mirror of
https://github.com/sonertari/SSLproxy
synced 2024-10-31 21:20:21 +00:00
4c8831bd90
Add VerifyPeer and AllowWrongHost options
2447 lines
72 KiB
C
2447 lines
72 KiB
C
/*
|
|
* Copyright (c) 2007-2015 Apple Inc. All rights reserved.
|
|
*
|
|
* @APPLE_OSREFERENCE_LICENSE_HEADER_START@
|
|
*
|
|
* This file contains Original Code and/or Modifications of Original Code
|
|
* as defined in and that are subject to the Apple Public Source License
|
|
* Version 2.0 (the 'License'). You may not use this file except in
|
|
* compliance with the License. The rights granted to you under the License
|
|
* may not be used to create, or enable the creation or redistribution of,
|
|
* unlawful or unlicensed copies of an Apple operating system, or to
|
|
* circumvent, violate, or enable the circumvention or violation of, any
|
|
* terms of an Apple operating system software license agreement.
|
|
*
|
|
* Please obtain a copy of the License at
|
|
* http://www.opensource.apple.com/apsl/ and read it before using this file.
|
|
*
|
|
* The Original Code and all software distributed under the License are
|
|
* distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
|
|
* EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
|
|
* INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
|
|
* FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
|
|
* Please see the License for the specific language governing rights and
|
|
* limitations under the License.
|
|
*
|
|
* @APPLE_OSREFERENCE_LICENSE_HEADER_END@
|
|
*/
|
|
|
|
/* $apfw: git commit b6bf13f8321283cd7ee82b1795e86506084b1b95 $ */
|
|
/* $OpenBSD: pfvar.h,v 1.259 2007/12/02 12:08:04 pascoe Exp $ */
|
|
|
|
/*
|
|
* Copyright (c) 2001 Daniel Hartmeier
|
|
* NAT64 - Copyright (c) 2010 Viagenie Inc. (http://www.viagenie.ca)
|
|
* All rights reserved.
|
|
*
|
|
* Redistribution and use in source and binary forms, with or without
|
|
* modification, are permitted provided that the following conditions
|
|
* are met:
|
|
*
|
|
* - Redistributions of source code must retain the above copyright
|
|
* notice, this list of conditions and the following disclaimer.
|
|
* - Redistributions in binary form must reproduce the above
|
|
* copyright notice, this list of conditions and the following
|
|
* disclaimer in the documentation and/or other materials provided
|
|
* with the distribution.
|
|
*
|
|
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
|
|
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
|
|
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
|
|
* FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
|
|
* COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
|
|
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
|
|
* BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
|
|
* LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
|
|
* CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
|
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
|
|
* ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
|
|
* POSSIBILITY OF SUCH DAMAGE.
|
|
*
|
|
*/
|
|
|
|
#ifndef _NET_PFVAR_H_
|
|
#define _NET_PFVAR_H_
|
|
|
|
#ifdef PRIVATE
|
|
/*
|
|
* XXX
|
|
* XXX Private interfaces. Do not include this file; use pfctl(8) instead.
|
|
* XXX
|
|
*/
|
|
#if PF || !defined(KERNEL)
|
|
|
|
#ifdef __cplusplus
|
|
extern "C" {
|
|
#endif
|
|
|
|
#include <stdbool.h>
|
|
#include <sys/param.h>
|
|
#include <sys/types.h>
|
|
#include <sys/queue.h>
|
|
#include <libkern/tree.h>
|
|
|
|
#include <net/radix.h>
|
|
#include <netinet/in.h>
|
|
#include <net/if_var.h>
|
|
#ifdef KERNEL
|
|
#include <kern/kern_types.h>
|
|
#include <kern/zalloc.h>
|
|
#include <kern/locks.h>
|
|
|
|
#include <machine/endian.h>
|
|
#include <sys/systm.h>
|
|
|
|
#if BYTE_ORDER == BIG_ENDIAN
|
|
#define htobe64(x) (x)
|
|
#else /* LITTLE ENDIAN */
|
|
#define htobe64(x) __DARWIN_OSSwapInt64(x)
|
|
#endif /* LITTLE_ENDIAN */
|
|
|
|
#define be64toh(x) htobe64(x)
|
|
|
|
extern lck_rw_t *pf_perim_lock;
|
|
extern lck_mtx_t *pf_lock;
|
|
|
|
struct pool {
|
|
struct zone *pool_zone; /* pointer to backend zone */
|
|
const char *pool_name; /* name of pool */
|
|
unsigned int pool_count; /* # of outstanding elements */
|
|
unsigned int pool_hiwat; /* high watermark */
|
|
unsigned int pool_limit; /* hard limit */
|
|
unsigned int pool_fails; /* # of failed allocs due to limit */
|
|
};
|
|
|
|
#define PR_NOWAIT FALSE
|
|
#define PR_WAITOK TRUE
|
|
|
|
__private_extern__ void pool_init(struct pool *, size_t, unsigned int,
|
|
unsigned int, int, const char *, void *);
|
|
__private_extern__ void pool_destroy(struct pool *);
|
|
__private_extern__ void pool_sethiwat(struct pool *, int);
|
|
__private_extern__ void pool_sethardlimit(struct pool *, int,
|
|
const char *, int);
|
|
__private_extern__ void *pool_get(struct pool *, int);
|
|
__private_extern__ void pool_put(struct pool *, void *);
|
|
__private_extern__ u_int64_t pf_time_second(void);
|
|
__private_extern__ u_int64_t pf_calendar_time_second(void);
|
|
#endif /* KERNEL */
|
|
|
|
union sockaddr_union {
|
|
struct sockaddr sa;
|
|
struct sockaddr_in sin;
|
|
struct sockaddr_in6 sin6;
|
|
};
|
|
|
|
#define PF_TCPS_PROXY_SRC ((TCP_NSTATES)+0)
|
|
#define PF_TCPS_PROXY_DST ((TCP_NSTATES)+1)
|
|
|
|
#define PF_MD5_DIGEST_LENGTH 16
|
|
#ifdef MD5_DIGEST_LENGTH
|
|
#if PF_MD5_DIGEST_LENGTH != MD5_DIGEST_LENGTH
|
|
#error
|
|
#endif /* PF_MD5_DIGEST_LENGTH != MD5_DIGEST_LENGTH */
|
|
#endif /* MD5_DIGEST_LENGTH */
|
|
|
|
#ifdef KERNEL
|
|
struct ip;
|
|
struct ip6_hdr;
|
|
struct tcphdr;
|
|
struct pf_grev1_hdr;
|
|
struct pf_esp_hdr;
|
|
#endif /* KERNEL */
|
|
|
|
#define PF_GRE_PPTP_VARIANT 0x01
|
|
|
|
enum { PF_INOUT, PF_IN, PF_OUT };
|
|
enum { PF_PASS, PF_DROP, PF_SCRUB, PF_NOSCRUB, PF_NAT, PF_NONAT,
|
|
PF_BINAT, PF_NOBINAT, PF_RDR, PF_NORDR, PF_SYNPROXY_DROP,
|
|
PF_DUMMYNET, PF_NODUMMYNET, PF_NAT64, PF_NONAT64 };
|
|
enum { PF_RULESET_SCRUB, PF_RULESET_FILTER, PF_RULESET_NAT,
|
|
PF_RULESET_BINAT, PF_RULESET_RDR, PF_RULESET_DUMMYNET,
|
|
PF_RULESET_MAX };
|
|
enum { PF_OP_NONE, PF_OP_IRG, PF_OP_EQ, PF_OP_NE, PF_OP_LT,
|
|
PF_OP_LE, PF_OP_GT, PF_OP_GE, PF_OP_XRG, PF_OP_RRG };
|
|
enum { PF_DEBUG_NONE, PF_DEBUG_URGENT, PF_DEBUG_MISC, PF_DEBUG_NOISY };
|
|
enum { PF_CHANGE_NONE, PF_CHANGE_ADD_HEAD, PF_CHANGE_ADD_TAIL,
|
|
PF_CHANGE_ADD_BEFORE, PF_CHANGE_ADD_AFTER,
|
|
PF_CHANGE_REMOVE, PF_CHANGE_GET_TICKET };
|
|
enum { PF_GET_NONE, PF_GET_CLR_CNTR };
|
|
|
|
/*
|
|
* Note about PFTM_*: real indices into pf_rule.timeout[] come before
|
|
* PFTM_MAX, special cases afterwards. See pf_state_expires().
