROP); - m_copyback(m, off, sizeof (*grev1), grev1); + if (pf_lazy_makewritable(pd, pd->mp, off + sizeof(*grev1)) == + NULL) { + return PF_DROP; + } + pbuf_copy_back(pd->mp, off, sizeof(*grev1), grev1); } - return (PF_PASS); + return PF_PASS; } static int @@ -7304,17 +8717,18 @@ pf_test_state_esp(struct pf_state **state, int direction, struct pfi_kif *kif, struct pf_esp_hdr *esp = pd->hdr.esp; int action; - memset(&key, 0, sizeof (key)); - key.af = pd->af; + memset(&key, 0, sizeof(key)); key.proto = IPPROTO_ESP; - if (direction == PF_IN) { - PF_ACPY(&key.ext.addr, pd->src, key.af); - PF_ACPY(&key.gwy.addr, pd->dst, key.af); + if (direction == PF_IN) { + key.af_gwy = pd->af; + PF_ACPY(&key.ext_gwy.addr, pd->src, key.af_gwy); + PF_ACPY(&key.gwy.addr, pd->dst, key.af_gwy); key.gwy.xport.spi = esp->spi; } else { - PF_ACPY(&key.lan.addr, pd->src, key.af); - PF_ACPY(&key.ext.addr, pd->dst, key.af); - key.ext.xport.spi = esp->spi; + key.af_lan = pd->af; + PF_ACPY(&key.lan.addr, pd->src, key.af_lan); + PF_ACPY(&key.ext_lan.addr, pd->dst, key.af_lan); + key.ext_lan.xport.spi = esp->spi; } *state = pf_find_state(kif, &key, direction); @@ -7341,13 +8755,14 @@ pf_test_state_esp(struct pf_state **state, int direction, struct pfi_kif *kif, esp->spi; if (RB_INSERT(pf_state_tree_ext_gwy, - &pf_statetbl_ext_gwy, sk)) + &pf_statetbl_ext_gwy, sk)) { pf_detach_state(s, PF_DT_SKIP_EXTGWY); - else + } else { *state = s; + } } } else { - key.ext.xport.spi = 0; + key.ext_lan.xport.spi = 0; s = pf_find_state(kif, &key, direction); if (s) { @@ -7355,28 +8770,30 @@ pf_test_state_esp(struct pf_state **state, int direction, struct pfi_kif *kif, RB_REMOVE(pf_state_tree_lan_ext, &pf_statetbl_lan_ext, sk); - sk->ext.xport.spi = esp->spi; + sk->ext_lan.xport.spi = esp->spi; if (RB_INSERT(pf_state_tree_lan_ext, - &pf_statetbl_lan_ext, sk)) + &pf_statetbl_lan_ext, sk)) { pf_detach_state(s, PF_DT_SKIP_LANEXT); - else + } else { *state = s; + } } } if (s) { if (*state == 0) { #if NPFSYNC - if (s->creatorid == pf_status.hostid) + if (s->creatorid == pf_status.hostid) { pfsync_delete_state(s); + } #endif s->timeout = PFTM_UNLINKED; hook_runloop(&s->unlink_hooks, - HOOK_REMOVE|HOOK_FREE); + HOOK_REMOVE | HOOK_FREE); pf_src_tree_remove_state(s); pf_free_state(s); - return (PF_DROP); + return PF_DROP; } } } @@ -7391,8 +8808,9 @@ pf_test_state_esp(struct pf_state **state, int direction, struct pfi_kif *kif, } } - if (pf_state_lookup_aux(state, kif, direction, &action)) - return (action); + if (pf_state_lookup_aux(state, kif, direction, &action)) { + return action; + } if (direction == (*state)->state_key->direction) { src = &(*state)->src; @@ -7403,8 +8821,9 @@ pf_test_state_esp(struct pf_state **state, int direction, struct pfi_kif *kif, } /* update states */ - if (src->state < PFESPS_INITIATING) + if (src->state < PFESPS_INITIATING) { src->state = PFESPS_INITIATING; + } /* update expire time */ (*state)->expire = pf_time_second(); @@ -7422,9 +8841,9 @@ pf_test_state_esp(struct pf_state **state, int direction, struct pfi_kif *kif, switch (pd->af) { #if INET case AF_INET: - pf_change_a(&pd->src->v4.s_addr, + pf_change_a(&pd->src->v4addr.s_addr, pd->ip_sum, - (*state)->state_key->gwy.addr.v4.s_addr, 0); + (*state)->state_key->gwy.addr.v4addr.s_addr, 0); break; #endif /* INET */ #if INET6 @@ -7438,9 +8857,9 @@ pf_test_state_esp(struct pf_state **state, int direction, struct pfi_kif *kif, switch (pd->af) { #if INET case AF_INET: - pf_change_a(&pd->dst->v4.s_addr, + pf_change_a(&pd->dst->v4addr.s_addr, pd->ip_sum, - (*state)->state_key->lan.addr.v4.s_addr, 0); + (*state)->state_key->lan.addr.v4addr.s_addr, 0); break; #endif /* INET */ #if INET6 @@ -7453,29 +8872,30 @@ pf_test_state_esp(struct pf_state **state, int direction, struct pfi_kif *kif, } } - return (PF_PASS); + return PF_PASS; } static int pf_test_state_other(struct pf_state **state, int direction, struct pfi_kif *kif, struct pf_pdesc *pd) { - struct pf_state_peer *src, *dst; - struct pf_state_key_cmp key; + struct pf_state_peer *src, *dst; + struct pf_state_key_cmp key; key.app_state = 0; - key.af = pd->af; key.proto = pd->proto; - if (direction == PF_IN) { - PF_ACPY(&key.ext.addr, pd->src, key.af); - PF_ACPY(&key.gwy.addr, pd->dst, key.af); - key.ext.xport.port = 0; + if (direction == PF_IN) { + key.af_gwy = pd->af; + PF_ACPY(&key.ext_gwy.addr, pd->src, key.af_gwy); + PF_ACPY(&key.gwy.addr, pd->dst, key.af_gwy); + key.ext_gwy.xport.port = 0; key.gwy.xport.port = 0; } else { - PF_ACPY(&key.lan.addr, pd->src, key.af); - PF_ACPY(&key.ext.addr, pd->dst, key.af); + key.af_lan = pd->af; + PF_ACPY(&key.lan.addr, pd->src, key.af_lan); + PF_ACPY(&key.ext_lan.addr, pd->dst, key.af_lan); key.lan.xport.port = 0; - key.ext.xport.port = 0; + key.ext_lan.xport.port = 0; } STATE_LOOKUP(); @@ -7489,17 +8909,20 @@ pf_test_state_other(struct pf_state **state, int direction, struct pfi_kif *kif, } /* update states */ - if (src->state < PFOTHERS_SINGLE) + if (src->state < PFOTHERS_SINGLE) { src->state = PFOTHERS_SINGLE; - if (dst->state == PFOTHERS_SINGLE) + } + if (dst->state == PFOTHERS_SINGLE) { dst->state = PFOTHERS_MULTIPLE; + } /* update expire time */ (*state)->expire = pf_time_second(); - if (src->state == PFOTHERS_MULTIPLE && dst->state == PFOTHERS_MULTIPLE) + if (src->state == PFOTHERS_MULTIPLE && dst->state == PFOTHERS_MULTIPLE) { (*state)->timeout = PFTM_OTHER_MULTIPLE; - else + } else { (*state)->timeout = PFTM_OTHER_SINGLE; + } /* translate source/destination address, if necessary */ if (STATE_ADDR_TRANSLATE((*state)->state_key)) { @@ -7507,9 +8930,9 @@ pf_test_state_other(struct pf_state **state, int direction, struct pfi_kif *kif, switch (pd->af) { #if INET case AF_INET: - pf_change_a(&pd->src->v4.s_addr, + pf_change_a(&pd->src->v4addr.s_addr, pd->ip_sum, - (*state)->state_key->gwy.addr.v4.s_addr, + (*state)->state_key->gwy.addr.v4addr.s_addr, 0); break; #endif /* INET */ @@ -7524,9 +8947,9 @@ pf_test_state_other(struct pf_state **state, int direction, struct pfi_kif *kif, switch (pd->af) { #if INET case AF_INET: - pf_change_a(&pd->dst->v4.s_addr, + pf_change_a(&pd->dst->v4addr.s_addr, pd->ip_sum, - (*state)->state_key->lan.addr.v4.s_addr, + (*state)->state_key->lan.addr.v4addr.s_addr, 0); break; #endif /* INET */ @@ -7540,7 +8963,7 @@ pf_test_state_other(struct pf_state **state, int direction, struct pfi_kif *kif, } } - return (PF_PASS); + return PF_PASS; } /* @@ -7548,14 +8971,14 @@ pf_test_state_other(struct pf_state **state, int direction, struct pfi_kif *kif, * h must be at "ipoff" on the mbuf chain. */ void * -pf_pull_hdr(struct mbuf *m, int off, void *p, int len, +pf_pull_hdr(pbuf_t *pbuf, int off, void *p, int len, u_short *actionp, u_short *reasonp, sa_family_t af) { switch (af) { #if INET case AF_INET: { - struct ip *h = mtod(m, struct ip *); - u_int16_t fragoff = (ntohs(h->ip_off) & IP_OFFMASK) << 3; + struct ip *h = pbuf->pb_data; + u_int16_t fragoff = (ntohs(h->ip_off) & IP_OFFMASK) << 3; if (fragoff) { if (fragoff >= len) { @@ -7564,113 +8987,114 @@ pf_pull_hdr(struct mbuf *m, int off, void *p, int len, ACTION_SET(actionp, PF_DROP); REASON_SET(reasonp, PFRES_FRAG); } - return (NULL); + return NULL; } - if (m->m_pkthdr.len < off + len || + if (pbuf->pb_packet_len < (unsigned)(off + len) || ntohs(h->ip_len) < off + len) { ACTION_SET(actionp, PF_DROP); REASON_SET(reasonp, PFRES_SHORT); - return (NULL); + return NULL; } break; } #endif /* INET */ #if INET6 case AF_INET6: { - struct ip6_hdr *h = mtod(m, struct ip6_hdr *); + struct ip6_hdr *h = pbuf->pb_data; - if (m->m_pkthdr.len < off + len || - (ntohs(h->ip6_plen) + sizeof (struct ip6_hdr)) < + if (pbuf->pb_packet_len < (unsigned)(off + len) || + (ntohs(h->ip6_plen) + sizeof(struct ip6_hdr)) < (unsigned)(off + len)) { ACTION_SET(actionp, PF_DROP); REASON_SET(reasonp, PFRES_SHORT); - return (NULL); + return NULL; } break; } #endif /* INET6 */ } - m_copydata(m, off, len, p); - return (p); + pbuf_copy_data(pbuf, off, len, p); + return p; } int pf_routable(struct pf_addr *addr, sa_family_t af, struct pfi_kif *kif) { #pragma unused(kif) - struct sockaddr_in *dst; - int ret = 1; + struct sockaddr_in *dst; + int ret = 1; #if INET6 - struct sockaddr_in6 *dst6; - struct route_in6 ro; + struct sockaddr_in6 *dst6; + struct route_in6 ro; #else - struct