|
|
*/
|
|
enum { PFTM_TCP_FIRST_PACKET, PFTM_TCP_OPENING, PFTM_TCP_ESTABLISHED,
|
|
PFTM_TCP_CLOSING, PFTM_TCP_FIN_WAIT, PFTM_TCP_CLOSED,
|
|
PFTM_UDP_FIRST_PACKET, PFTM_UDP_SINGLE, PFTM_UDP_MULTIPLE,
|
|
PFTM_ICMP_FIRST_PACKET, PFTM_ICMP_ERROR_REPLY,
|
|
PFTM_GREv1_FIRST_PACKET, PFTM_GREv1_INITIATING,
|
|
PFTM_GREv1_ESTABLISHED, PFTM_ESP_FIRST_PACKET, PFTM_ESP_INITIATING,
|
|
PFTM_ESP_ESTABLISHED, PFTM_OTHER_FIRST_PACKET, PFTM_OTHER_SINGLE,
|
|
PFTM_OTHER_MULTIPLE, PFTM_FRAG, PFTM_INTERVAL,
|
|
PFTM_ADAPTIVE_START, PFTM_ADAPTIVE_END, PFTM_SRC_NODE,
|
|
PFTM_TS_DIFF, PFTM_MAX, PFTM_PURGE, PFTM_UNLINKED };
|
|
|
|
/* PFTM default values */
|
|
#define PFTM_TCP_FIRST_PACKET_VAL 120 /* First TCP packet */
|
|
#define PFTM_TCP_OPENING_VAL 30 /* No response yet */
|
|
#define PFTM_TCP_ESTABLISHED_VAL (24 * 60 * 60) /* Established */
|
|
#define PFTM_TCP_CLOSING_VAL (15 * 60) /* Half closed */
|
|
#define PFTM_TCP_FIN_WAIT_VAL 45 /* Got both FINs */
|
|
#define PFTM_TCP_CLOSED_VAL 90 /* Got a RST */
|
|
#define PFTM_UDP_FIRST_PACKET_VAL 60 /* First UDP packet */
|
|
#define PFTM_UDP_SINGLE_VAL 30 /* Unidirectional */
|
|
#define PFTM_UDP_MULTIPLE_VAL 60 /* Bidirectional */
|
|
#define PFTM_ICMP_FIRST_PACKET_VAL 20 /* First ICMP packet */
|
|
#define PFTM_ICMP_ERROR_REPLY_VAL 10 /* Got error response */
|
|
#define PFTM_GREv1_FIRST_PACKET_VAL 120
|
|
#define PFTM_GREv1_INITIATING_VAL 30
|
|
#define PFTM_GREv1_ESTABLISHED_VAL 1800
|
|
#define PFTM_ESP_FIRST_PACKET_VAL 120
|
|
#define PFTM_ESP_INITIATING_VAL 30
|
|
#define PFTM_ESP_ESTABLISHED_VAL 900
|
|
#define PFTM_OTHER_FIRST_PACKET_VAL 60 /* First packet */
|
|
#define PFTM_OTHER_SINGLE_VAL 30 /* Unidirectional */
|
|
#define PFTM_OTHER_MULTIPLE_VAL 60 /* Bidirectional */
|
|
#define PFTM_FRAG_VAL 30 /* Fragment expire */
|
|
#define PFTM_INTERVAL_VAL 10 /* Expire interval */
|
|
#define PFTM_SRC_NODE_VAL 0 /* Source tracking */
|
|
#define PFTM_TS_DIFF_VAL 30 /* Allowed TS diff */
|
|
|
|
enum { PF_NOPFROUTE, PF_FASTROUTE, PF_ROUTETO, PF_DUPTO, PF_REPLYTO };
|
|
enum { PF_LIMIT_STATES,
|
|
PF_LIMIT_APP_STATES,
|
|
PF_LIMIT_SRC_NODES, PF_LIMIT_FRAGS,
|
|
PF_LIMIT_TABLES, PF_LIMIT_TABLE_ENTRIES, PF_LIMIT_MAX };
|
|
#define PF_POOL_IDMASK 0x0f
|
|
enum { PF_POOL_NONE, PF_POOL_BITMASK, PF_POOL_RANDOM,
|
|
PF_POOL_SRCHASH, PF_POOL_ROUNDROBIN };
|
|
enum { PF_ADDR_ADDRMASK, PF_ADDR_NOROUTE, PF_ADDR_DYNIFTL,
|
|
PF_ADDR_TABLE, PF_ADDR_RTLABEL, PF_ADDR_URPFFAILED,
|
|
PF_ADDR_RANGE };
|
|
#define PF_POOL_TYPEMASK 0x0f
|
|
#define PF_POOL_STICKYADDR 0x20
|
|
#define PF_WSCALE_FLAG 0x80
|
|
#define PF_WSCALE_MASK 0x0f
|
|
|
|
#define PF_LOG 0x01
|
|
#define PF_LOG_ALL 0x02
|
|
#define PF_LOG_SOCKET_LOOKUP 0x04
|
|
|
|
struct pf_addr {
|
|
union {
|
|
struct in_addr v4;
|
|
struct in6_addr v6;
|
|
u_int8_t addr8[16];
|
|
u_int16_t addr16[8];
|
|
u_int32_t addr32[4];
|
|
} pfa; /* 128-bit address */
|
|
#define v4 pfa.v4
|
|
#define v6 pfa.v6
|
|
#define addr8 pfa.addr8
|
|
#define addr16 pfa.addr16
|
|
#define addr32 pfa.addr32
|
|
};
|
|
|
|
#define PF_TABLE_NAME_SIZE 32
|
|
|
|
#define PFI_AFLAG_NETWORK 0x01
|
|
#define PFI_AFLAG_BROADCAST 0x02
|
|
#define PFI_AFLAG_PEER 0x04
|
|
#define PFI_AFLAG_MODEMASK 0x07
|
|
#define PFI_AFLAG_NOALIAS 0x08
|
|
|
|
#ifndef RTLABEL_LEN
|
|
#define RTLABEL_LEN 32
|
|
#endif
|
|
|
|
struct pf_addr_wrap {
|
|
union {
|
|
struct {
|
|
struct pf_addr addr;
|
|
struct pf_addr mask;
|
|
} a;
|
|
char ifname[IFNAMSIZ];
|
|
char tblname[PF_TABLE_NAME_SIZE];
|
|
char rtlabelname[RTLABEL_LEN];
|
|
u_int32_t rtlabel;
|
|
} v;
|
|
union {
|
|
#ifdef KERNEL
|
|
struct pfi_dynaddr *dyn __attribute__((aligned(8)));
|
|
struct pfr_ktable *tbl __attribute__((aligned(8)));
|
|
#else /* !KERNEL */
|
|
void *dyn __attribute__((aligned(8)));
|
|
void *tbl __attribute__((aligned(8)));
|
|
#endif /* !KERNEL */
|
|
int dyncnt __attribute__((aligned(8)));
|
|
int tblcnt __attribute__((aligned(8)));
|
|
} p __attribute__((aligned(8)));
|
|
u_int8_t type; /* PF_ADDR_* */
|
|
u_int8_t iflags; /* PFI_AFLAG_* */
|
|
};
|
|
|
|
struct pf_port_range {
|
|
u_int16_t port[2];
|
|
u_int8_t op;
|
|
};
|
|
|
|
union pf_rule_xport {
|
|
struct pf_port_range range;
|
|
u_int16_t call_id;
|
|
u_int32_t spi;
|
|
};
|
|
|
|
#ifdef KERNEL
|
|
struct pfi_dynaddr {
|
|
TAILQ_ENTRY(pfi_dynaddr) entry;
|
|
struct pf_addr pfid_addr4;
|
|
struct pf_addr pfid_mask4;
|
|
struct pf_addr pfid_addr6;
|
|
struct pf_addr pfid_mask6;
|
|
struct pfr_ktable *pfid_kt;
|
|
struct pfi_kif *pfid_kif;
|
|
void *pfid_hook_cookie;
|
|
int pfid_net; /* mask or 128 */
|
|
int pfid_acnt4; /* address count IPv4 */
|
|
int pfid_acnt6; /* address count IPv6 */
|
|
sa_family_t pfid_af; /* rule af */
|
|
u_int8_t pfid_iflags; /* PFI_AFLAG_* */
|
|
};
|
|
|
|
/*
|
|
* Address manipulation macros
|
|
*/
|
|
|
|
#if INET
|
|
#if !INET6
|
|
#define PF_INET_ONLY
|
|
#endif /* ! INET6 */
|
|
#endif /* INET */
|
|
|
|
#if INET6
|
|
#if !INET
|
|
#define PF_INET6_ONLY
|
|
#endif /* ! INET */
|
|
#endif /* INET6 */
|
|
|
|
#if INET
|
|
#if INET6
|
|
#define PF_INET_INET6
|
|
#endif /* INET6 */
|
|
#endif /* INET */
|
|
|
|
#else /* !KERNEL */
|
|
|
|
#define PF_INET_INET6
|
|
|
|
#endif /* !KERNEL */
|
|
|
|
/* Both IPv4 and IPv6 */
|
|
#ifdef PF_INET_INET6
|
|
|
|
#define PF_AEQ(a, b, c) \
|
|
((c == AF_INET && (a)->addr32[0] == (b)->addr32[0]) || \
|
|
((a)->addr32[3] == (b)->addr32[3] && \
|
|
(a)->addr32[2] == (b)->addr32[2] && \
|
|
(a)->addr32[1] == (b)->addr32[1] && \
|
|
(a)->addr32[0] == (b)->addr32[0])) \
|
|
|
|
#define PF_ANEQ(a, b, c) \
|
|
((c == AF_INET && (a)->addr32[0] != (b)->addr32[0]) || \
|
|
((a)->addr32[3] != (b)->addr32[3] || \
|
|
(a)->addr32[2] != (b)->addr32[2] || \
|
|
(a)->addr32[1] != (b)->addr32[1] || \
|
|
(a)->addr32[0] != (b)->addr32[0])) \
|
|
|
|
#define PF_ALEQ(a, b, c) \
|
|
((c == AF_INET && (a)->addr32[0] <= (b)->addr32[0]) || \
|
|
((a)->addr32[3] <= (b)->addr32[3] && \
|
|
(a)->addr32[2] <= (b)->addr32[2] && \
|
|
(a)->addr32[1] <= (b)->addr32[1] && \
|
|
(a)->addr32[0] <= (b)->addr32[0])) \
|
|
|
|
#define PF_AZERO(a, c) \
|
|
((c == AF_INET && !(a)->addr32[0]) || \
|
|
(!(a)->addr32[0] && !(a)->addr32[1] && \
|
|
!(a)->addr32[2] && !(a)->addr32[3])) \
|
|
|
|
#define PF_MATCHA(n, a, m, b, f) \
|
|
pf_match_addr(n, a, m, b, f)
|
|
|
|
#define PF_ACPY(a, b, f) \
|
|
pf_addrcpy(a, b, f)
|
|
|
|
#define PF_AINC(a, f) \
|
|
pf_addr_inc(a, f)
|
|
|
|
#define PF_POOLMASK(a, b, c, d, f) \
|
|
pf_poolmask(a, b, c, d, f)
|
|
|
|
#else
|
|
|
|
/* Just IPv6 */
|
|
|
|
#ifdef PF_INET6_ONLY
|
|
|
|
#define PF_AEQ(a, b, c) \
|
|
((a)->addr32[3] == (b)->addr32[3] && \
|
|
(a)->addr32[2] == (b)->addr32[2] && \
|
|
(a)->addr32[1] == (b)->addr32[1] && \
|
|
(a)->addr32[0] == (b)->addr32[0]) \
|
|
|
|
#define PF_ANEQ(a, b, c) \
|
|
((a)->addr32[3] != (b)->addr32[3] || \
|
|
(a)->addr32[2] != (b)->addr32[2] || \
|
|
(a)->addr32[1] != (b)->addr32[1] || \
|
|
(a)->addr32[0] != (b)->addr32[0]) \
|
|
|
|
#define PF_ALEQ(a, b, c) \
|
|
((a)->addr32[3] <= (b)->addr32[3] && \
|
|
(a)->addr32[2] <= (b)->addr32[2] && \
|
|
(a)->addr32[1] <= (b)->addr32[1] && \
|
|
(a)->addr32[0] <= (b)->addr32[0]) \
|
|
|
|
#define PF_AZERO(a, c) \
|
|
(!(a)->addr32[0] && \
|
|
!(a)->addr32[1] && \
|
|
!(a)->addr32[2] && \
|
|
!(a)->addr32[3]) \
|
|
|
|
#define PF_MATCHA(n, a, m, b, f) \
|
|
pf_match_addr(n, a, m, b, f)
|
|
|
|
#define PF_ACPY(a, b, f) \
|
|
pf_addrcpy(a, b, f)
|
|
|
|
#define PF_AINC(a, f) \
|
|
pf_addr_inc(a, f)
|
|
|
|
#define PF_POOLMASK(a, b, c, d, f) \
|
|
pf_poolmask(a, b, c, d, f)
|
|
|
|
#else
|
|
|
|
/* Just IPv4 */
|
|
#ifdef PF_INET_ONLY
|
|
|
|
#define PF_AEQ(a, b, c) \
|
|
((a)->addr32[0] == (b)->addr32[0])
|
|
|
|
#define PF_ANEQ(a, b, c) \
|
|
((a)->addr32[0] != (b)->addr32[0])
|
|
|
|
#define PF_ALEQ(a, b, c) \
|
|
((a)->addr32[0] <= (b)->addr32[0])
|
|
|
|
#define PF_AZERO(a, c) \
|
|
(!(a)->addr32[0])
|
|
|
|
#define PF_MATCHA(n, a, m, b, f) \
|
|
pf_match_addr(n, a, m, b, f)
|
|
|
|
#define PF_ACPY(a, b, f) \
|
|
(a)->v4.s_addr = (b)->v4.s_addr
|
|
|
|
#define PF_AINC(a, f) \
|
|
do { \
|
|
(a)->addr32[0] = htonl(ntohl((a)->addr32[0]) + 1); \
|
|
} while (0)
|
|
|
|
#define PF_POOLMASK(a, b, c, d, f) \
|
|
do { \
|
|
(a)->addr32[0] = ((b)->addr32[0] & (c)->addr32[0]) | \
|
|
(((c)->addr32[0] ^ 0xffffffff) & (d)->addr32[0]); \
|
|
} while (0)
|
|
|
|
#endif /* PF_INET_ONLY */
|
|
#endif /* PF_INET6_ONLY */
|
|
#endif /* PF_INET_INET6 */
|
|
|
|
#ifdef KERNEL
|
|
#define PF_MISMATCHAW(aw, x, af, neg, ifp) \
|
|
( \
|
|
(((aw)->type == PF_ADDR_NOROUTE && \
|
|
pf_routable((x), (af), NULL)) || \
|
|
(((aw)->type == PF_ADDR_URPFFAILED && (ifp) != NULL && \
|
|
pf_routable((x), (af), (ifp))) || \
|
|
((aw)->type == PF_ADDR_RTLABEL && \
|
|
!pf_rtlabel_match((x), (af), (aw))) || \
|
|
((aw)->type == PF_ADDR_TABLE && \
|
|
!pfr_match_addr((aw)->p.tbl, (x), (af))) || \
|
|
((aw)->type == PF_ADDR_DYNIFTL && \
|
|
!pfi_match_addr((aw)->p.dyn, (x), (af))) || \
|
|
((aw)->type == PF_ADDR_RANGE && \
|
|
!pf_match_addr_range(&(aw)->v.a.addr, \
|
|
&(aw)->v.a.mask, (x), (af))) || \
|
|
((aw)->type == PF_ADDR_ADDRMASK && \
|
|
!PF_AZERO(&(aw)->v.a.mask, (af)) && \
|
|
!PF_MATCHA(0, &(aw)->v.a.addr, \
|
|
&(aw)->v.a.mask, (x), (af))))) != \
|
|
(neg) \
|
|
)
|
|
#endif /* KERNEL */
|
|
|
|
struct pf_rule_uid {
|
|
uid_t uid[2];
|
|
u_int8_t op;
|
|
u_int8_t _pad[3];
|
|
};
|
|
|
|
struct pf_rule_gid {
|
|
uid_t gid[2];
|
|
u_int8_t op;
|
|
u_int8_t _pad[3];
|
|
};
|
|
|
|
struct pf_rule_addr {
|
|
struct pf_addr_wrap addr;
|
|
union pf_rule_xport xport;
|
|
u_int8_t neg;
|
|
};
|
|
|
|
struct pf_pooladdr {
|
|
struct pf_addr_wrap addr;
|
|
TAILQ_ENTRY(pf_pooladdr) entries;
|
|
#if !defined(__LP64__)
|
|
u_int32_t _pad[2];
|
|
#endif /* !__LP64__ */
|
|
char ifname[IFNAMSIZ];
|
|
#ifdef KERNEL
|
|
struct pfi_kif *kif __attribute__((aligned(8)));
|
|
#else /* !KERNEL */
|
|
void *kif __attribute__((aligned(8)));
|
|
#endif /* !KERNEL */
|
|
};
|
|
|
|
TAILQ_HEAD(pf_palist, pf_pooladdr);
|
|
|
|
struct pf_poolhashkey {
|
|
union {
|
|
u_int8_t key8[16];
|
|
u_int16_t key16[8];
|
|
u_int32_t key32[4];
|
|
} pfk; /* 128-bit hash key */
|
|
#define key8 pfk.key8
|
|
#define key16 pfk.key16
|
|
#define key32 pfk.key32
|
|
};
|
|
|
|
struct pf_pool {
|
|
struct pf_palist list;
|
|
#if !defined(__LP64__)
|
|
u_int32_t _pad[2];
|
|
#endif /* !__LP64__ */
|
|
#ifdef KERNEL
|
|
struct pf_pooladdr *cur __attribute__((aligned(8)));
|
|
#else /* !KERNEL */
|
|
void *cur __attribute__((aligned(8)));
|
|
#endif /* !KERNEL */
|
|
struct pf_poolhashkey key __attribute__((aligned(8)));
|
|
struct pf_addr counter;
|
|
int tblidx;
|
|
u_int16_t proxy_port[2];
|
|
u_int8_t port_op;
|
|
u_int8_t opts;
|
|
sa_family_t af;
|
|
};
|
|
|
|
|
|
/* A packed Operating System description for fingerprinting */
|
|
typedef u_int32_t pf_osfp_t;
|
|
#define PF_OSFP_ANY ((pf_osfp_t)0)
|
|
#define PF_OSFP_UNKNOWN ((pf_osfp_t)-1)
|
|
#define PF_OSFP_NOMATCH ((pf_osfp_t)-2)
|
|
|
|
struct pf_osfp_entry {
|
|
SLIST_ENTRY(pf_osfp_entry) fp_entry;
|
|
#if !defined(__LP64__)
|
|
u_int32_t _pad;
|
|
#endif /* !__LP64__ */
|
|
pf_osfp_t fp_os;
|
|
int fp_enflags;
|
|
#define PF_OSFP_EXPANDED 0x001 /* expanded entry */
|
|
#define PF_OSFP_GENERIC 0x002 /* generic signature */
|
|
#define PF_OSFP_NODETAIL 0x004 /* no p0f details */
|
|
#define PF_OSFP_LEN 32
|
|
char fp_class_nm[PF_OSFP_LEN];
|
|
char fp_version_nm[PF_OSFP_LEN];
|
|
char fp_subtype_nm[PF_OSFP_LEN];
|
|
};
|
|
#define PF_OSFP_ENTRY_EQ(a, b) \
|
|
((a)->fp_os == (b)->fp_os && \
|
|
memcmp((a)->fp_class_nm, (b)->fp_class_nm, PF_OSFP_LEN) == 0 && \
|
|
memcmp((a)->fp_version_nm, (b)->fp_version_nm, PF_OSFP_LEN) == 0 && \
|
|
memcmp((a)->fp_subtype_nm, (b)->fp_subtype_nm, PF_OSFP_LEN) == 0)
|
|
|
|
/* handle pf_osfp_t packing */
|
|
#define _FP_RESERVED_BIT 1 /* For the special negative #defines */
|
|
#define _FP_UNUSED_BITS 1
|
|
#define _FP_CLASS_BITS 10 /* OS Class (Windows, Linux) */
|
|
#define _FP_VERSION_BITS 10 /* OS version (95, 98, NT, 2.4.54, 3.2) */
|
|
#define _FP_SUBTYPE_BITS 10 /* patch level (NT SP4, SP3, ECN patch) */
|
|
#define PF_OSFP_UNPACK(osfp, class, version, subtype) do { \
|
|
(class) = ((osfp) >> (_FP_VERSION_BITS+_FP_SUBTYPE_BITS)) & \
|
|
((1 << _FP_CLASS_BITS) - 1); \
|
|
(version) = ((osfp) >> _FP_SUBTYPE_BITS) & \
|
|
((1 << _FP_VERSION_BITS) - 1);\
|
|
(subtype) = (osfp) & ((1 << _FP_SUBTYPE_BITS) - 1); \
|
|
} while (0)
|
|
#define PF_OSFP_PACK(osfp, class, version, subtype) do { \
|
|
(osfp) = ((class) & ((1 << _FP_CLASS_BITS) - 1)) << (_FP_VERSION_BITS \
|
|
+ _FP_SUBTYPE_BITS); \
|
|
(osfp) |= ((version) & ((1 << _FP_VERSION_BITS) - 1)) << \
|
|
_FP_SUBTYPE_BITS; \
|
|
(osfp) |= (subtype) & ((1 << _FP_SUBTYPE_BITS) - 1); \
|
|
} while (0)
|
|
|
|
/* the fingerprint of an OSes TCP SYN packet */
|
|
typedef u_int64_t pf_tcpopts_t;
|
|
struct pf_os_fingerprint {
|
|
SLIST_HEAD(pf_osfp_enlist, pf_osfp_entry) fp_oses; /* list of matches */
|
|
pf_tcpopts_t fp_tcpopts; /* packed TCP options */
|
|
u_int16_t fp_wsize; /* TCP window size */
|
|
u_int16_t fp_psize; /* ip->ip_len */
|
|
u_int16_t fp_mss; /* TCP MSS */
|
|
u_int16_t fp_flags;
|
|
#define PF_OSFP_WSIZE_MOD 0x0001 /* Window modulus */
|
|
#define PF_OSFP_WSIZE_DC 0x0002 /* Window don't care */
|
|
#define PF_OSFP_WSIZE_MSS 0x0004 /* Window multiple of MSS */
|
|
#define PF_OSFP_WSIZE_MTU 0x0008 /* Window multiple of MTU */
|
|
#define PF_OSFP_PSIZE_MOD 0x0010 /* packet size modulus */
|
|
#define PF_OSFP_PSIZE_DC 0x0020 /* packet size don't care */
|
|
#define PF_OSFP_WSCALE 0x0040 /* TCP window scaling */
|
|
#define PF_OSFP_WSCALE_MOD 0x0080 /* TCP window scale modulus */
|
|
#define PF_OSFP_WSCALE_DC 0x0100 /* TCP window scale dont-care */
|
|
#define PF_OSFP_MSS 0x0200 /* TCP MSS */
|
|
#define PF_OSFP_MSS_MOD 0x0400 /* TCP MSS modulus */
|
|
#define PF_OSFP_MSS_DC 0x0800 /* TCP MSS dont-care */
|
|
#define PF_OSFP_DF 0x1000 /* IPv4 don't fragment bit */
|
|
#define PF_OSFP_TS0 0x2000 /* Zero timestamp */
|
|
#define PF_OSFP_INET6 0x4000 /* IPv6 */
|
|
u_int8_t fp_optcnt; /* TCP option count */
|
|
u_int8_t fp_wscale; /* TCP window scaling */
|
|
u_int8_t fp_ttl; /* IPv4 TTL */
|
|
#define PF_OSFP_MAXTTL_OFFSET 40
|
|
/* TCP options packing */
|
|
#define PF_OSFP_TCPOPT_NOP 0x0 /* TCP NOP option */
|
|
#define PF_OSFP_TCPOPT_WSCALE 0x1 /* TCP window scaling option */
|
|
#define PF_OSFP_TCPOPT_MSS 0x2 /* TCP max segment size opt */
|
|
#define PF_OSFP_TCPOPT_SACK 0x3 /* TCP SACK OK option */
|
|
#define PF_OSFP_TCPOPT_TS 0x4 /* TCP timestamp option */
|
|
#define PF_OSFP_TCPOPT_BITS 3 /* bits used by each option */
|
|
#define PF_OSFP_MAX_OPTS \
|
|
(sizeof(((struct pf_os_fingerprint *)0)->fp_tcpopts) * 8) \
|
|
/ PF_OSFP_TCPOPT_BITS
|
|
|
|
SLIST_ENTRY(pf_os_fingerprint) fp_next;
|
|
};
|
|
|
|
struct pf_osfp_ioctl {
|
|
struct pf_osfp_entry fp_os;
|
|
pf_tcpopts_t fp_tcpopts; /* packed TCP options */
|
|
u_int16_t fp_wsize; /* TCP window size */
|
|
u_int16_t fp_psize; /* ip->ip_len */
|
|
u_int16_t fp_mss; /* TCP MSS */
|
|
u_int16_t fp_flags;