route ro; + struct route ro; #endif - bzero(&ro, sizeof (ro)); + bzero(&ro, sizeof(ro)); switch (af) { case AF_INET: dst = satosin(&ro.ro_dst); dst->sin_family = AF_INET; - dst->sin_len = sizeof (*dst); - dst->sin_addr = addr->v4; + dst->sin_len = sizeof(*dst); + dst->sin_addr = addr->v4addr; break; #if INET6 case AF_INET6: dst6 = (struct sockaddr_in6 *)&ro.ro_dst; dst6->sin6_family = AF_INET6; - dst6->sin6_len = sizeof (*dst6); - dst6->sin6_addr = addr->v6; + dst6->sin6_len = sizeof(*dst6); + dst6->sin6_addr = addr->v6addr; break; #endif /* INET6 */ default: - return (0); + return 0; } /* XXX: IFT_ENC is not currently used by anything*/ /* Skip checks for ipsec interfaces */ - if (kif != NULL && kif->pfik_ifp->if_type == IFT_ENC) + if (kif != NULL && kif->pfik_ifp->if_type == IFT_ENC) { goto out; + } /* XXX: what is the point of this? */ rtalloc((struct route *)&ro); out: ROUTE_RELEASE(&ro); - return (ret); + return ret; } int pf_rtlabel_match(struct pf_addr *addr, sa_family_t af, struct pf_addr_wrap *aw) { #pragma unused(aw) - struct sockaddr_in *dst; + struct sockaddr_in *dst; #if INET6 - struct sockaddr_in6 *dst6; - struct route_in6 ro; + struct sockaddr_in6 *dst6; + struct route_in6 ro; #else - struct route ro; + struct route ro; #endif - int ret = 0; + int ret = 0; - bzero(&ro, sizeof (ro)); + bzero(&ro, sizeof(ro)); switch (af) { case AF_INET: dst = satosin(&ro.ro_dst); dst->sin_family = AF_INET; - dst->sin_len = sizeof (*dst); - dst->sin_addr = addr->v4; + dst->sin_len = sizeof(*dst); + dst->sin_addr = addr->v4addr; break; #if INET6 case AF_INET6: dst6 = (struct sockaddr_in6 *)&ro.ro_dst; dst6->sin6_family = AF_INET6; - dst6->sin6_len = sizeof (*dst6); - dst6->sin6_addr = addr->v6; + dst6->sin6_len = sizeof(*dst6); + dst6->sin6_addr = addr->v6addr; break; #endif /* INET6 */ default: - return (0); + return 0; } /* XXX: what is the point of this? */ @@ -7678,50 +9102,64 @@ pf_rtlabel_match(struct pf_addr *addr, sa_family_t af, struct pf_addr_wrap *aw) ROUTE_RELEASE(&ro); - return (ret); + return ret; } #if INET static void -pf_route(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, +pf_route(pbuf_t **pbufp, struct pf_rule *r, int dir, struct ifnet *oifp, struct pf_state *s, struct pf_pdesc *pd) { #pragma unused(pd) - struct mbuf *m0, *m1; - struct route iproute; - struct route *ro = &iproute; - struct sockaddr_in *dst; - struct ip *ip; - struct ifnet *ifp = NULL; - struct pf_addr naddr; - struct pf_src_node *sn = NULL; - int error = 0; - uint32_t sw_csum; - - bzero(&iproute, sizeof (iproute)); - - if (m == NULL || *m == NULL || r == NULL || - (dir != PF_IN && dir != PF_OUT) || oifp == NULL) + struct mbuf *m0, *m1; + struct route iproute; + struct route *ro = &iproute; + struct sockaddr_in *dst; + struct ip *ip; + struct ifnet *ifp = NULL; + struct pf_addr naddr; + struct pf_src_node *sn = NULL; + int error = 0; + uint32_t sw_csum; + int interface_mtu = 0; + bzero(&iproute, sizeof(iproute)); + + if (pbufp == NULL || !pbuf_is_valid(*pbufp) || r == NULL || + (dir != PF_IN && dir != PF_OUT) || oifp == NULL) { panic("pf_route: invalid parameters"); + } if (pd->pf_mtag->pftag_routed++ > 3) { - m0 = *m; - *m = NULL; + pbuf_destroy(*pbufp); + *pbufp = NULL; + m0 = NULL; goto bad; } + /* + * Since this is something of an edge case and may involve the + * host stack (for routing, at least for now), we convert the + * incoming pbuf into an mbuf. + */ if (r->rt == PF_DUPTO) { - if ((m0 = m_copym(*m, 0, M_COPYALL, M_NOWAIT)) == NULL) - return; + m0 = pbuf_clone_to_mbuf(*pbufp); + } else if ((r->rt == PF_REPLYTO) == (r->direction == dir)) { + return; } else { - if ((r->rt == PF_REPLYTO) == (r->direction == dir)) - return; - m0 = *m; + /* We're going to consume this packet */ + m0 = pbuf_to_mbuf(*pbufp, TRUE); + *pbufp = NULL; + } + + if (m0 == NULL) { + goto bad; } - if (m0->m_len < (int)sizeof (struct ip)) { + /* We now have the packet in an mbuf (m0) */ + + if (m0->m_len < (int)sizeof(struct ip)) { DPFPRINTF(PF_DEBUG_URGENT, - ("pf_route: m0->m_len < sizeof (struct ip)\n")); + ("pf_route: packet length < sizeof (struct ip)\n")); goto bad; } @@ -7729,7 +9167,7 @@ pf_route(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, dst = satosin((void *)&ro->ro_dst); dst->sin_family = AF_INET; - dst->sin_len = sizeof (*dst); + dst->sin_len = sizeof(*dst); dst->sin_addr = ip->ip_dst; if (r->rt == PF_FASTROUTE) { @@ -7743,8 +9181,9 @@ pf_route(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, RT_LOCK(ro->ro_rt); ro->ro_rt->rt_use++; - if (ro->ro_rt->rt_flags & RTF_GATEWAY) + if (ro->ro_rt->rt_flags & RTF_GATEWAY) { dst = satosin((void *)ro->ro_rt->rt_gateway); + } RT_UNLOCK(ro->ro_rt); } else { if (TAILQ_EMPTY(&r->rpool.list)) { @@ -7755,28 +9194,32 @@ pf_route(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, if (s == NULL) { pf_map_addr(AF_INET, r, (struct pf_addr *)&ip->ip_src, &naddr, NULL, &sn); - if (!PF_AZERO(&naddr, AF_INET)) - dst->sin_addr.s_addr = naddr.v4.s_addr; + if (!PF_AZERO(&naddr, AF_INET)) { + dst->sin_addr.s_addr = naddr.v4addr.s_addr; + } ifp = r->rpool.cur->kif ? r->rpool.cur->kif->pfik_ifp : NULL; } else { - if (!PF_AZERO(&s->rt_addr, AF_INET)) + if (!PF_AZERO(&s->rt_addr, AF_INET)) { dst->sin_addr.s_addr = - s->rt_addr.v4.s_addr; + s->rt_addr.v4addr.s_addr; + } ifp = s->rt_kif ? s->rt_kif->pfik_ifp : NULL; } } - if (ifp == NULL) + if (ifp == NULL) { goto bad; + } if (oifp != ifp) { - if (pf_test(PF_OUT, ifp, &m0, NULL, NULL) != PF_PASS) + if (pf_test_mbuf(PF_OUT, ifp, &m0, NULL, NULL) != PF_PASS) { goto bad; - else if (m0 == NULL) + } else if (m0 == NULL) { goto done; - if (m0->m_len < (int)sizeof (struct ip)) { + } + if (m0->m_len < (int)sizeof(struct ip)) { DPFPRINTF(PF_DEBUG_URGENT, - ("pf_route: m0->m_len < sizeof (struct ip)\n")); + ("pf_route: packet length < sizeof (struct ip)\n")); goto bad; } ip = mtod(m0, struct ip *); @@ -7786,7 +9229,15 @@ pf_route(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, ip_output_checksum(ifp, m0, ((ip->ip_hl) << 2), ntohs(ip->ip_len), &sw_csum); - if (ntohs(ip->ip_len) <= ifp->if_mtu || TSO_IPV4_OK(ifp, m0) || + interface_mtu = ifp->if_mtu; + + if (INTF_ADJUST_MTU_FOR_CLAT46(ifp)) { + interface_mtu = IN6_LINKMTU(ifp); + /* Further adjust the size for CLAT46 expansion */ + interface_mtu -= CLAT46_HDR_EXPANSION_OVERHD; + } + + if (ntohs(ip->ip_len) <= interface_mtu || TSO_IPV4_OK(ifp, m0) || (!(ip->ip_off & htons(IP_DF)) && (ifp->if_hwassist & CSUM_FRAGMENT))) { ip->ip_sum = 0; @@ -7809,10 +9260,11 @@ pf_route(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, ipstat.ips_cantfrag++; if (r->rt != PF_DUPTO) { icmp_error(m0, ICMP_UNREACH, ICMP_UNREACH_NEEDFRAG, 0, - ifp->if_mtu); + interface_mtu); goto done; - } else + } else { goto bad; + } } m1 = m0; @@ -7822,7 +9274,7 @@ pf_route(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, NTOHS(ip->ip_off); NTOHS(ip->ip_len); #endif - error = ip_fragment(m0, ifp, ifp->if_mtu, sw_csum); + error = ip_fragment(m0, ifp, interface_mtu, sw_csum); if (error) { m0 = NULL; @@ -7832,65 +9284,79 @@ pf_route(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, for (m0 = m1; m0; m0 = m1) { m1 = m0->m_nextpkt; m0->m_nextpkt = 0; - if (error == 0) + if (error == 0) { error = ifnet_output(ifp, PF_INET, m0, ro->ro_rt, sintosa(dst)); - else + } else { m_freem(m0); + } } - if (error == 0) + if (error == 0) { ipstat.ips_fragmented++; + } done: - if (r->rt != PF_DUPTO) - *m = NULL; - ROUTE_RELEASE(&iproute); return; bad: - m_freem(m0); + if (m0) { + m_freem(m0); + } goto done; } #endif /* INET */ #if INET6 static void -pf_route6(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, +pf_route6(pbuf_t **pbufp, struct pf_rule *r, int dir, struct ifnet *oifp, struct pf_state *s, struct pf_pdesc *pd) { #pragma unused(pd) - struct mbuf *m0; - struct route_in6 ip6route; - struct route_in6 *ro; - struct sockaddr_in6 *dst; - struct ip6_hdr *ip6; - struct ifnet *ifp = NULL; - struct pf_addr naddr; - struct pf_src_node *sn = NULL; - int error = 0; - - if (m == NULL || *m == NULL || r == NULL || - (dir != PF_IN && dir != PF_OUT) || oifp == NULL) + struct mbuf *m0; + struct route_in6 ip6route; + struct route_in6 *ro; + struct sockaddr_in6 *dst; + struct ip6_hdr *ip6; + struct ifnet *ifp = NULL; + struct pf_addr