|
|
u_int8_t fp_optcnt; /* TCP option count */
|
|
u_int8_t fp_wscale; /* TCP window scaling */
|
|
u_int8_t fp_ttl; /* IPv4 TTL */
|
|
|
|
int fp_getnum; /* DIOCOSFPGET number */
|
|
};
|
|
|
|
|
|
union pf_rule_ptr {
|
|
struct pf_rule *ptr __attribute__((aligned(8)));
|
|
u_int32_t nr __attribute__((aligned(8)));
|
|
} __attribute__((aligned(8)));
|
|
|
|
#define PF_ANCHOR_NAME_SIZE 64
|
|
|
|
struct pf_rule {
|
|
struct pf_rule_addr src;
|
|
struct pf_rule_addr dst;
|
|
#define PF_SKIP_IFP 0
|
|
#define PF_SKIP_DIR 1
|
|
#define PF_SKIP_AF 2
|
|
#define PF_SKIP_PROTO 3
|
|
#define PF_SKIP_SRC_ADDR 4
|
|
#define PF_SKIP_SRC_PORT 5
|
|
#define PF_SKIP_DST_ADDR 6
|
|
#define PF_SKIP_DST_PORT 7
|
|
#define PF_SKIP_COUNT 8
|
|
union pf_rule_ptr skip[PF_SKIP_COUNT];
|
|
#define PF_RULE_LABEL_SIZE 64
|
|
char label[PF_RULE_LABEL_SIZE];
|
|
#define PF_QNAME_SIZE 64
|
|
char ifname[IFNAMSIZ];
|
|
char qname[PF_QNAME_SIZE];
|
|
char pqname[PF_QNAME_SIZE];
|
|
#define PF_TAG_NAME_SIZE 64
|
|
char tagname[PF_TAG_NAME_SIZE];
|
|
char match_tagname[PF_TAG_NAME_SIZE];
|
|
|
|
char overload_tblname[PF_TABLE_NAME_SIZE];
|
|
|
|
TAILQ_ENTRY(pf_rule) entries;
|
|
#if !defined(__LP64__)
|
|
u_int32_t _pad[2];
|
|
#endif /* !__LP64__ */
|
|
struct pf_pool rpool;
|
|
|
|
u_int64_t evaluations;
|
|
u_int64_t packets[2];
|
|
u_int64_t bytes[2];
|
|
|
|
u_int64_t ticket;
|
|
#define PF_OWNER_NAME_SIZE 64
|
|
char owner[PF_OWNER_NAME_SIZE];
|
|
u_int32_t priority;
|
|
|
|
#ifdef KERNEL
|
|
struct pfi_kif *kif __attribute__((aligned(8)));
|
|
#else /* !KERNEL */
|
|
void *kif __attribute__((aligned(8)));
|
|
#endif /* !KERNEL */
|
|
struct pf_anchor *anchor __attribute__((aligned(8)));
|
|
#ifdef KERNEL
|
|
struct pfr_ktable *overload_tbl __attribute__((aligned(8)));
|
|
#else /* !KERNEL */
|
|
void *overload_tbl __attribute__((aligned(8)));
|
|
#endif /* !KERNEL */
|
|
|
|
pf_osfp_t os_fingerprint __attribute__((aligned(8)));
|
|
|
|
unsigned int rtableid;
|
|
u_int32_t timeout[PFTM_MAX];
|
|
u_int32_t states;
|
|
u_int32_t max_states;
|
|
u_int32_t src_nodes;
|
|
u_int32_t max_src_nodes;
|
|
u_int32_t max_src_states;
|
|
u_int32_t max_src_conn;
|
|
struct {
|
|
u_int32_t limit;
|
|
u_int32_t seconds;
|
|
} max_src_conn_rate;
|
|
u_int32_t qid;
|
|
u_int32_t pqid;
|
|
u_int32_t rt_listid;
|
|
u_int32_t nr;
|
|
u_int32_t prob;
|
|
uid_t cuid;
|
|
pid_t cpid;
|
|
|
|
u_int16_t return_icmp;
|
|
u_int16_t return_icmp6;
|
|
u_int16_t max_mss;
|
|
u_int16_t tag;
|
|
u_int16_t match_tag;
|
|
|
|
struct pf_rule_uid uid;
|
|
struct pf_rule_gid gid;
|
|
|
|
u_int32_t rule_flag;
|
|
u_int8_t action;
|
|
u_int8_t direction;
|
|
u_int8_t log;
|
|
u_int8_t logif;
|
|
u_int8_t quick;
|
|
u_int8_t ifnot;
|
|
u_int8_t match_tag_not;
|
|
u_int8_t natpass;
|
|
|
|
#define PF_STATE_NORMAL 0x1
|
|
#define PF_STATE_MODULATE 0x2
|
|
#define PF_STATE_SYNPROXY 0x3
|
|
u_int8_t keep_state;
|
|
sa_family_t af;
|
|
u_int8_t proto;
|
|
u_int8_t type;
|
|
u_int8_t code;
|
|
u_int8_t flags;
|
|
u_int8_t flagset;
|
|
u_int8_t min_ttl;
|
|
u_int8_t allow_opts;
|
|
u_int8_t rt;
|
|
u_int8_t return_ttl;
|
|
|
|
/* service class categories */
|
|
#define SCIDX_MASK 0x0f
|
|
#define SC_BE 0x10
|
|
#define SC_BK_SYS 0x11
|
|
#define SC_BK 0x12
|
|
#define SC_RD 0x13
|
|
#define SC_OAM 0x14
|
|
#define SC_AV 0x15
|
|
#define SC_RV 0x16
|
|
#define SC_VI 0x17
|
|
#define SC_VO 0x18
|
|
#define SC_CTL 0x19
|
|
|
|
/* diffserve code points */
|
|
#define DSCP_MASK 0xfc
|
|
#define DSCP_CUMASK 0x03
|
|
#define DSCP_EF 0xb8
|
|
#define DSCP_AF11 0x28
|
|
#define DSCP_AF12 0x30
|
|
#define DSCP_AF13 0x38
|
|
#define DSCP_AF21 0x48
|
|
#define DSCP_AF22 0x50
|
|
#define DSCP_AF23 0x58
|
|
#define DSCP_AF31 0x68
|
|
#define DSCP_AF32 0x70
|
|
#define DSCP_AF33 0x78
|
|
#define DSCP_AF41 0x88
|
|
#define DSCP_AF42 0x90
|
|
#define DSCP_AF43 0x98
|
|
#define AF_CLASSMASK 0xe0
|
|
#define AF_DROPPRECMASK 0x18
|
|
u_int8_t tos;
|
|
u_int8_t anchor_relative;
|
|
u_int8_t anchor_wildcard;
|
|
|
|
#define PF_FLUSH 0x01
|
|
#define PF_FLUSH_GLOBAL 0x02
|
|
u_int8_t flush;
|
|
|
|
u_int8_t proto_variant;
|
|
u_int8_t extfilter; /* Filter mode [PF_EXTFILTER_xxx] */
|
|
u_int8_t extmap; /* Mapping mode [PF_EXTMAP_xxx] */
|
|
u_int32_t dnpipe;
|
|
u_int32_t dntype;
|
|
};
|
|
|
|
/* pf device identifiers */
|
|
#define PFDEV_PF 0
|
|
#define PFDEV_PFM 1
|
|
#define PFDEV_MAX 2
|
|
|
|
/* rule flags */
|
|
#define PFRULE_DROP 0x0000
|
|
#define PFRULE_RETURNRST 0x0001
|
|
#define PFRULE_FRAGMENT 0x0002
|
|
#define PFRULE_RETURNICMP 0x0004
|
|
#define PFRULE_RETURN 0x0008
|
|
#define PFRULE_NOSYNC 0x0010
|
|
#define PFRULE_SRCTRACK 0x0020 /* track source states */
|
|
#define PFRULE_RULESRCTRACK 0x0040 /* per rule */
|
|
|
|
/* scrub flags */
|
|
#define PFRULE_NODF 0x0100
|
|
#define PFRULE_FRAGCROP 0x0200 /* non-buffering frag cache */
|
|
#define PFRULE_FRAGDROP 0x0400 /* drop funny fragments */
|
|
#define PFRULE_RANDOMID 0x0800
|
|
#define PFRULE_REASSEMBLE_TCP 0x1000
|
|
|
|
/* rule flags for TOS/DSCP/service class differentiation */
|
|
#define PFRULE_TOS 0x2000
|
|
#define PFRULE_DSCP 0x4000
|
|
#define PFRULE_SC 0x8000
|
|
|
|
/* rule flags again */
|
|
#define PFRULE_IFBOUND 0x00010000 /* if-bound */
|
|
#define PFRULE_PFM 0x00020000 /* created by pfm device */
|
|
|
|
#define PFSTATE_HIWAT 10000 /* default state table size */
|
|
#define PFSTATE_ADAPT_START 6000 /* default adaptive timeout start */
|
|
#define PFSTATE_ADAPT_END 12000 /* default adaptive timeout end */
|
|
|
|
#define PFAPPSTATE_HIWAT 10000 /* default same as state table */
|
|
|
|
enum pf_extmap {
|
|
PF_EXTMAP_APD = 1, /* Address-port-dependent mapping */
|
|
PF_EXTMAP_AD, /* Address-dependent mapping */
|
|
PF_EXTMAP_EI /* Endpoint-independent mapping */
|
|
};
|
|
|
|
enum pf_extfilter {
|
|
PF_EXTFILTER_APD = 1, /* Address-port-dependent filtering */
|
|
PF_EXTFILTER_AD, /* Address-dependent filtering */
|
|
PF_EXTFILTER_EI /* Endpoint-independent filtering */
|
|
};
|
|
|
|
struct pf_threshold {
|
|
u_int32_t limit;
|
|
#define PF_THRESHOLD_MULT 1000
|
|
#define PF_THRESHOLD_MAX 0xffffffff / PF_THRESHOLD_MULT
|
|
u_int32_t seconds;
|
|
u_int32_t count;
|
|
u_int32_t last;
|
|
};
|
|
|
|
struct pf_src_node {
|
|
RB_ENTRY(pf_src_node) entry;
|
|
struct pf_addr addr;
|
|
struct pf_addr raddr;
|
|
union pf_rule_ptr rule;
|
|
#ifdef KERNEL
|
|
struct pfi_kif *kif;
|
|
#else /* !KERNEL */
|
|
void *kif;
|
|
#endif /* !KERNEL */
|
|
u_int64_t bytes[2];
|
|
u_int64_t packets[2];
|
|
u_int32_t states;
|
|
u_int32_t conn;
|
|
struct pf_threshold conn_rate;
|
|
u_int64_t creation;
|
|
u_int64_t expire;
|
|
sa_family_t af;
|
|
u_int8_t ruletype;
|
|
};
|
|
|
|
#define PFSNODE_HIWAT 10000 /* default source node table size */
|
|
|
|
#ifdef KERNEL
|
|
struct pf_state_scrub {
|
|
struct timeval pfss_last; /* time received last packet */
|
|
u_int32_t pfss_tsecr; /* last echoed timestamp */
|
|
u_int32_t pfss_tsval; /* largest timestamp */
|
|
u_int32_t pfss_tsval0; /* original timestamp */
|
|
u_int16_t pfss_flags;
|
|
#define PFSS_TIMESTAMP 0x0001 /* modulate timestamp */
|
|
#define PFSS_PAWS 0x0010 /* stricter PAWS checks */
|
|
#define PFSS_PAWS_IDLED 0x0020 /* was idle too long. no PAWS */
|
|
#define PFSS_DATA_TS 0x0040 /* timestamp on data packets */
|
|
#define PFSS_DATA_NOTS 0x0080 /* no timestamp on data packets */
|
|
u_int8_t pfss_ttl; /* stashed TTL */
|
|
u_int8_t pad;
|
|
u_int32_t pfss_ts_mod; /* timestamp modulation */
|
|
};
|
|
#endif /* KERNEL */
|
|
|
|
union pf_state_xport {
|
|
u_int16_t port;
|
|
u_int16_t call_id;
|
|
u_int32_t spi;
|
|
};
|
|
|
|
struct pf_state_host {
|
|
struct pf_addr addr;
|
|
union pf_state_xport xport;
|
|
};
|
|
|
|
#ifdef KERNEL
|
|
struct pf_state_peer {
|
|
u_int32_t seqlo; /* Max sequence number sent */
|
|
u_int32_t seqhi; /* Max the other end ACKd + win */
|
|
u_int32_t seqdiff; /* Sequence number modulator */
|
|
u_int16_t max_win; /* largest window (pre scaling) */
|
|
u_int8_t state; /* active state level */
|
|
u_int8_t wscale; /* window scaling factor */
|
|
u_int16_t mss; /* Maximum segment size option */
|
|
u_int8_t tcp_est; /* Did we reach TCPS_ESTABLISHED */
|
|
struct pf_state_scrub *scrub; /* state is scrubbed */
|
|
u_int8_t pad[3];
|
|
};
|
|
|
|
TAILQ_HEAD(pf_state_queue, pf_state);
|
|
|
|
struct pf_state;
|
|
struct pf_pdesc;
|
|
struct pf_app_state;
|
|
|
|
typedef void (*pf_app_handler)(struct pf_state *, int, int, struct pf_pdesc *,
|
|
struct pfi_kif *);
|
|
|
|
typedef int (*pf_app_compare)(struct pf_app_state *, struct pf_app_state *);
|
|
|
|
struct pf_pptp_state {
|
|
struct pf_state *grev1_state;
|
|
};
|
|
|
|
struct pf_grev1_state {
|
|
struct pf_state *pptp_state;
|
|
};
|
|
|
|
struct pf_ike_state {
|
|
u_int64_t cookie;
|
|
};
|
|
|
|
struct pf_app_state {
|
|
pf_app_handler handler;
|
|
pf_app_compare compare_lan_ext;
|
|
pf_app_compare compare_ext_gwy;
|
|
union {
|
|
struct pf_pptp_state pptp;
|
|
struct pf_grev1_state grev1;
|
|
struct pf_ike_state ike;
|
|
} u;
|
|
};
|
|
|
|
/* keep synced with struct pf_state, used in RB_FIND */
|
|
struct pf_state_key_cmp {
|
|
struct pf_state_host lan;
|
|
struct pf_state_host gwy;
|
|
struct pf_state_host ext_lan;
|
|
struct pf_state_host ext_gwy;
|
|
sa_family_t af_lan;
|
|
sa_family_t af_gwy;
|
|
u_int8_t proto;
|
|
u_int8_t direction;
|
|
u_int8_t proto_variant;
|
|
struct pf_app_state *app_state;
|
|
};
|
|
|
|
TAILQ_HEAD(pf_statelist, pf_state);
|
|
|
|
struct pf_state_key {
|
|
struct pf_state_host lan;
|
|
struct pf_state_host gwy;
|
|
struct pf_state_host ext_lan;
|
|
struct pf_state_host ext_gwy;
|
|
sa_family_t af_lan;
|
|
sa_family_t af_gwy;
|
|
u_int8_t proto;
|
|
u_int8_t direction;
|
|
u_int8_t proto_variant;
|
|
struct pf_app_state *app_state;
|
|
u_int32_t flowsrc;
|
|
u_int32_t flowhash;
|
|
|
|
RB_ENTRY(pf_state_key) entry_lan_ext;
|
|
RB_ENTRY(pf_state_key) entry_ext_gwy;
|
|
struct pf_statelist states;
|
|
u_int32_t refcnt;
|
|
};
|
|
|
|
|
|
/* keep synced with struct pf_state, used in RB_FIND */
|
|
struct pf_state_cmp {
|
|
u_int64_t id;
|
|
u_int32_t creatorid;
|
|
u_int32_t pad;
|
|
};
|
|
|
|
/* flowhash key (12-bytes multiple for performance) */
|
|
struct pf_flowhash_key {
|
|
struct pf_state_host ap1; /* address+port blob 1 */
|
|
struct pf_state_host ap2; /* address+port blob 2 */
|
|
u_int32_t af;
|
|
u_int32_t proto;
|
|
};
|
|
#endif /* KERNEL */
|
|
|
|
struct hook_desc;
|
|
TAILQ_HEAD(hook_desc_head, hook_desc);
|
|
|
|
#ifdef KERNEL
|
|
struct pf_state {
|
|
u_int64_t id;
|
|
u_int32_t creatorid;
|
|
u_int32_t pad;
|
|
|
|
TAILQ_ENTRY(pf_state) entry_list;
|
|
TAILQ_ENTRY(pf_state) next;
|
|
RB_ENTRY(pf_state) entry_id;
|
|
struct pf_state_peer src;
|
|
struct pf_state_peer dst;
|
|
union pf_rule_ptr rule;
|
|
union pf_rule_ptr anchor;
|
|
union pf_rule_ptr nat_rule;
|
|
struct pf_addr rt_addr;
|
|
struct hook_desc_head unlink_hooks;
|
|
struct pf_state_key *state_key;
|
|
struct pfi_kif *kif;
|
|
struct pfi_kif *rt_kif;
|
|
struct pf_src_node *src_node;
|
|
struct pf_src_node *nat_src_node;
|
|
u_int64_t packets[2];
|
|
u_int64_t bytes[2];
|
|
u_int64_t creation;
|
|
u_int64_t expire;
|
|
u_int64_t pfsync_time;
|
|
u_int16_t tag;
|
|
u_int8_t log;
|
|
u_int8_t allow_opts;
|
|
u_int8_t timeout;
|
|
u_int8_t sync_flags;
|
|
};
|
|
#endif /* KERNEL */
|
|
|
|
#define PFSTATE_NOSYNC 0x01
|
|
#define PFSTATE_FROMSYNC 0x02
|
|
#define PFSTATE_STALE 0x04
|
|
|
|
#define __packed __attribute__((__packed__))
|
|
|
|
/*
|
|
* Unified state structures for pulling states out of the kernel
|
|
* used by pfsync(4) and the pf(4) ioctl.
|
|
*/
|
|
struct pfsync_state_scrub {
|
|
u_int16_t pfss_flags;
|
|
u_int8_t pfss_ttl; /* stashed TTL */
|
|
#define PFSYNC_SCRUB_FLAG_VALID 0x01
|
|
u_int8_t scrub_flag;
|
|
u_int32_t pfss_ts_mod; /* timestamp modulation */
|
|
} __packed;
|
|
|
|
struct pfsync_state_host {
|
|
struct pf_addr addr;
|
|
union pf_state_xport xport;
|
|
u_int16_t pad[2];
|
|
} __packed;
|
|
|
|
struct pfsync_state_peer {
|
|
struct pfsync_state_scrub scrub; /* state is scrubbed */
|
|
u_int32_t seqlo; /* Max sequence number sent */
|
|
u_int32_t seqhi; /* Max the other end ACKd + win */
|
|
u_int32_t seqdiff; /* Sequence number modulator */
|
|
u_int16_t max_win; /* largest window (pre scaling) */
|
|
u_int16_t mss; /* Maximum segment size option */
|
|
u_int8_t state; /* active state level */
|
|
u_int8_t wscale; /* window scaling factor */
|
|
u_int8_t pad[6];
|
|
} __packed;
|
|
|
|
struct pfsync_state {
|
|
u_int32_t id[2];
|
|
char ifname[IFNAMSIZ];
|
|
struct pfsync_state_host lan;
|
|
struct pfsync_state_host gwy;
|
|
struct pfsync_state_host ext_lan;
|
|
struct pfsync_state_host ext_gwy;
|
|
struct pfsync_state_peer src;
|
|
struct pfsync_state_peer dst;
|
|
struct pf_addr rt_addr;
|
|
struct hook_desc_head unlink_hooks;
|
|
#if !defined(__LP64__)
|
|
u_int32_t _pad[2];
|
|
#endif /* !__LP64__ */
|
|
u_int32_t rule;
|
|
u_int32_t anchor;
|
|
u_int32_t nat_rule;
|
|
u_int64_t creation;
|
|
u_int64_t expire;
|
|
u_int32_t packets[2][2];
|
|
u_int32_t bytes[2][2];
|
|
u_int32_t creatorid;
|
|
u_int16_t tag;
|
|
sa_family_t af_lan;
|
|
sa_family_t af_gwy;
|
|
u_int8_t proto;
|
|
u_int8_t direction;
|
|
u_int8_t log;
|
|
u_int8_t allow_opts;
|
|
u_int8_t timeout;
|
|
u_int8_t sync_flags;
|
|
u_int8_t updates;
|
|
u_int8_t proto_variant;
|
|
u_int8_t __pad;
|
|
u_int32_t flowhash;
|
|
} __packed;
|
|
|
|
#define PFSYNC_FLAG_COMPRESS 0x01
|
|
#define PFSYNC_FLAG_STALE 0x02
|
|
#define PFSYNC_FLAG_SRCNODE 0x04
|
|
#define PFSYNC_FLAG_NATSRCNODE 0x08
|
|
|
|
#ifdef KERNEL
|
|
/* for copies to/from userland via pf_ioctl() */
|
|
#define pf_state_peer_to_pfsync(s, d) do { \
|
|
(d)->seqlo = (s)->seqlo; \
|
|
(d)->seqhi = (s)->seqhi; \
|
|
(d)->seqdiff = (s)->seqdiff; \
|
|
(d)->max_win = (s)->max_win; \
|
|
(d)->mss = (s)->mss; \
|
|
(d)->state = (s)->state; \
|
|
(d)->wscale = (s)->wscale; \
|
|
if ((s)->scrub) { \
|
|
(d)->scrub.pfss_flags = \
|
|
(s)->scrub->pfss_flags & PFSS_TIMESTAMP; \
|
|
(d)->scrub.pfss_ttl = (s)->scrub->pfss_ttl; \
|
|
(d)->scrub.pfss_ts_mod = (s)->scrub->pfss_ts_mod; \
|
|
(d)->scrub.scrub_flag = PFSYNC_SCRUB_FLAG_VALID; \
|
|
} \
|
|
} while (0)
|
|
|
|
#define pf_state_peer_from_pfsync(s, d) do { \
|
|
(d)->seqlo = (s)->seqlo; \
|
|
(d)->seqhi = (s)->seqhi; \
|
|
(d)->seqdiff = (s)->seqdiff; \
|
|
(d)->max_win = (s)->max_win; \
|
|
(d)->mss = ntohs((s)->mss); \
|
|
(d)->state = (s)->state; \
|
|
(d)->wscale = (s)->wscale; \
|
|
if ((s)->scrub.scrub_flag == PFSYNC_SCRUB_FLAG_VALID && \
|
|
(d)->scrub != NULL) { \
|
|
(d)->scrub->pfss_flags = \
|
|
ntohs((s)->scrub.pfss_flags) & PFSS_TIMESTAMP; \
|
|
(d)->scrub->pfss_ttl = (s)->scrub.pfss_ttl; \
|
|
(d)->scrub->pfss_ts_mod = (s)->scrub.pfss_ts_mod; \
|
|
} \
|
|
} while (0)
|
|
#endif /* KERNEL */
|
|
|
|
#define pf_state_counter_to_pfsync(s, d) do { \
|
|
d[0] = (s>>32)&0xffffffff; \
|
|
d[1] = s&0xffffffff; \
|
|
} while (0)
|
|
|
|
#define pf_state_counter_from_pfsync(s) \
|
|
(((u_int64_t)(s[0])<<32) | (u_int64_t)(s[1]))
|
|
|
|
|
|
|
|