naddr; + struct pf_src_node *sn = NULL; + int error = 0; + struct pf_mtag *pf_mtag; + + if (pbufp == NULL || !pbuf_is_valid(*pbufp) || r == NULL || + (dir != PF_IN && dir != PF_OUT) || oifp == NULL) { panic("pf_route6: invalid parameters"); + } if (pd->pf_mtag->pftag_routed++ > 3) { - m0 = *m; - *m = NULL; + pbuf_destroy(*pbufp); + *pbufp = NULL; + m0 = NULL; goto bad; } + /* + * Since this is something of an edge case and may involve the + * host stack (for routing, at least for now), we convert the + * incoming pbuf into an mbuf. + */ if (r->rt == PF_DUPTO) { - if ((m0 = m_copym(*m, 0, M_COPYALL, M_NOWAIT)) == NULL) - return; + m0 = pbuf_clone_to_mbuf(*pbufp); + } else if ((r->rt == PF_REPLYTO) == (r->direction == dir)) { + return; } else { - if ((r->rt == PF_REPLYTO) == (r->direction == dir)) - return; - m0 = *m; + /* We're about to consume this packet */ + m0 = pbuf_to_mbuf(*pbufp, TRUE); + *pbufp = NULL; + } + + if (m0 == NULL) { + goto bad; } - if (m0->m_len < (int)sizeof (struct ip6_hdr)) { + if (m0->m_len < (int)sizeof(struct ip6_hdr)) { DPFPRINTF(PF_DEBUG_URGENT, ("pf_route6: m0->m_len < sizeof (struct ip6_hdr)\n")); goto bad; @@ -7898,18 +9364,16 @@ pf_route6(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, ip6 = mtod(m0, struct ip6_hdr *); ro = &ip6route; - bzero((caddr_t)ro, sizeof (*ro)); + bzero((caddr_t)ro, sizeof(*ro)); dst = (struct sockaddr_in6 *)&ro->ro_dst; dst->sin6_family = AF_INET6; - dst->sin6_len = sizeof (*dst); + dst->sin6_len = sizeof(*dst); dst->sin6_addr = ip6->ip6_dst; - /* Cheat. XXX why only in the v6 case??? */ + /* Cheat. XXX why only in the v6addr case??? */ if (r->rt == PF_FASTROUTE) { - struct pf_mtag *pf_mtag; - - if ((pf_mtag = pf_get_mtag(m0)) == NULL) - goto bad; + pf_mtag = pf_get_mtag(m0); + ASSERT(pf_mtag != NULL); pf_mtag->pftag_flags |= PF_TAG_GENERATED; ip6_output(m0, NULL, NULL, 0, NULL, NULL, NULL); return; @@ -7921,31 +9385,53 @@ pf_route6(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, goto bad; } if (s == NULL) { - pf_map_addr(AF_INET6, r, (struct pf_addr *)&ip6->ip6_src, + pf_map_addr(AF_INET6, r, (struct pf_addr *)(uintptr_t)&ip6->ip6_src, &naddr, NULL, &sn); - if (!PF_AZERO(&naddr, AF_INET6)) + if (!PF_AZERO(&naddr, AF_INET6)) { PF_ACPY((struct pf_addr *)&dst->sin6_addr, &naddr, AF_INET6); + } ifp = r->rpool.cur->kif ? r->rpool.cur->kif->pfik_ifp : NULL; } else { - if (!PF_AZERO(&s->rt_addr, AF_INET6)) + if (!PF_AZERO(&s->rt_addr, AF_INET6)) { PF_ACPY((struct pf_addr *)&dst->sin6_addr, &s->rt_addr, AF_INET6); + } ifp = s->rt_kif ? s->rt_kif->pfik_ifp : NULL; } - if (ifp == NULL) + if (ifp == NULL) { goto bad; + } if (oifp != ifp) { - if (pf_test6(PF_OUT, ifp, &m0, NULL, NULL) != PF_PASS) + if (pf_test6_mbuf(PF_OUT, ifp, &m0, NULL, NULL) != PF_PASS) { goto bad; - else if (m0 == NULL) + } else if (m0 == NULL) { goto done; - if (m0->m_len < (int)sizeof (struct ip6_hdr)) { + } + if (m0->m_len < (int)sizeof(struct ip6_hdr)) { DPFPRINTF(PF_DEBUG_URGENT, ("pf_route6: m0->m_len " "< sizeof (struct ip6_hdr)\n")); goto bad; } + pf_mtag = pf_get_mtag(m0); + /* + * send refragmented packets. + */ + if ((pf_mtag->pftag_flags & PF_TAG_REFRAGMENTED) != 0) { + pf_mtag->pftag_flags &= ~PF_TAG_REFRAGMENTED; + /* + * nd6_output() frees packet chain in both success and + * failure cases. + */ + error = nd6_output(ifp, ifp, m0, dst, NULL, NULL); + m0 = NULL; + if (error) { + DPFPRINTF(PF_DEBUG_URGENT, ("pf_route6:" + "dropped refragmented packet\n")); + } + goto done; + } ip6 = mtod(m0, struct ip6_hdr *); } @@ -7953,25 +9439,28 @@ pf_route6(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp, * If the packet is too large for the outgoing interface, * send back an icmp6 error. */ - if (IN6_IS_SCOPE_EMBED(&dst->sin6_addr)) + if (IN6_IS_SCOPE_EMBED(&dst->sin6_addr)) { dst->sin6_addr.s6_addr16[1] = htons(ifp->if_index); + } if ((unsigned)m0->m_pkthdr.len <= ifp->if_mtu) { error = nd6_output(ifp, ifp, m0, dst, NULL, NULL); } else { in6_ifstat_inc(ifp, ifs6_in_toobig); - if (r->rt != PF_DUPTO) + if (r->rt != PF_DUPTO) { icmp6_error(m0, ICMP6_PACKET_TOO_BIG, 0, ifp->if_mtu); - else + } else { goto bad; + } } done: - if (r->rt != PF_DUPTO) - *m = NULL; return; bad: - m_freem(m0); + if (m0) { + m_freem(m0); + m0 = NULL; + } goto done; } #endif /* INET6 */ @@ -7984,7 +9473,7 @@ bad: * returns 0 when the checksum is valid, otherwise returns 1. */ static int -pf_check_proto_cksum(struct mbuf *m, int off, int len, u_int8_t p, +pf_check_proto_cksum(pbuf_t *pbuf, int off, int len, u_int8_t p, sa_family_t af) { u_int16_t sum; @@ -7998,11 +9487,11 @@ pf_check_proto_cksum(struct mbuf *m, int off, int len, u_int8_t p, * is partially-computed (only 16-bit summation), do it in * software below. */ - if ((m->m_pkthdr.csum_flags & + if ((*pbuf->pb_csum_flags & (CSUM_DATA_VALID | CSUM_PSEUDO_HDR)) == (CSUM_DATA_VALID | CSUM_PSEUDO_HDR) && - (m->m_pkthdr.csum_data ^ 0xffff) == 0) { - return (0); + (*pbuf->pb_csum_data ^ 0xffff) == 0) { + return 0; } break; case IPPROTO_ICMP: @@ -8011,39 +9500,51 @@ pf_check_proto_cksum(struct mbuf *m, int off, int len, u_int8_t p, #endif /* INET6 */ break; default: - return (1); + return 1; + } + if (off < (int)sizeof(struct ip) || len < (int)sizeof(struct udphdr)) { + return 1; + } + if (pbuf->pb_packet_len < (unsigned)(off + len)) { + return 1; } - if (off < (int)sizeof (struct ip) || len < (int)sizeof (struct udphdr)) - return (1); - if (m->m_pkthdr.len < off + len) - return (1); switch (af) { #if INET case AF_INET: if (p == IPPROTO_ICMP) { - if (m->m_len < off) - return (1); +#if 0 + if (m->m_len < off) { + return 1; + } m->m_data += off; m->m_len -= off; sum = in_cksum(m, len); m->m_data -= off; m->m_len += off; +#else + if (pbuf->pb_contig_len < (unsigned)off) { + return 1; + } + sum = pbuf_inet_cksum(pbuf, 0, off, len); +#endif } else { - if (m->m_len < (int)sizeof (struct ip)) - return (1); - sum = inet_cksum(m, p, off, len); + if (pbuf->pb_contig_len < (int)sizeof(struct ip)) { + return 1; + } + sum = pbuf_inet_cksum(pbuf, p, off, len); } break; #endif /* INET */ #if INET6 case AF_INET6: - if (m->m_len < (int)sizeof (struct ip6_hdr)) - return (1); - sum = inet6_cksum(m, p, off, len); + if (pbuf->pb_contig_len < (int)sizeof(struct ip6_hdr)) { + return 1; + } + sum = pbuf_inet6_cksum(pbuf, p, off, len); break; #endif /* INET6 */ default: - return (1); + return 1; } if (sum) { switch (p) { @@ -8062,98 +9563,122 @@ pf_check_proto_cksum(struct mbuf *m, int off, int len, u_int8_t p, break; #endif /* INET6 */ } - return (1); + return 1; } - return (0); + return 0; } #if INET -#define PF_APPLE_UPDATE_PDESC_IPv4() \ - do { \ - if (m && pd.mp && m != pd.mp) { \ - m = pd.mp; \ - h = mtod(m, struct ip *); \ - pd.pf_mtag = pf_get_mtag(m); \ - } \ +#define PF_APPLE_UPDATE_PDESC_IPv4() \ + do { \ + if (pbuf && pd.mp && pbuf != pd.mp) { \ + pbuf = pd.mp; \ + h = pbuf->pb_data; \ + pd.pf_mtag = pf_get_mtag_pbuf(pbuf); \ + } \ } while (0) int -pf_test(int dir, struct ifnet *ifp, struct mbuf **m0, +pf_test_mbuf(int dir, struct ifnet *ifp, struct mbuf **m0, + struct ether_header *eh, struct ip_fw_args *fwa) +{ + pbuf_t pbuf_store, *pbuf; + int rv; + + pbuf_init_mbuf(&pbuf_store, *m0, (*m0)->m_pkthdr.rcvif); + pbuf = &pbuf_store; + + rv = pf_test(dir, ifp, &pbuf, eh, fwa); + + if (pbuf_is_valid(pbuf)) { + *m0 = pbuf->pb_mbuf; + pbuf->pb_mbuf = NULL; + pbuf_destroy(pbuf); + } else { + *m0 = NULL; + } + + return rv; +} + +int +pf_test(int dir, struct ifnet *ifp, pbuf_t **pbufp, struct ether_header *eh, struct ip_fw_args *fwa) { #if !DUMMYNET #pragma unused(fwa) #endif - struct pfi_kif *kif; - u_short action = PF_PASS, reason = 0, log = 0; - struct mbuf *m = *m0; - struct ip *h = 0; - struct pf_rule *a = NULL, *r = &pf_default_rule, *tr, *nr; - struct pf_state *s = NULL; - struct pf_state_key *sk = NULL; - struct pf_ruleset *ruleset = NULL; - struct pf_pdesc pd; - int off, dirndx, pqid = 0; + struct pfi_kif *kif; + u_short action = PF_PASS, reason = 0, log = 0; + pbuf_t *pbuf = *pbufp; + struct ip *h = 0; + struct pf_rule *a = NULL, *r = &pf_default_rule, *tr, *nr; + struct pf_state *s = NULL; + struct pf_state_key *sk = NULL; + struct pf_ruleset *ruleset = NULL; + struct pf_pdesc pd; + int off, dirndx, pqid = 0; + + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); - - if (!pf_status.running) - return (PF_PASS); + if (!pf_status.running) { + return PF_PASS; + } - memset(&pd, 0, sizeof (pd)); + memset(&pd, 0, sizeof(pd)); - if ((pd.pf_mtag = pf_get_mtag(m)) == NULL) { + if ((pd.pf_mtag = pf_get_mtag_pbuf(pbuf)) == NULL) { DPFPRINTF(PF_DEBUG_URGENT, - ("pf_test: pf_get_mtag returned NULL\n")); - return (PF_DROP); + ("pf_test: pf_get_mtag_pbuf returned NULL\n")); + return PF_DROP; } - if (pd.pf_mtag->pftag_flags & PF_TAG_GENERATED) - return (PF_PASS); + if (pd.pf_mtag->pftag_flags & PF_TAG_GENERATED) { + return PF_PASS; + } kif = (struct pfi_kif *)ifp->if_pf_kif; if (kif == NULL) { DPFPRINTF(PF_DEBUG_URGENT, ("pf_test: kif == NULL, if_name %s\n", ifp->if_name)); - return (PF_DROP); + return PF_DROP; + } + if (kif->pfik_flags & PFI_IFLAG_SKIP) { + return PF_PASS; } - if (kif->pfik_flags & PFI_IFLAG_SKIP) - return (PF_PASS); - VERIFY(m->m_flags & M_PKTHDR); + if (pbuf->pb_packet_len < (int)sizeof(*h)) { + REASON_SET(&reason, PFRES_SHORT); + return PF_DROP; + } /* initialize enough of pd for the done label */ - h = mtod(m, struct ip *); - pd.mp = m; + h = pbuf->pb_data; + pd.mp = pbuf; pd.lmw = 0; - pd.pf_mtag = pf_get_mtag(m); + pd.pf_mtag = pf_get_mtag_pbuf(pbuf); pd.src = (struct pf_addr *)&h->ip_src; pd.dst = (struct pf_addr *)&h->ip_dst; - PF_ACPY(&pd.baddr, dir == PF_OUT ? pd.src : pd.dst, AF_INET); + PF_ACPY(&pd.baddr, pd.src, AF_INET); + PF_ACPY(&pd.bdaddr, pd.dst, AF_INET); pd.ip_sum = &h->ip_sum; pd.proto = h->ip_p; pd.proto_variant = 0; pd.af = AF_INET; pd.tos = h->ip_tos; + pd.ttl = h->ip_ttl; pd.tot_len = ntohs(h->ip_len); pd.eh = eh; - if (m->m_pkthdr.len < (int)sizeof (*h)) { - action = PF_DROP; - REASON_SET(&reason, PFRES_SHORT); - log = 1; - goto done; - } - #if DUMMYNET - if (fwa != NULL && fwa->fwa_pf_rule != NULL) + if (fwa != NULL && fwa->fwa_pf_rule != NULL) { goto nonormalize; + } #endif /* DUMMYNET */ /* We do IP header normalization and packet reassembly here */ - action = pf_normalize_ip(m0, dir, kif, &reason, &pd); - pd.mp = m = *m0; + action = pf_normalize_ip(pbuf, dir, kif, &reason, &pd); if (action != PF_PASS || pd.lmw < 0) { action = PF_DROP; goto done; @@ -8162,11 +9687,11 @@ pf_test(int dir, struct ifnet *ifp, struct mbuf **m0, #if DUMMYNET nonormalize: #endif /* DUMMYNET */ - m = *m0; /* pf_normalize messes with m0 */ - h = mtod(m, struct ip *); + /* pf_normalize can mess with pb_data */ + h = pbuf->pb_data; off = h->ip_hl << 2; - if (off < (int)sizeof (*h)) { + if (off < (int)sizeof(*h)) { action = PF_DROP; REASON_SET(&reason, PFRES_SHORT); log = 1; @@ -8175,23 +9700,25 @@ nonormalize: pd.src = (struct pf_addr *)&h->ip_src; pd.dst = (struct pf_addr *)&h->ip_dst; - PF_ACPY(&pd.baddr, dir == PF_OUT ? pd.src : pd.dst, AF_INET); + PF_ACPY(&pd.baddr, pd.src, AF_INET); + PF_ACPY(&pd.bdaddr, pd.dst, AF_INET); pd.ip_sum = &h->ip_sum; pd.proto = h->ip_p; pd.proto_variant = 0; - pd.mp = m; + pd.mp = pbuf; pd.lmw = 0; - pd.pf_mtag = pf_get_mtag(m); + pd.pf_mtag = pf_get_mtag_pbuf(pbuf); pd.af = AF_INET; pd.tos = h->ip_tos; - pd.sc = MBUF_SCIDX(mbuf_get_service_class(m)); + pd.ttl = h->ip_ttl; + pd.sc = MBUF_SCIDX(pbuf_get_service_class(pbuf)); pd.tot_len = ntohs(h->ip_len); pd.eh = eh; - if (m->m_pkthdr.pkt_flags & PKTF_FLOW_ID) { - pd.flowsrc = m->m_pkthdr.pkt_flowsrc; - pd.flowhash = m->m_pkthdr.pkt_flowid; - pd.pktflags = (m->m_pkthdr.pkt_flags & PKTF_FLOW_MASK); + if (*pbuf->pb_flags & PKTF_FLOW_ID) { + pd.flowsrc = *pbuf->pb_flowsrc; + pd.flowhash = *pbuf->pb_flowid; + pd.pktflags = *pbuf->pb_flags & PKTF_FLOW_MASK; } /* handle fragments that didn't get reassembled by normalization */ @@ -8199,48 +9726,54 @@ nonormalize: pd.flags |= PFDESC_IP_FRAG; #if DUMMYNET /* Traffic goes through dummynet first */ - action = pf_test_dummynet(&r, dir, kif, &m, &pd, fwa); - if (action == PF_DROP || m == NULL) { - *m0 = NULL; - return (action); + action = pf_test_dummynet(&r, dir, kif, &pbuf, &pd, fwa); + if (action == PF_DROP || pbuf == NULL) { + *pbufp = NULL; + return action; } #endif /* DUMMYNET */ - action = pf_test_fragment(&r, dir, kif, m, h, + action = pf_test_fragment(&r, dir, kif, pbuf, h, &pd, &a, &ruleset); goto done; } switch (h->ip_p) { - case IPPROTO_TCP: { - struct tcphdr th; + struct tcphdr th; pd.hdr.tcp = &th; - if (!pf_pull_hdr(m, off, &th, sizeof (th), + if (!pf_pull_hdr(pbuf, off, &th, sizeof(th), &action, &reason, AF_INET)) { log = action != PF_PASS; goto done; } pd.p_len = pd.tot_len - off - (th.th_off << 2); - if ((th.th_flags & TH_ACK) && pd.p_len == 0) + if ((th.th_flags & TH_ACK) && pd.p_len == 0) { pqid = 1; + } #if DUMMYNET /* Traffic goes through dummynet first */ - action = pf_test_dummynet(&r, dir, kif, &m, &pd, fwa); - if (action == PF_DROP || m == NULL) { - *m0 = NULL; - return (action); + action = pf_test_dummynet(&r, dir, kif, &pbuf, &pd, fwa); + if (action == PF_DROP || pbuf == NULL) { + *pbufp = NULL; + return action; } #endif /* DUMMYNET */ - action = pf_normalize_tcp(dir, kif, m, 0, off, h, &pd); - if (pd.lmw < 0) + action = pf_normalize_tcp(dir, kif, pbuf, 0, off, h, &pd); + if (pd.lmw < 0) { goto done; + } PF_APPLE_UPDATE_PDESC_IPv4(); - if (action == PF_DROP) + if (action == PF_DROP) { goto done; - action = pf_test_state_tcp(&s, dir, kif, m, off, h, &pd, + } + action = pf_test_state_tcp(&s, dir, kif, pbuf, off, h, &pd, &reason); - if (pd.lmw < 0) + if (action == PF_NAT64) { + goto done; + } + if (pd.lmw < 0) { goto done; + } PF_APPLE_UPDATE_PDESC_IPv4(); if (action == PF_PASS) { #if NPFSYNC @@ -8249,40 +9782,45 @@ nonormalize: r = s->rule.ptr; a = s->anchor.ptr; log = s->log; - } else if (s == NULL) + } else if (s == NULL) { action = pf_test_rule(&r, &s, dir, kif, - m, off, h, &pd, &a, &ruleset, NULL); + pbuf, off, h, &pd, &a, &ruleset, NULL); + } break; } case IPPROTO_UDP: { - struct udphdr uh; + struct udphdr uh; pd.hdr.udp = &uh; - if (!pf_pull_hdr(m, off, &uh, sizeof (uh), + if (!pf_pull_hdr(pbuf, off, &uh, sizeof(uh), &action, &reason, AF_INET)) { log = action != PF_PASS; goto done; } if (uh.uh_dport == 0 || - ntohs(uh.uh_ulen) > m->m_pkthdr.len - off || - ntohs(uh.uh_ulen) < sizeof (struct udphdr)) { + ntohs(uh.uh_ulen) > pbuf->pb_packet_len - off || + ntohs(uh.uh_ulen) < sizeof(struct udphdr)) { action = PF_DROP; REASON_SET(&reason, PFRES_SHORT); goto done; } #if DUMMYNET /* Traffic goes through dummynet first */ - action = pf_test_dummynet(&r, dir, kif, &m, &pd, fwa); - if (action == PF_DROP || m == NULL) { - *m0 = NULL; - return (action); + action = pf_test_dummynet(&r, dir, kif, &pbuf, &pd, fwa); + if (action == PF_DROP || pbuf == NULL) { + *pbufp = NULL; + return action; } #endif /* DUMMYNET */ - action = pf_test_state_udp(&s, dir, kif, m, off, h, &pd, + action = pf_test_state_udp(&s, dir, kif, pbuf, off, h, &pd, &reason); - if (pd.lmw < 0) + if (action == PF_NAT64) { + goto done; + } + if (pd.lmw < 0) { goto done; + } PF_APPLE_UPDATE_PDESC_IPv4(); if (action == PF_PASS) { #if NPFSYNC @@ -8291,33 +9829,38 @@ nonormalize: r = s->rule.ptr; a = s->anchor.ptr; log = s->log; - } else if (s == NULL) + } else if (s == NULL) { action = pf_test_rule(&r, &s, dir, kif, - m, off, h, &pd, &a, &ruleset, NULL); + pbuf, off, h, &pd, &a, &ruleset, NULL); + } break; } case IPPROTO_ICMP: { - struct icmp ih; + struct icmp ih; pd.hdr.icmp = &ih; - if (!pf_pull_hdr(m, off, &ih, ICMP_MINLEN, + if (!pf_pull_hdr(pbuf, off, &ih, ICMP_MINLEN, &action, &reason, AF_INET)) { log = action != PF_PASS; goto done; } #if DUMMYNET /* Traffic goes through dummynet first */ - action = pf_test_dummynet(&r, dir, kif, &m, &pd, fwa); - if (action == PF_DROP || m == NULL) { - *m0 = NULL; - return (action); + action = pf_test_dummynet(&r, dir, kif, &pbuf, &pd, fwa); + if (action == PF_DROP || pbuf == NULL) { + *pbufp = NULL; + return action; } #endif /* DUMMYNET */ - action = pf_test_state_icmp(&s, dir, kif, m, off, h, &pd, + action = pf_test_state_icmp(&s, dir, kif, pbuf, off, h, &pd, &reason); - if (pd.lmw < 0) + if (action == PF_NAT64) { goto done; + } + if (pd.lmw < 0) { + goto done; + } PF_APPLE_UPDATE_PDESC_IPv4(); if (action == PF_PASS) { #if NPFSYNC @@ -8326,32 +9869,34 @@ nonormalize: r = s->rule.ptr; a = s->anchor.ptr; log = s->log; - } else if (s == NULL) + } else if (s == NULL) { action = pf_test_rule(&r, &s, dir, kif, - m, off, h, &pd, &a, &ruleset, NULL); + pbuf, off, h, &pd, &a, &ruleset, NULL); + } break; } case IPPROTO_ESP: { - struct pf_esp_hdr esp; + struct pf_esp_hdr esp; pd.hdr.esp = &esp; - if (!pf_pull_hdr(m, off, &esp, sizeof (esp), &action, &reason, + if (!pf_pull_hdr(pbuf, off, &esp, sizeof(esp), &action, &reason, AF_INET)) { log = action != PF_PASS; goto done; } #if DUMMYNET /* Traffic goes through dummynet first */ - action = pf_test_dummynet(&r, dir, kif, &m, &pd, fwa); - if (action == PF_DROP || m == NULL) { - *m0 = NULL; - return (action); + action = pf_test_dummynet(&r, dir, kif, &pbuf, &pd, fwa); + if (action == PF_DROP || pbuf == NULL) { + *pbufp = NULL; + return action; } #endif /* DUMMYNET */ action = pf_test_state_esp(&s, dir, kif, off, &pd); - if (pd.lmw < 0) + if (pd.lmw < 0) { goto done; + } PF_APPLE_UPDATE_PDESC_IPv4(); if (action == PF_PASS) { #if NPFSYNC @@ -8360,39 +9905,42 @@ nonormalize: r = s->rule.ptr; a = s->anchor.ptr; log = s->log; - } else if (s == NULL) + } else if (s == NULL) { action = pf_test_rule(&r, &s, dir, kif, - m, off, h, &pd, &a, &ruleset, NULL); + pbuf, off, h, &pd, &a, &ruleset, NULL); + } break; } case IPPROTO_GRE: { - struct pf_grev1_hdr grev1; + struct pf_grev1_hdr grev1; pd.hdr.grev1 = &grev1; - if (!pf_pull_hdr(m, off, &grev1, sizeof (grev1), &action, + if (!pf_pull_hdr(pbuf, off, &grev1, sizeof(grev1), &action, &reason, AF_INET)) { log = (action != PF_PASS); goto done; } #if DUMMYNET /* Traffic goes through dummynet first */ - action = pf_test_dummynet(&r, dir, kif, &m, &pd, fwa); - if (action == PF_DROP || m == NULL) { - *m0 = NULL; - return (action); + action = pf_test_dummynet(&r, dir, kif, &pbuf, &pd, fwa); + if (action == PF_DROP || pbuf == NULL) { + *pbufp = NULL; + return action; } #endif /* DUMMYNET */ if ((ntohs(grev1.flags) & PF_GRE_FLAG_VERSION_MASK) == 1 && ntohs(grev1.protocol_type) == PF_GRE_PPP_ETHERTYPE) { if (ntohs(grev1.payload_length) > - m->m_pkthdr.len - off) { + pbuf->pb_packet_len - off) { action = PF_DROP; REASON_SET(&reason, PFRES_SHORT); goto done; } pd.proto_variant = PF_GRE_PPTP_VARIANT; action = pf_test_state_grev1(&s, dir, kif, off, &pd); - if (pd.lmw < 0) goto done; + if (pd.lmw < 0) { + goto done; + } PF_APPLE_UPDATE_PDESC_IPv4(); if (action == PF_PASS) { #if NPFSYNC @@ -8403,10 +9951,11 @@ nonormalize: log = s->log; break; } else if (s == NULL) { - action = pf_test_rule(&r, &s, dir, kif, m, off, - h, &pd, &a, &ruleset, NULL); - if (action == PF_PASS) + action = pf_test_rule(&r, &s, dir, kif, pbuf, + off, h, &pd, &a, &ruleset, NULL); + if (action == PF_PASS) { break; + } } } @@ -8416,15 +9965,16 @@ nonormalize: default: #if DUMMYNET /* Traffic goes through dummynet first */ - action = pf_test_dummynet(&r, dir, kif, &m, &pd, fwa); - if (action == PF_DROP || m == NULL) { - *m0 = NULL; - return (action); + action = pf_test_dummynet(&r, dir, kif, &pbuf, &pd, fwa); + if (action == PF_DROP || pbuf == NULL) { + *pbufp = NULL; + return action; } #endif /* DUMMYNET */ action = pf_test_state_other(&s, dir, kif, &pd); - if (pd.lmw < 0) + if (pd.lmw < 0) { goto done; + } PF_APPLE_UPDATE_PDESC_IPv4(); if (action == PF_PASS) { #if NPFSYNC @@ -8433,72 +9983,77 @@ nonormalize: r = s->rule.ptr; a = s->anchor.ptr; log = s->log; - } else if (s == NULL) - action = pf_test_rule(&r, &s, dir, kif, m, off, h, + } else if (s == NULL) { + action = pf_test_rule(&r, &s, dir, kif, pbuf, off, h, &pd, &a, &ruleset, NULL); + } break; } done: - *m0 = pd.mp; - PF_APPLE_UPDATE_PDESC_IPv4(); - - if (action == PF_PASS && h->ip_hl > 5 && - !((s && s->allow_opts) || r->allow_opts)) { - action = PF_DROP; - REASON_SET(&reason, PFRES_IPOPTIONS); - log = 1; - DPFPRINTF(PF_DEBUG_MISC, - ("pf: dropping packet with ip options [hlen=%u]\n", - (unsigned int) h->ip_hl)); + if (action == PF_NAT64) { + *pbufp = NULL; + return action; } - if ((s && s->tag) || PF_RTABLEID_IS_VALID(r->rtableid) || - (pd.pktflags & PKTF_FLOW_ID)) - (void) pf_tag_packet(m, pd.pf_mtag, s ? s->tag : 0, - r->rtableid, &pd); + *pbufp = pd.mp; + PF_APPLE_UPDATE_PDESC_IPv4(); + + if (action != PF_DROP) { + if (action == PF_PASS && h->ip_hl > 5 && + !((s && s->allow_opts) || r->allow_opts)) { + action = PF_DROP; + REASON_SET(&reason, PFRES_IPOPTIONS); + log = 1; + DPFPRINTF(PF_DEBUG_MISC, + ("pf: dropping packet with ip options [hlen=%u]\n", + (unsigned int) h->ip_hl)); + } - if (action == PF_PASS) { -#if PF_ALTQ - if (altq_allowed && r->qid) { - if (pqid || (pd.tos & IPTOS_LOWDELAY)) - pd.pf_mtag->pftag_qid = r->pqid; - else - pd.pf_mtag->pftag_qid = r->qid; + if ((s && s->tag) || PF_RTABLEID_IS_VALID(r->rtableid) || + (pd.pktflags & PKTF_FLOW_ID)) { + (void) pf_tag_packet(pbuf, pd.pf_mtag, s ? s->tag : 0, + r->rtableid, &pd); } -#endif /* PF_ALTQ */ + + if (action == PF_PASS) { #if PF_ECN - /* add hints for ecn */ - pd.pf_mtag->pftag_hdr = h; - /* record address family */ - pd.pf_mtag->pftag_flags &= ~PF_TAG_HDR_INET6; - pd.pf_mtag->pftag_flags |= PF_TAG_HDR_INET; + /* add hints for ecn */ + pd.pf_mtag->pftag_hdr = h; + /* record address family */ + pd.pf_mtag->pftag_flags &= ~PF_TAG_HDR_INET6; + pd.pf_mtag->pftag_flags |= PF_TAG_HDR_INET; #endif /* PF_ECN */ - /* record protocol */ - m->m_pkthdr.pkt_proto = pd.proto; - } + /* record protocol */ + *pbuf->pb_proto = pd.proto; - /* - * connections redirected to loopback should not match sockets - * bound specifically to loopback due to security implications, - * see tcp_input() and in_pcblookup_listen(). - */ - if (dir == PF_IN && action == PF_PASS && (pd.proto == IPPROTO_TCP || - pd.proto == IPPROTO_UDP) && s != NULL && s->nat_rule.ptr != NULL && - (s->nat_rule.ptr->action == PF_RDR || - s->nat_rule.ptr->action == PF_BINAT) && - (ntohl(pd.dst->v4.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET) - pd.pf_mtag->pftag_flags |= PF_TAG_TRANSLATE_LOCALHOST; + /* + * connections redirected to loopback should not match sockets + * bound specifically to loopback due to security implications, + * see tcp_input() and in_pcblookup_listen(). + */ + if (dir == PF_IN && (pd.proto == IPPROTO_TCP || + pd.proto == IPPROTO_UDP) && s != NULL && + s->nat_rule.ptr != NULL && + (s->nat_rule.ptr->action == PF_RDR || + s->nat_rule.ptr->action == PF_BINAT) && + (ntohl(pd.dst->v4addr.s_addr) >> IN_CLASSA_NSHIFT) + == IN_LOOPBACKNET) { + pd.pf_mtag->pftag_flags |= PF_TAG_TRANSLATE_LOCALHOST; + } + } + } if (log) { struct pf_rule *lr; if (s != NULL && s->nat_rule.ptr != NULL && - s->nat_rule.ptr->log & PF_LOG_ALL) + s->nat_rule.ptr->log & PF_LOG_ALL) { lr = s->nat_rule.ptr; - else + } else { lr = r; - PFLOG_PACKET(kif, h, m, AF_INET, dir, reason, lr, a, ruleset, + } + PFLOG_PACKET(kif, h, pbuf, AF_INET, dir, reason, lr, a, ruleset, &pd); } @@ -8544,157 +10099,189 @@ done: tr = nr; x = (sk == NULL || sk->direction == dir) ? &pd.baddr : &pd.naddr; - } else + } else { x = (sk == NULL || sk->direction == dir) ? &pd.naddr : &pd.baddr; + } if (x == &pd.baddr || s == NULL) { /* we need to change the address */ - if (dir == PF_OUT) + if (dir == PF_OUT) { pd.src = x; - else + } else { pd.dst = x; + } } } - if (tr->src.addr.type == PF_ADDR_TABLE) + if (tr->src.addr.type == PF_ADDR_TABLE) { pfr_update_stats(tr->src.addr.p.tbl, (sk == NULL || sk->direction == dir) ? pd.src : pd.dst, pd.af, pd.tot_len, dir == PF_OUT, r->action == PF_PASS, tr->src.neg); - if (tr->dst.addr.type == PF_ADDR_TABLE) + } + if (tr->dst.addr.type == PF_ADDR_TABLE) { pfr_update_stats(tr->dst.addr.p.tbl, (sk == NULL || sk->direction == dir) ? pd.dst : pd.src, pd.af, pd.tot_len, dir == PF_OUT, r->action == PF_PASS, tr->dst.neg); + } } - VERIFY(m == NULL || pd.mp == NULL || pd.mp == m); + VERIFY(pbuf == NULL || pd.mp == NULL || pd.mp == pbuf); - if (*m0) { + if (*pbufp) { if (pd.lmw < 0) { REASON_SET(&reason, PFRES_MEMORY); action = PF_DROP; } if (action == PF_DROP) { - m_freem(*m0); - *m0 = NULL; - return (PF_DROP); + pbuf_destroy(*pbufp); + *pbufp = NULL; + return PF_DROP; } - *m0 = m; + *pbufp = pbuf; } if (action == PF_SYNPROXY_DROP) { - m_freem(*m0); - *m0 = NULL; + pbuf_destroy(*pbufp); + *pbufp = NULL; action = PF_PASS; - } else if (r->rt) - /* pf_route can free the mbuf causing *m0 to become NULL */ - pf_route(m0, r, dir, kif->pfik_ifp, s, &pd); + } else if (r->rt) { + /* pf_route can free the pbuf causing *pbufp to become NULL */ + pf_route(pbufp, r, dir, kif->pfik_ifp, s, &pd); + } - return (action); + return action; } #endif /* INET */ #if INET6 -#define PF_APPLE_UPDATE_PDESC_IPv6() \ - do { \ - if (m && pd.mp && m != pd.mp) { \ - if (n == m) \ - n = pd.mp; \ - m = pd.mp; \ - h = mtod(m, struct ip6_hdr *); \ - } \ +#define PF_APPLE_UPDATE_PDESC_IPv6() \ + do { \ + if (pbuf && pd.mp && pbuf != pd.mp) { \ + pbuf = pd.mp; \ + } \ + h = pbuf->pb_data; \ } while (0) int -pf_test6(int dir, struct ifnet *ifp, struct mbuf **m0, +pf_test6_mbuf(int dir, struct ifnet *ifp, struct mbuf **m0, + struct ether_header *eh, struct ip_fw_args *fwa) +{ + pbuf_t pbuf_store, *pbuf; + int rv; + + pbuf_init_mbuf(&pbuf_store, *m0, (*m0)->m_pkthdr.rcvif); + pbuf = &pbuf_store; + + rv = pf_test6(dir, ifp, &pbuf, eh, fwa); + + if (pbuf_is_valid(pbuf)) { + *m0 = pbuf->pb_mbuf; + pbuf->pb_mbuf = NULL; + pbuf_destroy(pbuf); + } else { + *m0 = NULL; + } + + return rv; +} + +int +pf_test6(int dir, struct ifnet *ifp, pbuf_t **pbufp, struct ether_header *eh, struct ip_fw_args *fwa) { #if !DUMMYNET #pragma unused(fwa) #endif - struct pfi_kif *kif; - u_short action = PF_PASS, reason = 0, log = 0; - struct mbuf *m = *m0, *n = NULL; - struct ip6_hdr *h; - struct pf_rule *a = NULL, *r = &pf_default_rule, *tr, *nr; - struct pf_state *s = NULL; - struct pf_state_key *sk = NULL; - struct pf_ruleset *ruleset = NULL; - struct pf_pdesc pd; - int off, terminal = 0, dirndx, rh_cnt = 0; - u_int8_t nxt; - - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); - - if (!pf_status.running) - return (PF_PASS); - - memset(&pd, 0, sizeof (pd)); - - if ((pd.pf_mtag = pf_get_mtag(m)) == NULL) { + struct pfi_kif *kif; + u_short action = PF_PASS, reason = 0, log = 0; + pbuf_t *pbuf = *pbufp; + struct ip6_hdr *h; + struct pf_rule *a = NULL, *r = &pf_default_rule, *tr, *nr; + struct pf_state *s = NULL; + struct pf_state_key *sk = NULL; + struct pf_ruleset *ruleset = NULL; + struct pf_pdesc pd; + int off, terminal = 0, dirndx, rh_cnt = 0; + u_int8_t nxt; + boolean_t fwd = FALSE; + + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); + + ASSERT(ifp != NULL); + if ((dir == PF_OUT) && (pbuf->pb_ifp) && (ifp != pbuf->pb_ifp)) { + fwd = TRUE; + } + + if (!pf_status.running) { + return PF_PASS; + } + + memset(&pd, 0, sizeof(pd)); + + if ((pd.pf_mtag = pf_get_mtag_pbuf(pbuf)) == NULL) { DPFPRINTF(PF_DEBUG_URGENT, - ("pf_test6: pf_get_mtag returned NULL\n")); - return (PF_DROP); + ("pf_test6: pf_get_mtag_pbuf returned NULL\n")); + return PF_DROP; } - if (pd.pf_mtag->pftag_flags & PF_TAG_GENERATED) - return (PF_PASS); + if (pd.pf_mtag->pftag_flags & PF_TAG_GENERATED) { + return PF_PASS; + } kif = (struct pfi_kif *)ifp->if_pf_kif; if (kif == NULL) { DPFPRINTF(PF_DEBUG_URGENT, ("pf_test6: kif == NULL, if_name %s\n", ifp->if_name)); - return (PF_DROP); + return PF_DROP; + } + if (kif->pfik_flags & PFI_IFLAG_SKIP) { + return PF_PASS; } - if (kif->pfik_flags & PFI_IFLAG_SKIP) - return (PF_PASS); - - VERIFY(m->m_flags & M_PKTHDR); - h = mtod(m, struct ip6_hdr *); + if (pbuf->pb_packet_len < (int)sizeof(*h)) { + REASON_SET(&reason, PFRES_SHORT); + return PF_DROP; + } + h = pbuf->pb_data; nxt = h->ip6_nxt; - off = ((caddr_t)h - m->m_data) + sizeof(struct ip6_hdr); - pd.mp = m; + off = ((caddr_t)h - (caddr_t)pbuf->pb_data) + sizeof(struct ip6_hdr); + pd.mp = pbuf; pd.lmw = 0; - pd.pf_mtag = pf_get_mtag(m); - pd.src = (struct pf_addr *)&h->ip6_src; - pd.dst = (struct pf_addr *)&h->ip6_dst; - PF_ACPY(&pd.baddr, dir == PF_OUT ? pd.src : pd.dst, AF_INET6); + pd.pf_mtag = pf_get_mtag_pbuf(pbuf); + pd.src = (struct pf_addr *)(uintptr_t)&h->ip6_src; + pd.dst = (struct pf_addr *)(uintptr_t)&h->ip6_dst; + PF_ACPY(&pd.baddr, pd.src, AF_INET6); + PF_ACPY(&pd.bdaddr, pd.dst, AF_INET6); pd.ip_sum = NULL; pd.af = AF_INET6; pd.proto = nxt; pd.proto_variant = 0; pd.tos = 0; - pd.sc = MBUF_SCIDX(mbuf_get_service_class(m)); + pd.ttl = h->ip6_hlim; + pd.sc = MBUF_SCIDX(pbuf_get_service_class(pbuf)); pd.tot_len = ntohs(h->ip6_plen) + sizeof(struct ip6_hdr); pd.eh = eh; - if (m->m_pkthdr.pkt_flags & PKTF_FLOW_ID) { - pd.flowsrc = m->m_pkthdr.pkt_flowsrc; - pd.flowhash = m->m_pkthdr.pkt_flowid; - pd.pktflags = (m->m_pkthdr.pkt_flags & PKTF_FLOW_MASK); - } - - if (m->m_pkthdr.len < (int)sizeof (*h)) { - action = PF_DROP; - REASON_SET(&reason, PFRES_SHORT); - log = 1; - goto done; + if (*pbuf->pb_flags & PKTF_FLOW_ID) { + pd.flowsrc = *pbuf->pb_flowsrc; + pd.flowhash = *pbuf->pb_flowid; + pd.pktflags = (*pbuf->pb_flags & PKTF_FLOW_MASK); } #if DUMMYNET - if (fwa != NULL && fwa->fwa_pf_rule != NULL) + if (fwa != NULL && fwa->fwa_pf_rule != NULL) { goto nonormalize; + } #endif /* DUMMYNET */ /* We do IP header normalization and packet reassembly here */ - action = pf_normalize_ip6(m0, dir, kif, &reason, &pd); - pd.mp = m = *m0; + action = pf_normalize_ip6(pbuf, dir, kif, &reason, &pd); if (action != PF_PASS || pd.lmw < 0) { action = PF_DROP; goto done; @@ -8703,7 +10290,7 @@ pf_test6(int dir, struct ifnet *ifp, struct mbuf **m0, #if DUMMYNET nonormalize: #endif /* DUMMYNET */ - h = mtod(m, struct ip6_hdr *); + h = pbuf->pb_data; #if 1 /* @@ -8712,34 +10299,35 @@ nonormalize: */ if (htons(h->ip6_plen) == 0) { action = PF_DROP; - REASON_SET(&reason, PFRES_NORM); /*XXX*/ + REASON_SET(&reason, PFRES_NORM); /*XXX*/ goto done; } #endif - - pd.src = (struct pf_addr *)&h->ip6_src; - pd.dst = (struct pf_addr *)&h->ip6_dst; - PF_ACPY(&pd.baddr, dir == PF_OUT ? pd.src : pd.dst, AF_INET6); + pd.src = (struct pf_addr *)(uintptr_t)&h->ip6_src; + pd.dst = (struct pf_addr *)(uintptr_t)&h->ip6_dst; + PF_ACPY(&pd.baddr, pd.src, AF_INET6); + PF_ACPY(&pd.bdaddr, pd.dst, AF_INET6); pd.ip_sum = NULL; pd.af = AF_INET6; pd.tos = 0; - pd.tot_len = ntohs(h->ip6_plen) + sizeof (struct ip6_hdr); + pd.ttl = h->ip6_hlim; + pd.tot_len = ntohs(h->ip6_plen) + sizeof(struct ip6_hdr); pd.eh = eh; - off = ((caddr_t)h - m->m_data) + sizeof (struct ip6_hdr); + off = ((caddr_t)h - (caddr_t)pbuf->pb_data) + sizeof(struct ip6_hdr); pd.proto = h->ip6_nxt; pd.proto_variant = 0; - pd.mp = m; + pd.mp = pbuf; pd.lmw = 0; - pd.pf_mtag = pf_get_mtag(m); + pd.pf_mtag = pf_get_mtag_pbuf(pbuf); do { - switch (nxt) { + switch (pd.proto) { case IPPROTO_FRAGMENT: { struct ip6_frag ip6f; pd.flags |= PFDESC_IP_FRAG; - if (!pf_pull_hdr(m, off, &ip6f, sizeof ip6f, NULL, + if (!pf_pull_hdr(pbuf, off, &ip6f, sizeof ip6f, NULL, &reason, pd.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: IPv6 short fragment header\n")); @@ -8748,17 +10336,18 @@ nonormalize: log = 1; goto done; } - pd.proto = nxt = ip6f.ip6f_nxt; + pd.proto = ip6f.ip6f_nxt; #if DUMMYNET /* Traffic goes through dummynet first */ - action = pf_test_dummynet(&r, dir, kif, &m, &pd, fwa); - if (action == PF_DROP || m == NULL) { - *m0 = NULL; - return (action); + action = pf_test_dummynet(&r, dir, kif, &pbuf, &pd, + fwa); + if (action == PF_DROP || pbuf == NULL) { + *pbufp = NULL; + return action; } #endif /* DUMMYNET */ - action = pf_test_fragment(&r, dir, kif, m, h, &pd, &a, - &ruleset); + action = pf_test_fragment(&r, dir, kif, pbuf, h, &pd, + &a, &ruleset); if (action == PF_DROP) { REASON_SET(&reason, PFRES_FRAG); log = 1; @@ -8767,15 +10356,15 @@ nonormalize: } case IPPROTO_ROUTING: ++rh_cnt; - /* FALL THROUGH */ + /* FALL THROUGH */ case IPPROTO_AH: case IPPROTO_HOPOPTS: case IPPROTO_DSTOPTS: { /* get next header and header length */ - struct ip6_ext opt6; + struct ip6_ext opt6; - if (!pf_pull_hdr(m, off, &opt6, sizeof(opt6), + if (!pf_pull_hdr(pbuf, off, &opt6, sizeof(opt6), NULL, &reason, pd.af)) { DPFPRINTF(PF_DEBUG_MISC, ("pf: IPv6 short opt\n")); @@ -8783,11 +10372,12 @@ nonormalize: log = 1; goto done; } - if (pd.proto == IPPROTO_AH) + if (pd.proto == IPPROTO_AH) { off += (opt6.ip6e_len + 2) * 4; - else + } else { off += (opt6.ip6e_len + 1) * 8; - nxt = opt6.ip6e_nxt; + } + pd.proto = opt6.ip6e_nxt; /* goto the next header */ break; } @@ -8797,17 +10387,13 @@ nonormalize: } } while (!terminal); - /* if there's no routing header, use unmodified mbuf for checksumming */ - if (!n) - n = m; switch (pd.proto) { - case IPPROTO_TCP: { - struct tcphdr th; + struct tcphdr th; pd.hdr.tcp = &th; - if (!pf_pull_hdr(m, off, &th, sizeof (th), + if (!pf_pull_hdr(pbuf, off, &th, sizeof(th), &action, &reason, AF_INET6)) { log = action != PF_PASS; goto done; @@ -8815,22 +10401,28 @@ nonormalize: pd.p_len = pd.tot_len - off - (th.th_off << 2); #if DUMMYNET /* Traffic goes through dummynet first */ - action = pf_test_dummynet(&r, dir, kif, &m, &pd, fwa); - if (action == PF_DROP || m == NULL) { - *m0 = NULL; - return (action); + action = pf_test_dummynet(&r, dir, kif, &pbuf, &pd, fwa); + if (action == PF_DROP || pbuf == NULL) { + *pbufp = NULL; + return action; } #endif /* DUMMYNET */ - action = pf_normalize_tcp(dir, kif, m, 0, off, h, &pd); - if (pd.lmw < 0) + action = pf_normalize_tcp(dir, kif, pbuf, 0, off, h, &pd); + if (pd.lmw < 0) { goto done; + } PF_APPLE_UPDATE_PDESC_IPv6(); - if (action == PF_DROP) + if (action == PF_DROP) { goto done; - action = pf_test_state_tcp(&s, dir, kif, m, off, h, &pd, + } + action = pf_test_state_tcp(&s, dir, kif, pbuf, off, h, &pd, &reason); - if (pd.lmw < 0) + if (action == PF_NAT64) { goto done; + } + if (pd.lmw < 0) { + goto done; + } PF_APPLE_UPDATE_PDESC_IPv6(); if (action == PF_PASS) { #if NPFSYNC @@ -8839,40 +10431,45 @@ nonormalize: r = s->rule.ptr; a = s->anchor.ptr; log = s->log; - } else if (s == NULL) + } else if (s == NULL) { action = pf_test_rule(&r, &s, dir, kif, - m, off, h, &pd, &a, &ruleset, NULL); + pbuf, off, h, &pd, &a, &ruleset, NULL); + } break; } case IPPROTO_UDP: { - struct udphdr uh; + struct udphdr uh; pd.hdr.udp = &uh; - if (!pf_pull_hdr(m, off, &uh, sizeof (uh), + if (!pf_pull_hdr(pbuf, off, &uh, sizeof(uh), &action, &reason, AF_INET6)) { log = action != PF_PASS; goto done; } if (uh.uh_dport == 0 || - ntohs(uh.uh_ulen) > m->m_pkthdr.len - off || - ntohs(uh.uh_ulen) < sizeof (struct udphdr)) { + ntohs(uh.uh_ulen) > pbuf->pb_packet_len - off || + ntohs(uh.uh_ulen) < sizeof(struct udphdr)) { action = PF_DROP; REASON_SET(&reason, PFRES_SHORT); goto done; } #if DUMMYNET /* Traffic goes through dummynet first */ - action = pf_test_dummynet(&r, dir, kif, &m, &pd, fwa); - if (action == PF_DROP || m == NULL) { - *m0 = NULL; - return (action); + action = pf_test_dummynet(&r, dir, kif, &pbuf, &pd, fwa); + if (action == PF_DROP || pbuf == NULL) { + *pbufp = NULL; + return action; } #endif /* DUMMYNET */ - action = pf_test_state_udp(&s, dir, kif, m, off, h, &pd, + action = pf_test_state_udp(&s, dir, kif, pbuf, off, h, &pd, &reason); - if (pd.lmw < 0) + if (action == PF_NAT64) { + goto done; + } + if (pd.lmw < 0) { goto done; + } PF_APPLE_UPDATE_PDESC_IPv6(); if (action == PF_PASS) { #if NPFSYNC @@ -8881,33 +10478,38 @@ nonormalize: r = s->rule.ptr; a = s->anchor.ptr; log = s->log; - } else if (s == NULL) + } else if (s == NULL) { action = pf_test_rule(&r, &s, dir, kif, - m, off, h, &pd, &a, &ruleset, NULL); + pbuf, off, h, &pd, &a, &ruleset, NULL); + } break; } case IPPROTO_ICMPV6: { - struct icmp6_hdr ih; + struct icmp6_hdr ih; pd.hdr.icmp6 = &ih; - if (!pf_pull_hdr(m, off, &ih, sizeof (ih), + if (!pf_pull_hdr(pbuf, off, &ih, sizeof(ih), &action, &reason, AF_INET6)) { log = action != PF_PASS; goto done; } #if DUMMYNET /* Traffic goes through dummynet first */ - action = pf_test_dummynet(&r, dir, kif, &m, &pd, fwa); - if (action == PF_DROP || m == NULL) { - *m0 = NULL; - return (action); + action = pf_test_dummynet(&r, dir, kif, &pbuf, &pd, fwa); + if (action == PF_DROP || pbuf == NULL) { + *pbufp = NULL; + return action; } #endif /* DUMMYNET */ action = pf_test_state_icmp(&s, dir, kif, - m, off, h, &pd, &reason); - if (pd.lmw < 0) + pbuf, off, h, &pd, &reason); + if (action == PF_NAT64) { + goto done; + } + if (pd.lmw < 0) { goto done; + } PF_APPLE_UPDATE_PDESC_IPv6(); if (action == PF_PASS) { #if NPFSYNC @@ -8916,32 +10518,34 @@ nonormalize: r = s->rule.ptr; a = s->anchor.ptr; log = s->log; - } else if (s == NULL) + } else if (s == NULL) { action = pf_test_rule(&r, &s, dir, kif, - m, off, h, &pd, &a, &ruleset, NULL); + pbuf, off, h, &pd, &a, &ruleset, NULL); + } break; } case IPPROTO_ESP: { - struct pf_esp_hdr esp; + struct pf_esp_hdr esp; pd.hdr.esp = &esp; - if (!pf_pull_hdr(m, off, &esp, sizeof (esp), &action, &reason, - AF_INET6)) { + if (!pf_pull_hdr(pbuf, off, &esp, sizeof(esp), &action, + &reason, AF_INET6)) { log = action != PF_PASS; goto done; } #if DUMMYNET /* Traffic goes through dummynet first */ - action = pf_test_dummynet(&r, dir, kif, &m, &pd, fwa); - if (action == PF_DROP || m == NULL) { - *m0 = NULL; - return (action); + action = pf_test_dummynet(&r, dir, kif, &pbuf, &pd, fwa); + if (action == PF_DROP || pbuf == NULL) { + *pbufp = NULL; + return action; } #endif /* DUMMYNET */ action = pf_test_state_esp(&s, dir, kif, off, &pd); - if (pd.lmw < 0) + if (pd.lmw < 0) { goto done; + } PF_APPLE_UPDATE_PDESC_IPv6(); if (action == PF_PASS) { #if NPFSYNC @@ -8950,40 +10554,42 @@ nonormalize: r = s->rule.ptr; a = s->anchor.ptr; log = s->log; - } else if (s == NULL) + } else if (s == NULL) { action = pf_test_rule(&r, &s, dir, kif, - m, off, h, &pd, &a, &ruleset, NULL); + pbuf, off, h, &pd, &a, &ruleset, NULL); + } break; } case IPPROTO_GRE: { - struct pf_grev1_hdr grev1; + struct pf_grev1_hdr grev1; pd.hdr.grev1 = &grev1; - if (!pf_pull_hdr(m, off, &grev1, sizeof (grev1), &action, + if (!pf_pull_hdr(pbuf, off, &grev1, sizeof(grev1), &action, &reason, AF_INET6)) { log = (action != PF_PASS); goto done; } #if DUMMYNET /* Traffic goes through dummynet first */ - action = pf_test_dummynet(&r, dir, kif, &m, &pd, fwa); - if (action == PF_DROP || m == NULL) { - *m0 = NULL; - return (action); + action = pf_test_dummynet(&r, dir, kif, &pbuf, &pd, fwa); + if (action == PF_DROP || pbuf == NULL) { + *pbufp = NULL; + return action; } #endif /* DUMMYNET */ if ((ntohs(grev1.flags) & PF_GRE_FLAG_VERSION_MASK) == 1 && ntohs(grev1.protocol_type) == PF_GRE_PPP_ETHERTYPE) { if (ntohs(grev1.payload_length) > - m->m_pkthdr.len - off) { + pbuf->pb_packet_len - off) { action = PF_DROP; REASON_SET(&reason, PFRES_SHORT); goto done; } action = pf_test_state_grev1(&s, dir, kif, off, &pd); - if (pd.lmw < 0) + if (pd.lmw < 0) { goto done; + } PF_APPLE_UPDATE_PDESC_IPv6(); if (action == PF_PASS) { #if NPFSYNC @@ -8994,10 +10600,11 @@ nonormalize: log = s->log; break; } else if (s == NULL) { - action = pf_test_rule(&r, &s, dir, kif, m, off, - h, &pd, &a, &ruleset, NULL); - if (action == PF_PASS) + action = pf_test_rule(&r, &s, dir, kif, pbuf, + off, h, &pd, &a, &ruleset, NULL); + if (action == PF_PASS) { break; + } } } @@ -9007,15 +10614,16 @@ nonormalize: default: #if DUMMYNET /* Traffic goes through dummynet first */ - action = pf_test_dummynet(&r, dir, kif, &m, &pd, fwa); - if (action == PF_DROP || m == NULL) { - *m0 = NULL; - return (action); + action = pf_test_dummynet(&r, dir, kif, &pbuf, &pd, fwa); + if (action == PF_DROP || pbuf == NULL) { + *pbufp = NULL; + return action; } #endif /* DUMMYNET */ action = pf_test_state_other(&s, dir, kif, &pd); - if (pd.lmw < 0) + if (pd.lmw < 0) { goto done; + } PF_APPLE_UPDATE_PDESC_IPv6(); if (action == PF_PASS) { #if NPFSYNC @@ -9024,72 +10632,71 @@ nonormalize: r = s->rule.ptr; a = s->anchor.ptr; log = s->log; - } else if (s == NULL) - action = pf_test_rule(&r, &s, dir, kif, m, off, h, + } else if (s == NULL) { + action = pf_test_rule(&r, &s, dir, kif, pbuf, off, h, &pd, &a, &ruleset, NULL); + } break; } done: - *m0 = pd.mp; - PF_APPLE_UPDATE_PDESC_IPv6(); - - if (n != m) { - m_freem(n); - n = NULL; + if (action == PF_NAT64) { + *pbufp = NULL; + return action; } - /* handle dangerous IPv6 extension headers. */ - if (action == PF_PASS && rh_cnt && - !((s && s->allow_opts) || r->allow_opts)) { - action = PF_DROP; - REASON_SET(&reason, PFRES_IPOPTIONS); - log = 1; - DPFPRINTF(PF_DEBUG_MISC, - ("pf: dropping packet with dangerous v6 headers\n")); - } + *pbufp = pd.mp; + PF_APPLE_UPDATE_PDESC_IPv6(); - if ((s && s->tag) || PF_RTABLEID_IS_VALID(r->rtableid) || - (pd.pktflags & PKTF_FLOW_ID)) - (void) pf_tag_packet(m, pd.pf_mtag, s ? s->tag : 0, - r->rtableid, &pd); + /* handle dangerous IPv6 extension headers. */ + if (action != PF_DROP) { + if (action == PF_PASS && rh_cnt && + !