TAILQ_HEAD(pf_rulequeue, pf_rule);
|
|
|
|
struct pf_anchor;
|
|
|
|
struct pf_ruleset {
|
|
struct {
|
|
struct pf_rulequeue queues[2];
|
|
struct {
|
|
struct pf_rulequeue *ptr;
|
|
struct pf_rule **ptr_array;
|
|
u_int32_t rcount;
|
|
u_int32_t ticket;
|
|
int open;
|
|
} active, inactive;
|
|
} rules[PF_RULESET_MAX];
|
|
struct pf_anchor *anchor;
|
|
u_int32_t tticket;
|
|
int tables;
|
|
int topen;
|
|
};
|
|
|
|
RB_HEAD(pf_anchor_global, pf_anchor);
|
|
RB_HEAD(pf_anchor_node, pf_anchor);
|
|
struct pf_anchor {
|
|
RB_ENTRY(pf_anchor) entry_global;
|
|
RB_ENTRY(pf_anchor) entry_node;
|
|
struct pf_anchor *parent;
|
|
struct pf_anchor_node children;
|
|
char name[PF_ANCHOR_NAME_SIZE];
|
|
char path[MAXPATHLEN];
|
|
struct pf_ruleset ruleset;
|
|
int refcnt; /* anchor rules */
|
|
int match;
|
|
char owner[PF_OWNER_NAME_SIZE];
|
|
};
|
|
#ifdef KERNEL
|
|
RB_PROTOTYPE_SC(__private_extern__, pf_anchor_global, pf_anchor, entry_global,
|
|
pf_anchor_compare);
|
|
RB_PROTOTYPE_SC(__private_extern__, pf_anchor_node, pf_anchor, entry_node,
|
|
pf_anchor_compare);
|
|
#else /* !KERNEL */
|
|
RB_PROTOTYPE(pf_anchor_global, pf_anchor, entry_global, pf_anchor_compare);
|
|
RB_PROTOTYPE(pf_anchor_node, pf_anchor, entry_node, pf_anchor_compare);
|
|
#endif /* !KERNEL */
|
|
|
|
#define PF_RESERVED_ANCHOR "_pf"
|
|
|
|
#define PFR_TFLAG_PERSIST 0x00000001
|
|
#define PFR_TFLAG_CONST 0x00000002
|
|
#define PFR_TFLAG_ACTIVE 0x00000004
|
|
#define PFR_TFLAG_INACTIVE 0x00000008
|
|
#define PFR_TFLAG_REFERENCED 0x00000010
|
|
#define PFR_TFLAG_REFDANCHOR 0x00000020
|
|
#define PFR_TFLAG_USRMASK 0x00000003
|
|
#define PFR_TFLAG_SETMASK 0x0000003C
|
|
#define PFR_TFLAG_ALLMASK 0x0000003F
|
|
|
|
struct pfr_table {
|
|
char pfrt_anchor[MAXPATHLEN];
|
|
char pfrt_name[PF_TABLE_NAME_SIZE];
|
|
u_int32_t pfrt_flags;
|
|
u_int8_t pfrt_fback;
|
|
};
|
|
|
|
enum { PFR_FB_NONE, PFR_FB_MATCH, PFR_FB_ADDED, PFR_FB_DELETED,
|
|
PFR_FB_CHANGED, PFR_FB_CLEARED, PFR_FB_DUPLICATE,
|
|
PFR_FB_NOTMATCH, PFR_FB_CONFLICT, PFR_FB_MAX };
|
|
|
|
struct pfr_addr {
|
|
union {
|
|
struct in_addr _pfra_ip4addr;
|
|
struct in6_addr _pfra_ip6addr;
|
|
} pfra_u;
|
|
u_int8_t pfra_af;
|
|
u_int8_t pfra_net;
|
|
u_int8_t pfra_not;
|
|
u_int8_t pfra_fback;
|
|
};
|
|
#define pfra_ip4addr pfra_u._pfra_ip4addr
|
|
#define pfra_ip6addr pfra_u._pfra_ip6addr
|
|
|
|
enum { PFR_DIR_IN, PFR_DIR_OUT, PFR_DIR_MAX };
|
|
enum { PFR_OP_BLOCK, PFR_OP_PASS, PFR_OP_ADDR_MAX, PFR_OP_TABLE_MAX };
|
|
#define PFR_OP_XPASS PFR_OP_ADDR_MAX
|
|
|
|
struct pfr_astats {
|
|
struct pfr_addr pfras_a;
|
|
#if !defined(__LP64__)
|
|
u_int32_t _pad;
|
|
#endif /* !__LP64__ */
|
|
u_int64_t pfras_packets[PFR_DIR_MAX][PFR_OP_ADDR_MAX];
|
|
u_int64_t pfras_bytes[PFR_DIR_MAX][PFR_OP_ADDR_MAX];
|
|
u_int64_t pfras_tzero;
|
|
};
|
|
|
|
enum { PFR_REFCNT_RULE, PFR_REFCNT_ANCHOR, PFR_REFCNT_MAX };
|
|
|
|
struct pfr_tstats {
|
|
struct pfr_table pfrts_t;
|
|
u_int64_t pfrts_packets[PFR_DIR_MAX][PFR_OP_TABLE_MAX];
|
|
u_int64_t pfrts_bytes[PFR_DIR_MAX][PFR_OP_TABLE_MAX];
|
|
u_int64_t pfrts_match;
|
|
u_int64_t pfrts_nomatch;
|
|
u_int64_t pfrts_tzero;
|
|
int pfrts_cnt;
|
|
int pfrts_refcnt[PFR_REFCNT_MAX];
|
|
#if !defined(__LP64__)
|
|
u_int32_t _pad;
|
|
#endif /* !__LP64__ */
|
|
};
|
|
#define pfrts_name pfrts_t.pfrt_name
|
|
#define pfrts_flags pfrts_t.pfrt_flags
|
|
|
|
#ifdef KERNEL
|
|
SLIST_HEAD(pfr_kentryworkq, pfr_kentry);
|
|
struct pfr_kentry {
|
|
struct radix_node pfrke_node[2];
|
|
union sockaddr_union pfrke_sa;
|
|
u_int64_t pfrke_packets[PFR_DIR_MAX][PFR_OP_ADDR_MAX];
|
|
u_int64_t pfrke_bytes[PFR_DIR_MAX][PFR_OP_ADDR_MAX];
|
|
SLIST_ENTRY(pfr_kentry) pfrke_workq;
|
|
u_int64_t pfrke_tzero;
|
|
u_int8_t pfrke_af;
|
|
u_int8_t pfrke_net;
|
|
u_int8_t pfrke_not;
|
|
u_int8_t pfrke_mark;
|
|
u_int8_t pfrke_intrpool;
|
|
};
|
|
|
|
SLIST_HEAD(pfr_ktableworkq, pfr_ktable);
|
|
RB_HEAD(pfr_ktablehead, pfr_ktable);
|
|
struct pfr_ktable {
|
|
struct pfr_tstats pfrkt_ts;
|
|
RB_ENTRY(pfr_ktable) pfrkt_tree;
|
|
SLIST_ENTRY(pfr_ktable) pfrkt_workq;
|
|
struct radix_node_head *pfrkt_ip4;
|
|
struct radix_node_head *pfrkt_ip6;
|
|
struct pfr_ktable *pfrkt_shadow;
|
|
struct pfr_ktable *pfrkt_root;
|
|
struct pf_ruleset *pfrkt_rs;
|
|
u_int64_t pfrkt_larg;
|
|
u_int32_t pfrkt_nflags;
|
|
};
|
|
#define pfrkt_t pfrkt_ts.pfrts_t
|
|
#define pfrkt_name pfrkt_t.pfrt_name
|
|
#define pfrkt_anchor pfrkt_t.pfrt_anchor
|
|
#define pfrkt_ruleset pfrkt_t.pfrt_ruleset
|
|
#define pfrkt_flags pfrkt_t.pfrt_flags
|
|
#define pfrkt_cnt pfrkt_ts.pfrts_cnt
|
|
#define pfrkt_refcnt pfrkt_ts.pfrts_refcnt
|
|
#define pfrkt_packets pfrkt_ts.pfrts_packets
|
|
#define pfrkt_bytes pfrkt_ts.pfrts_bytes
|
|
#define pfrkt_match pfrkt_ts.pfrts_match
|
|
#define pfrkt_nomatch pfrkt_ts.pfrts_nomatch
|
|
#define pfrkt_tzero pfrkt_ts.pfrts_tzero
|
|
|
|
RB_HEAD(pf_state_tree_lan_ext, pf_state_key);
|
|
RB_PROTOTYPE_SC(__private_extern__, pf_state_tree_lan_ext, pf_state_key,
|
|
entry_lan_ext, pf_state_compare_lan_ext);
|
|
|
|
RB_HEAD(pf_state_tree_ext_gwy, pf_state_key);
|
|
RB_PROTOTYPE_SC(__private_extern__, pf_state_tree_ext_gwy, pf_state_key,
|
|
entry_ext_gwy, pf_state_compare_ext_gwy);
|
|
|
|
RB_HEAD(pfi_ifhead, pfi_kif);
|
|
|
|
/* state tables */
|
|
extern struct pf_state_tree_lan_ext pf_statetbl_lan_ext;
|
|
extern struct pf_state_tree_ext_gwy pf_statetbl_ext_gwy;
|
|
|
|
/* keep synced with pfi_kif, used in RB_FIND */
|
|
struct pfi_kif_cmp {
|
|
char pfik_name[IFNAMSIZ];
|
|
};
|
|
|
|
struct pfi_kif {
|
|
char pfik_name[IFNAMSIZ];
|
|
RB_ENTRY(pfi_kif) pfik_tree;
|
|
u_int64_t pfik_packets[2][2][2];
|
|
u_int64_t pfik_bytes[2][2][2];
|
|
u_int64_t pfik_tzero;
|
|
int pfik_flags;
|
|
void *pfik_ah_cookie;
|
|
struct ifnet *pfik_ifp;
|
|
int pfik_states;
|
|
int pfik_rules;
|
|
TAILQ_HEAD(, pfi_dynaddr) pfik_dynaddrs;
|
|
};
|
|
|
|
enum pfi_kif_refs {
|
|
PFI_KIF_REF_NONE,
|
|
PFI_KIF_REF_STATE,
|
|
PFI_KIF_REF_RULE
|
|
};
|
|
|
|
struct pfi_uif {
|
|
#else /* !KERNEL */
|
|
struct pfi_kif {
|
|
#endif /* !KERNEL */
|
|
char pfik_name[IFNAMSIZ];
|
|
u_int64_t pfik_packets[2][2][2];
|
|
u_int64_t pfik_bytes[2][2][2];
|
|
u_int64_t pfik_tzero;
|
|
int pfik_flags;
|
|
int pfik_states;
|
|
int pfik_rules;
|
|
#if !defined(__LP64__)
|
|
u_int32_t _pad;
|
|
#endif /* !__LP64__ */
|
|
};
|
|
|
|
#define PFI_IFLAG_SKIP 0x0100 /* skip filtering on interface */
|
|
|
|
#ifdef KERNEL
|
|
struct pf_pdesc {
|
|
struct {
|
|
int done;
|
|
uid_t uid;
|
|
gid_t gid;
|
|
pid_t pid;
|
|
} lookup;
|
|
u_int64_t tot_len; /* Make Mickey money */
|
|
union {
|
|
struct tcphdr *tcp;
|
|
struct udphdr *udp;
|
|
struct icmp *icmp;
|
|
#if INET6
|
|
struct icmp6_hdr *icmp6;
|
|
#endif /* INET6 */
|
|
struct pf_grev1_hdr *grev1;
|
|
struct pf_esp_hdr *esp;
|
|
void *any;
|
|
} hdr;
|
|
|
|
/* XXX TODO: Change baddr and naddr to *saddr */
|
|
struct pf_addr baddr; /* src address before translation */
|
|
struct pf_addr bdaddr; /* dst address before translation */
|
|
struct pf_addr naddr; /* src address after translation */
|
|
struct pf_addr ndaddr; /* dst address after translation */
|
|
struct pf_rule *nat_rule; /* nat/rdr rule applied to packet */
|
|
struct pf_addr *src;
|
|
struct pf_addr *dst;
|
|
struct ether_header
|
|
*eh;
|
|
struct mbuf *mp;
|
|
int lmw; /* lazy writable offset */
|
|
struct pf_mtag *pf_mtag;
|
|
u_int16_t *ip_sum;
|
|
u_int32_t off; /* protocol header offset */
|
|
u_int32_t hdrlen; /* protocol header length */
|
|
u_int32_t p_len; /* total length of payload */
|
|
u_int16_t flags; /* Let SCRUB trigger behavior in */
|
|
/* state code. Easier than tags */
|
|
#define PFDESC_TCP_NORM 0x0001 /* TCP shall be statefully scrubbed */
|
|
#define PFDESC_IP_REAS 0x0002 /* IP frags would've been reassembled */
|
|
#define PFDESC_IP_FRAG 0x0004 /* This is a fragment */
|
|
sa_family_t af;
|
|
sa_family_t naf; /* address family after translation */
|
|
u_int8_t proto;
|
|
u_int8_t tos;
|
|
u_int8_t ttl;
|
|
u_int8_t proto_variant;
|
|
mbuf_svc_class_t sc; /* mbuf service class (MBUF_SVC) */
|
|
u_int32_t pktflags; /* mbuf packet flags (PKTF) */
|
|
u_int32_t flowsrc; /* flow source (FLOWSRC) */
|
|
u_int32_t flowhash; /* flow hash to identify the sender */
|
|
};
|
|
#endif /* KERNEL */
|
|
|
|
/* flags for RDR options */
|
|
#define PF_DPORT_RANGE 0x01 /* Dest port uses range */
|
|
#define PF_RPORT_RANGE 0x02 /* RDR'ed port uses range */
|
|
|
|
/* Reasons code for passing/dropping a packet */
|
|
#define PFRES_MATCH 0 /* Explicit match of a rule */
|
|
#define PFRES_BADOFF 1 /* Bad offset for pull_hdr */
|
|
#define PFRES_FRAG 2 /* Dropping following fragment */
|
|
#define PFRES_SHORT 3 /* Dropping short packet */
|
|
#define PFRES_NORM 4 /* Dropping by normalizer */
|
|
#define PFRES_MEMORY 5 /* Dropped due to lacking mem */
|
|
#define PFRES_TS 6 /* Bad TCP Timestamp (RFC1323) */
|
|
#define PFRES_CONGEST 7 /* Congestion (of ipintrq) */
|
|
#define PFRES_IPOPTIONS 8 /* IP option */
|
|
#define PFRES_PROTCKSUM 9 /* Protocol checksum invalid */
|
|
#define PFRES_BADSTATE 10 /* State mismatch */
|
|
#define PFRES_STATEINS 11 /* State insertion failure */
|
|
#define PFRES_MAXSTATES 12 /* State limit */
|
|
#define PFRES_SRCLIMIT 13 /* Source node/conn limit */
|
|
#define PFRES_SYNPROXY 14 /* SYN proxy */
|
|
#define PFRES_DUMMYNET 15 /* Dummynet */
|
|
#define PFRES_MAX 16 /* total+1 */
|
|
|
|
#define PFRES_NAMES { \
|
|
"match", \
|
|
"bad-offset", \
|
|
"fragment", \
|
|
"short", \
|
|
"normalize", \
|
|
"memory", \
|
|
"bad-timestamp", \
|
|
"congestion", \
|
|
"ip-option", \
|
|
"proto-cksum", \
|
|
"state-mismatch", \
|
|
"state-insert", \
|
|
"state-limit", \
|
|
"src-limit", \
|
|
"synproxy", \
|
|
"dummynet", \
|
|
NULL \
|
|
}
|
|
|
|
/* Counters for other things we want to keep track of */
|
|
#define LCNT_STATES 0 /* states */
|
|
#define LCNT_SRCSTATES 1 /* max-src-states */
|
|
#define LCNT_SRCNODES 2 /* max-src-nodes */
|
|
#define LCNT_SRCCONN 3 /* max-src-conn */
|
|
#define LCNT_SRCCONNRATE 4 /* max-src-conn-rate */
|
|
#define LCNT_OVERLOAD_TABLE 5 /* entry added to overload table */
|
|
#define LCNT_OVERLOAD_FLUSH 6 /* state entries flushed */
|
|
#define LCNT_MAX 7 /* total+1 */
|
|
|
|
#define LCNT_NAMES { \
|
|
"max states per rule", \
|
|
"max-src-states", \
|
|
"max-src-nodes", \
|
|
"max-src-conn", \
|
|
"max-src-conn-rate", \
|
|
"overload table insertion", \
|
|
"overload flush states", \
|
|
NULL \
|
|
}
|
|
|
|
/* UDP state enumeration */
|
|
#define PFUDPS_NO_TRAFFIC 0
|
|
#define PFUDPS_SINGLE 1
|
|
#define PFUDPS_MULTIPLE 2
|
|
|
|
#define PFUDPS_NSTATES 3 /* number of state levels */
|
|
|
|
#define PFUDPS_NAMES { \
|
|
"NO_TRAFFIC", \
|
|
"SINGLE", \
|
|
"MULTIPLE", \
|
|
NULL \
|
|
}
|
|
|
|
/* GREv1 protocol state enumeration */
|
|
#define PFGRE1S_NO_TRAFFIC 0
|
|
#define PFGRE1S_INITIATING 1
|
|
#define PFGRE1S_ESTABLISHED 2
|
|
|
|
#define PFGRE1S_NSTATES 3 /* number of state levels */
|
|
|
|
#define PFGRE1S_NAMES { \
|
|
"NO_TRAFFIC", \
|
|
"INITIATING", \
|
|
"ESTABLISHED", \
|
|
NULL \
|
|
}
|
|
|
|
#define PFESPS_NO_TRAFFIC 0
|
|
#define PFESPS_INITIATING 1
|
|
#define PFESPS_ESTABLISHED 2
|
|
|
|
#define PFESPS_NSTATES 3 /* number of state levels */
|
|
|
|
#define PFESPS_NAMES { "NO_TRAFFIC", "INITIATING", "ESTABLISHED", NULL }
|
|
|
|
/* Other protocol state enumeration */
|
|
#define PFOTHERS_NO_TRAFFIC 0
|
|
#define PFOTHERS_SINGLE 1
|
|
#define PFOTHERS_MULTIPLE 2
|
|
|
|
#define PFOTHERS_NSTATES 3 /* number of state levels */
|
|
|
|
#define PFOTHERS_NAMES { \
|
|
"NO_TRAFFIC", \
|
|
"SINGLE", \
|
|
"MULTIPLE", \
|
|
NULL \
|
|
}
|
|
|
|
#define FCNT_STATE_SEARCH 0
|
|
#define FCNT_STATE_INSERT 1
|
|
#define FCNT_STATE_REMOVALS 2
|
|
#define FCNT_MAX 3
|
|
|
|
#define SCNT_SRC_NODE_SEARCH 0
|
|
#define SCNT_SRC_NODE_INSERT 1
|
|
#define SCNT_SRC_NODE_REMOVALS 2
|
|
#define SCNT_MAX 3
|
|
|
|
#ifdef KERNEL
|
|
#define ACTION_SET(a, x) \
|
|
do { \
|
|
if ((a) != NULL) \
|
|
*(a) = (x); \
|
|
} while (0)
|
|
|
|
#define REASON_SET(a, x) \
|
|
do { \
|
|
if ((a) != NULL) \
|
|
*(a) = (x); \
|
|
if (x < PFRES_MAX) \
|
|
pf_status.counters[x]++; \
|
|
} while (0)
|
|
#endif /* KERNEL */
|
|
|
|
struct pf_status {
|
|
u_int64_t counters[PFRES_MAX];
|
|
u_int64_t lcounters[LCNT_MAX]; /* limit counters */
|
|
u_int64_t fcounters[FCNT_MAX];
|
|
u_int64_t scounters[SCNT_MAX];
|
|
u_int64_t pcounters[2][2][3];
|
|
u_int64_t bcounters[2][2];
|
|
u_int64_t stateid;
|
|
u_int32_t running;
|
|
u_int32_t states;
|
|
u_int32_t src_nodes;
|
|
u_int64_t since __attribute__((aligned(8)));
|
|
u_int32_t debug;
|
|
u_int32_t hostid;
|
|
char ifname[IFNAMSIZ];
|
|
u_int8_t pf_chksum[PF_MD5_DIGEST_LENGTH];
|
|
};
|
|
|
|
struct cbq_opts {
|
|
u_int32_t minburst;
|
|
u_int32_t maxburst;
|
|
u_int32_t pktsize;
|
|
u_int32_t maxpktsize;
|
|
u_int32_t ns_per_byte;
|
|
u_int32_t maxidle;
|
|
int32_t minidle;
|
|
u_int32_t offtime;
|
|
u_int32_t flags;
|
|
};
|
|
|
|
struct priq_opts {
|
|
u_int32_t flags;
|
|
};
|
|
|
|
struct qfq_opts {
|
|
u_int32_t flags;
|
|
u_int32_t lmax;
|
|
};
|
|
|
|
struct hfsc_opts {
|
|
/* real-time service curve */
|
|
u_int64_t rtsc_m1; /* slope of the 1st segment in bps */
|
|
u_int64_t rtsc_d; /* the x-projection of m1 in msec */
|
|
u_int64_t rtsc_m2; /* slope of the 2nd segment in bps */
|
|
u_int32_t rtsc_fl; /* service curve flags */
|
|
#if !defined(__LP64__)
|
|
u_int32_t _pad;
|
|
#endif /* !__LP64__ */
|
|
/* link-sharing service curve */
|
|
u_int64_t lssc_m1;
|
|
u_int64_t lssc_d;
|
|
u_int64_t lssc_m2;
|
|
u_int32_t lssc_fl;
|
|
#if !defined(__LP64__)
|
|
u_int32_t __pad;
|
|
#endif /* !__LP64__ */
|
|
/* upper-limit service curve */
|
|
u_int64_t ulsc_m1;
|
|
u_int64_t ulsc_d;
|
|
u_int64_t ulsc_m2;
|
|
u_int32_t ulsc_fl;
|
|
u_int32_t flags; /* scheduler flags */
|
|
};
|
|
|
|
struct fairq_opts {
|
|
u_int32_t nbuckets; /* hash buckets */
|
|
u_int32_t flags;
|
|
u_int64_t hogs_m1; /* hog detection bandwidth */
|
|
|
|
/* link-sharing service curve */
|
|
u_int64_t lssc_m1;
|
|
u_int64_t lssc_d;
|
|
u_int64_t lssc_m2;
|
|
};
|
|
|
|
/* bandwidth types */
|
|
#define PF_ALTQ_BW_ABSOLUTE 1 /* bw in absolute value (bps) */
|
|
#define PF_ALTQ_BW_PERCENT 2 /* bandwidth in percentage */
|
|
|
|
/* ALTQ rule flags */
|
|
#define PF_ALTQF_TBR 0x1 /* enable Token Bucket Regulator */
|
|
|
|
/* queue rule flags */
|
|
#define PF_ALTQ_QRF_WEIGHT 0x1 /* weight instead of priority */
|
|
|
|
struct pf_altq {
|
|
char ifname[IFNAMSIZ];
|
|
|
|
/* discipline-specific state */
|
|
void *altq_disc __attribute__((aligned(8)));
|
|
TAILQ_ENTRY(pf_altq) entries __attribute__((aligned(8)));
|
|
#if !defined(__LP64__)
|
|
u_int32_t _pad[2];
|
|
#endif /* !__LP64__ */
|
|
|
|
u_int32_t aflags; /* ALTQ rule flags */
|
|
u_int32_t bwtype; /* bandwidth type */
|
|
|
|
/* scheduler spec */
|
|
u_int32_t scheduler; /* scheduler type */
|
|
u_int32_t tbrsize; /* tokenbucket regulator size */
|
|
u_int64_t ifbandwidth; /* interface bandwidth */
|
|
|
|
/* queue spec */
|
|
char qname[PF_QNAME_SIZE]; /* queue name */
|
|