((s && s->allow_opts) || r->allow_opts)) { + action = PF_DROP; + REASON_SET(&reason, PFRES_IPOPTIONS); + log = 1; + DPFPRINTF(PF_DEBUG_MISC, + ("pf: dropping packet with dangerous v6addr headers\n")); + } - if (action == PF_PASS) { -#if PF_ALTQ - if (altq_allowed && r->qid) { - if (pd.tos & IPTOS_LOWDELAY) - pd.pf_mtag->pftag_qid = r->pqid; - else - pd.pf_mtag->pftag_qid = r->qid; + if ((s && s->tag) || PF_RTABLEID_IS_VALID(r->rtableid) || + (pd.pktflags & PKTF_FLOW_ID)) { + (void) pf_tag_packet(pbuf, pd.pf_mtag, s ? s->tag : 0, + r->rtableid, &pd); } -#endif /* PF_ALTQ */ + + if (action == PF_PASS) { #if PF_ECN - /* add hints for ecn */ - pd.pf_mtag->pftag_hdr = h; - /* record address family */ - pd.pf_mtag->pftag_flags &= ~PF_TAG_HDR_INET; - pd.pf_mtag->pftag_flags |= PF_TAG_HDR_INET6; + /* add hints for ecn */ + pd.pf_mtag->pftag_hdr = h; + /* record address family */ + pd.pf_mtag->pftag_flags &= ~PF_TAG_HDR_INET; + pd.pf_mtag->pftag_flags |= PF_TAG_HDR_INET6; #endif /* PF_ECN */ - /* record protocol */ - m->m_pkthdr.pkt_proto = pd.proto; + /* record protocol */ + *pbuf->pb_proto = pd.proto; + if (dir == PF_IN && (pd.proto == IPPROTO_TCP || + pd.proto == IPPROTO_UDP) && s != NULL && + s->nat_rule.ptr != NULL && + (s->nat_rule.ptr->action == PF_RDR || + s->nat_rule.ptr->action == PF_BINAT) && + IN6_IS_ADDR_LOOPBACK(&pd.dst->v6addr)) { + pd.pf_mtag->pftag_flags |= PF_TAG_TRANSLATE_LOCALHOST; + } + } } - if (dir == PF_IN && action == PF_PASS && (pd.proto == IPPROTO_TCP || - pd.proto == IPPROTO_UDP) && s != NULL && s->nat_rule.ptr != NULL && - (s->nat_rule.ptr->action == PF_RDR || - s->nat_rule.ptr->action == PF_BINAT) && - IN6_IS_ADDR_LOOPBACK(&pd.dst->v6)) - pd.pf_mtag->pftag_flags |= PF_TAG_TRANSLATE_LOCALHOST; if (log) { struct pf_rule *lr; if (s != NULL && s->nat_rule.ptr != NULL && - s->nat_rule.ptr->log & PF_LOG_ALL) + s->nat_rule.ptr->log & PF_LOG_ALL) { lr = s->nat_rule.ptr; - else + } else { lr = r; - PFLOG_PACKET(kif, h, m, AF_INET6, dir, reason, lr, a, ruleset, + } + PFLOG_PACKET(kif, h, pbuf, AF_INET6, dir, reason, lr, a, ruleset, &pd); } @@ -9140,22 +10747,25 @@ done: &pd.naddr : &pd.baddr; } if (x == &pd.baddr || s == NULL) { - if (dir == PF_OUT) + if (dir == PF_OUT) { pd.src = x; - else + } else { pd.dst = x; + } } } - if (tr->src.addr.type == PF_ADDR_TABLE) + if (tr->src.addr.type == PF_ADDR_TABLE) { pfr_update_stats(tr->src.addr.p.tbl, (sk == NULL || sk->direction == dir) ? pd.src : pd.dst, pd.af, pd.tot_len, dir == PF_OUT, r->action == PF_PASS, tr->src.neg); - if (tr->dst.addr.type == PF_ADDR_TABLE) + } + if (tr->dst.addr.type == PF_ADDR_TABLE) { pfr_update_stats(tr->dst.addr.p.tbl, (sk == NULL || sk->direction == dir) ? pd.dst : pd.src, pd.af, pd.tot_len, dir == PF_OUT, r->action == PF_PASS, tr->dst.neg); + } } #if 0 @@ -9163,43 +10773,45 @@ done: m_freem(*m0); *m0 = NULL; action = PF_PASS; - } else if (r->rt) + } else if (r->rt) { /* pf_route6 can free the mbuf causing *m0 to become NULL */ pf_route6(m0, r, dir, kif->pfik_ifp, s, &pd); + } #else - VERIFY(m == NULL || pd.mp == NULL || pd.mp == m); + VERIFY(pbuf == NULL || pd.mp == NULL || pd.mp == pbuf); - if (*m0) { + if (*pbufp) { if (pd.lmw < 0) { REASON_SET(&reason, PFRES_MEMORY); action = PF_DROP; } if (action == PF_DROP) { - m_freem(*m0); - *m0 = NULL; - return (PF_DROP); + pbuf_destroy(*pbufp); + *pbufp = NULL; + return PF_DROP; } - *m0 = m; + *pbufp = pbuf; } if (action == PF_SYNPROXY_DROP) { - m_freem(*m0); - *m0 = NULL; + pbuf_destroy(*pbufp); + *pbufp = NULL; action = PF_PASS; } else if (r->rt) { - if (action == PF_PASS) { - m = *m0; - h = mtod(m, struct ip6_hdr *); - } - - /* pf_route6 can free the mbuf causing *m0 to become NULL */ - pf_route6(m0, r, dir, kif->pfik_ifp, s, &pd); + /* pf_route6 can free the mbuf causing *pbufp to become NULL */ + pf_route6(pbufp, r, dir, kif->pfik_ifp, s, &pd); } #endif /* 0 */ - return (action); + /* if reassembled packet passed, create new fragments */ + struct pf_fragment_tag *ftag = NULL; + if ((action == PF_PASS) && (*pbufp != NULL) && (fwd) && + ((ftag = pf_find_fragment_tag_pbuf(*pbufp)) != NULL)) { + action = pf_refragment6(ifp, pbufp, ftag); + } + return action; } #endif /* INET6 */ @@ -9207,7 +10819,7 @@ static int pf_check_congestion(struct ifqueue *ifq) { #pragma unused(ifq) - return (0); + return 0; } void @@ -9215,7 +10827,7 @@ pool_init(struct pool *pp, size_t size, unsigned int align, unsigned int ioff, int flags, const char *wchan, void *palloc) { #pragma unused(align, ioff, flags, palloc) - bzero(pp, sizeof (*pp)); + bzero(pp, sizeof(*pp)); pp->pool_zone = zinit(size, 1024 * size, PAGE_SIZE, wchan); if (pp->pool_zone != NULL) { zone_change(pp->pool_zone, Z_EXPAND, TRUE); @@ -9235,7 +10847,7 @@ pool_destroy(struct pool *pp) void pool_sethiwat(struct pool *pp, int n) { - pp->pool_hiwat = n; /* Currently unused */ + pp->pool_hiwat = n; /* Currently unused */ } void @@ -9250,7 +10862,7 @@ pool_get(struct pool *pp, int flags) { void *buf; - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); if (pp->pool_count > pp->pool_limit) { DPFPRINTF(PF_DEBUG_NOISY, @@ -9258,7 +10870,7 @@ pool_get(struct pool *pp, int flags) pp->pool_name != NULL ? pp->pool_name : "unknown", pp->pool_limit)); pp->pool_fails++; - return (NULL); + return NULL; } buf = zalloc_canblock(pp->pool_zone, (flags & (PR_NOWAIT | PR_WAITOK))); @@ -9266,13 +10878,13 @@ pool_get(struct pool *pp, int flags) pp->pool_count++; VERIFY(pp->pool_count != 0); } - return (buf); + return buf; } void pool_put(struct pool *pp, void *v) { - lck_mtx_assert(pf_lock, LCK_MTX_ASSERT_OWNED); + LCK_MTX_ASSERT(pf_lock, LCK_MTX_ASSERT_OWNED); zfree(pp->pool_zone, v); VERIFY(pp->pool_count != 0); @@ -9280,18 +10892,72 @@ pool_put(struct pool *pp, void *v) } struct pf_mtag * -pf_find_mtag(struct mbuf *m) +pf_find_mtag_pbuf(pbuf_t *pbuf) { - if (!(m->m_flags & M_PKTHDR)) - return (NULL); + return pbuf->pb_pftag; +} - return (m_pftag(m)); +struct pf_mtag * +pf_find_mtag(struct mbuf *m) +{ + return m_pftag(m); } struct pf_mtag * pf_get_mtag(struct mbuf *m) { - return (pf_find_mtag(m)); + return pf_find_mtag(m); +} + +struct pf_mtag * +pf_get_mtag_pbuf(pbuf_t *pbuf) +{ + return pf_find_mtag_pbuf(pbuf); +} + +struct pf_fragment_tag * +pf_copy_fragment_tag(struct mbuf *m, struct pf_fragment_tag *ftag, int how) +{ + struct m_tag *tag; + struct pf_mtag *pftag = pf_find_mtag(m); + + tag = m_tag_create(KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_PF_REASS, + sizeof(*ftag), how, m); + if (tag == NULL) { + return NULL; + } else { + m_tag_prepend(m, tag); + tag = tag + 1; + } + bcopy(ftag, tag, sizeof(*ftag)); + pftag->pftag_flags |= PF_TAG_REASSEMBLED; + return (struct pf_fragment_tag *)tag; +} + +struct pf_fragment_tag * +pf_find_fragment_tag(struct mbuf *m) +{ + struct m_tag *tag; + struct pf_fragment_tag *ftag; + struct pf_mtag *pftag = pf_find_mtag(m); + + tag = m_tag_locate(m, KERNEL_MODULE_TAG_ID, KERNEL_TAG_TYPE_PF_REASS, + NULL); + VERIFY((tag == NULL) || (pftag->pftag_flags & PF_TAG_REASSEMBLED)); + if (tag != NULL) { + tag = tag + 1; + } + ftag = (struct pf_fragment_tag *)tag; + return ftag; +} + +struct pf_fragment_tag * +pf_find_fragment_tag_pbuf(pbuf_t *pbuf) +{ + struct pf_mtag *mtag = pf_find_mtag_pbuf(pbuf); + + return (mtag->pftag_flags & PF_TAG_REASSEMBLED) ? + pbuf->pb_pf_fragtag : NULL; } uint64_t @@ -9300,7 +10966,7 @@ pf_time_second(void) struct timeval t; microuptime(&t); - return (t.tv_sec); + return t.tv_sec; } uint64_t @@ -9309,7 +10975,7 @@ pf_calendar_time_second(void) struct timeval t; getmicrotime(&t); - return (t.tv_sec); + return t.tv_sec; } static void * @@ -9318,17 +10984,19 @@ hook_establish(struct hook_desc_head *head, int tail, hook_fn_t fn, void *arg) struct hook_desc *hd; hd = _MALLOC(sizeof(*hd), M_DEVBUF, M_WAITOK); - if (hd == NULL) - return (NULL); + if (hd == NULL) { + return NULL; + } hd->hd_fn = fn; hd->hd_arg = arg; - if (tail) + if (tail) { TAILQ_INSERT_TAIL(head, hd, hd_list); - else + } else { TAILQ_INSERT_HEAD(head, hd, hd_list); + } - return (hd); + return hd; } static void @@ -9337,16 +11005,19 @@ hook_runloop(struct hook_desc_head *head, int flags) struct hook_desc *hd; if (!(flags & HOOK_REMOVE)) { - if (!(flags & HOOK_ABORT)) + if (!(flags & HOOK_ABORT)) { TAILQ_FOREACH(hd, head, hd_list) - hd->hd_fn(hd->hd_arg); + hd->hd_fn(hd->hd_arg); + } } else { while (!!(hd = TAILQ_FIRST(head))) { TAILQ_REMOVE(head, hd, hd_list); - if (!(flags & HOOK_ABORT)) + if (!(flags & HOOK_ABORT)) { hd->hd_fn(hd->hd_arg); - if (flags & HOOK_FREE) + } + if (flags & HOOK_FREE) { _FREE(hd, M_DEVBUF); + } } } }