char parent[PF_QNAME_SIZE]; /* parent name */
|
|
u_int32_t parent_qid; /* parent queue id */
|
|
u_int32_t qrflags; /* queue rule flags */
|
|
union {
|
|
u_int32_t priority; /* priority */
|
|
u_int32_t weight; /* weight */
|
|
};
|
|
u_int32_t qlimit; /* queue size limit */
|
|
u_int32_t flags; /* misc flags */
|
|
#if !defined(__LP64__)
|
|
u_int32_t __pad;
|
|
#endif /* !__LP64__ */
|
|
u_int64_t bandwidth; /* queue bandwidth */
|
|
union {
|
|
struct cbq_opts cbq_opts;
|
|
struct priq_opts priq_opts;
|
|
struct hfsc_opts hfsc_opts;
|
|
struct fairq_opts fairq_opts;
|
|
struct qfq_opts qfq_opts;
|
|
} pq_u;
|
|
|
|
u_int32_t qid; /* return value */
|
|
};
|
|
|
|
struct pf_tagname {
|
|
TAILQ_ENTRY(pf_tagname) entries;
|
|
char name[PF_TAG_NAME_SIZE];
|
|
u_int16_t tag;
|
|
int ref;
|
|
};
|
|
|
|
#define PFFRAG_FRENT_HIWAT 5000 /* Number of fragment entries */
|
|
#define PFFRAG_FRAG_HIWAT 1000 /* Number of fragmented packets */
|
|
#define PFFRAG_FRCENT_HIWAT 50000 /* Number of fragment cache entries */
|
|
#define PFFRAG_FRCACHE_HIWAT 10000 /* Number of fragment descriptors */
|
|
|
|
#define PFR_KTABLE_HIWAT 1000 /* Number of tables */
|
|
#define PFR_KENTRY_HIWAT 200000 /* Number of table entries */
|
|
#define PFR_KENTRY_HIWAT_SMALL 100000 /* Number of table entries (tiny hosts) */
|
|
|
|
/*
|
|
* ioctl parameter structures
|
|
*/
|
|
|
|
struct pfioc_pooladdr {
|
|
u_int32_t action;
|
|
u_int32_t ticket;
|
|
u_int32_t nr;
|
|
u_int32_t r_num;
|
|
u_int8_t r_action;
|
|
u_int8_t r_last;
|
|
u_int8_t af;
|
|
char anchor[MAXPATHLEN];
|
|
struct pf_pooladdr addr;
|
|
};
|
|
|
|
struct pfioc_rule {
|
|
u_int32_t action;
|
|
u_int32_t ticket;
|
|
u_int32_t pool_ticket;
|
|
u_int32_t nr;
|
|
char anchor[MAXPATHLEN];
|
|
char anchor_call[MAXPATHLEN];
|
|
struct pf_rule rule;
|
|
};
|
|
|
|
struct pfioc_natlook {
|
|
struct pf_addr saddr;
|
|
struct pf_addr daddr;
|
|
struct pf_addr rsaddr;
|
|
struct pf_addr rdaddr;
|
|
union pf_state_xport sxport;
|
|
union pf_state_xport dxport;
|
|
union pf_state_xport rsxport;
|
|
union pf_state_xport rdxport;
|
|
sa_family_t af;
|
|
u_int8_t proto;
|
|
u_int8_t proto_variant;
|
|
u_int8_t direction;
|
|
};
|
|
|
|
struct pfioc_state {
|
|
struct pfsync_state state;
|
|
};
|
|
|
|
struct pfioc_src_node_kill {
|
|
/* XXX returns the number of src nodes killed in psnk_af */
|
|
sa_family_t psnk_af;
|
|
struct pf_rule_addr psnk_src;
|
|
struct pf_rule_addr psnk_dst;
|
|
};
|
|
|
|
struct pfioc_state_addr_kill {
|
|
struct pf_addr_wrap addr;
|
|
u_int8_t reserved_[3];
|
|
u_int8_t neg;
|
|
union pf_rule_xport xport;
|
|
};
|
|
|
|
struct pfioc_state_kill {
|
|
/* XXX returns the number of states killed in psk_af */
|
|
sa_family_t psk_af;
|
|
u_int8_t psk_proto;
|
|
u_int8_t psk_proto_variant;
|
|
u_int8_t _pad;
|
|
struct pfioc_state_addr_kill psk_src;
|
|
struct pfioc_state_addr_kill psk_dst;
|
|
char psk_ifname[IFNAMSIZ];
|
|
char psk_ownername[PF_OWNER_NAME_SIZE];
|
|
};
|
|
|
|
struct pfioc_states {
|
|
int ps_len;
|
|
union {
|
|
caddr_t psu_buf;
|
|
struct pfsync_state *psu_states;
|
|
} ps_u __attribute__((aligned(8)));
|
|
#define ps_buf ps_u.psu_buf
|
|
#define ps_states ps_u.psu_states
|
|
};
|
|
|
|
#ifdef KERNEL
|
|
struct pfioc_states_32 {
|
|
int ps_len;
|
|
union {
|
|
user32_addr_t psu_buf;
|
|
user32_addr_t psu_states;
|
|
} ps_u __attribute__((aligned(8)));
|
|
};
|
|
|
|
struct pfioc_states_64 {
|
|
int ps_len;
|
|
union {
|
|
user64_addr_t psu_buf;
|
|
user64_addr_t psu_states;
|
|
} ps_u __attribute__((aligned(8)));
|
|
};
|
|
#endif /* KERNEL */
|
|
|
|
#define PFTOK_PROCNAME_LEN 64
|
|
#pragma pack(1)
|
|
struct pfioc_token {
|
|
u_int64_t token_value;
|
|
u_int64_t timestamp;
|
|
pid_t pid;
|
|
char proc_name[PFTOK_PROCNAME_LEN];
|
|
};
|
|
#pragma pack()
|
|
|
|
struct pfioc_kernel_token {
|
|
SLIST_ENTRY(pfioc_kernel_token) next;
|
|
struct pfioc_token token;
|
|
};
|
|
|
|
struct pfioc_remove_token {
|
|
u_int64_t token_value;
|
|
u_int64_t refcount;
|
|
};
|
|
|
|
struct pfioc_tokens {
|
|
int size;
|
|
union {
|
|
caddr_t pgtu_buf;
|
|
struct pfioc_token *pgtu_tokens;
|
|
} pgt_u __attribute__((aligned(8)));
|
|
#define pgt_buf pgt_u.pgtu_buf
|
|
#define pgt_tokens pgt_u.pgtu_tokens
|
|
};
|
|
|
|
#ifdef KERNEL
|
|
struct pfioc_tokens_32 {
|
|
int size;
|
|
union {
|
|
user32_addr_t pgtu_buf;
|
|
user32_addr_t pgtu_tokens;
|
|
} pgt_u __attribute__((aligned(8)));
|
|
};
|
|
|
|
struct pfioc_tokens_64 {
|
|
int size;
|
|
union {
|
|
user64_addr_t pgtu_buf;
|
|
user64_addr_t pgtu_tokens;
|
|
} pgt_u __attribute__((aligned(8)));
|
|
};
|
|
#endif /* KERNEL */
|
|
|
|
|
|
struct pfioc_src_nodes {
|
|
int psn_len;
|
|
union {
|
|
caddr_t psu_buf;
|
|
struct pf_src_node *psu_src_nodes;
|
|
} psn_u __attribute__((aligned(8)));
|
|
#define psn_buf psn_u.psu_buf
|
|
#define psn_src_nodes psn_u.psu_src_nodes
|
|
};
|
|
|
|
#ifdef KERNEL
|
|
struct pfioc_src_nodes_32 {
|
|
int psn_len;
|
|
union {
|
|
user32_addr_t psu_buf;
|
|
user32_addr_t psu_src_nodes;
|
|
} psn_u __attribute__((aligned(8)));
|
|
};
|
|
|
|
struct pfioc_src_nodes_64 {
|
|
int psn_len;
|
|
union {
|
|
user64_addr_t psu_buf;
|
|
user64_addr_t psu_src_nodes;
|
|
} psn_u __attribute__((aligned(8)));
|
|
};
|
|
#endif /* KERNEL */
|
|
|
|
struct pfioc_if {
|
|
char ifname[IFNAMSIZ];
|
|
};
|
|
|
|
struct pfioc_tm {
|
|
int timeout;
|
|
int seconds;
|
|
};
|
|
|
|
struct pfioc_limit {
|
|
int index;
|
|
unsigned limit;
|
|
};
|
|
|
|
struct pfioc_altq {
|
|
u_int32_t action;
|
|
u_int32_t ticket;
|
|
u_int32_t nr;
|
|
struct pf_altq altq __attribute__((aligned(8)));
|
|
};
|
|
|
|
struct pfioc_qstats {
|
|
u_int32_t ticket;
|
|
u_int32_t nr;
|
|
void *buf __attribute__((aligned(8)));
|
|
int nbytes __attribute__((aligned(8)));
|
|
u_int8_t scheduler;
|
|
};
|
|
|
|
struct pfioc_ruleset {
|
|
u_int32_t nr;
|
|
char path[MAXPATHLEN];
|
|
char name[PF_ANCHOR_NAME_SIZE];
|
|
};
|
|
|
|
#define PF_RULESET_ALTQ (PF_RULESET_MAX)
|
|
#define PF_RULESET_TABLE (PF_RULESET_MAX+1)
|
|
struct pfioc_trans {
|
|
int size; /* number of elements */
|
|
int esize; /* size of each element in bytes */
|
|
struct pfioc_trans_e {
|
|
int rs_num;
|
|
char anchor[MAXPATHLEN];
|
|
u_int32_t ticket;
|
|
} *array __attribute__((aligned(8)));
|
|
};
|
|
|
|
#ifdef KERNEL
|
|
struct pfioc_trans_32 {
|
|
int size; /* number of elements */
|
|
int esize; /* size of each element in bytes */
|
|
user32_addr_t array __attribute__((aligned(8)));
|
|
};
|
|
|
|
struct pfioc_trans_64 {
|
|
int size; /* number of elements */
|
|
int esize; /* size of each element in bytes */
|
|
user64_addr_t array __attribute__((aligned(8)));
|
|
};
|
|
#endif /* KERNEL */
|
|
|
|
|
|
#define PFR_FLAG_ATOMIC 0x00000001
|
|
#define PFR_FLAG_DUMMY 0x00000002
|
|
#define PFR_FLAG_FEEDBACK 0x00000004
|
|
#define PFR_FLAG_CLSTATS 0x00000008
|
|
#define PFR_FLAG_ADDRSTOO 0x00000010
|
|
#define PFR_FLAG_REPLACE 0x00000020
|
|
#define PFR_FLAG_ALLRSETS 0x00000040
|
|
#define PFR_FLAG_ALLMASK 0x0000007F
|
|
#ifdef KERNEL
|
|
#define PFR_FLAG_USERIOCTL 0x10000000
|
|
#endif /* KERNEL */
|
|
|
|
struct pfioc_table {
|
|
struct pfr_table pfrio_table;
|
|
void *pfrio_buffer __attribute__((aligned(8)));
|
|
int pfrio_esize __attribute__((aligned(8)));
|
|
int pfrio_size;
|
|
int pfrio_size2;
|
|
int pfrio_nadd;
|
|
int pfrio_ndel;
|
|
int pfrio_nchange;
|
|
int pfrio_flags;
|
|
u_int32_t pfrio_ticket;
|
|
};
|
|
#define pfrio_exists pfrio_nadd
|
|
#define pfrio_nzero pfrio_nadd
|
|
#define pfrio_nmatch pfrio_nadd
|
|
#define pfrio_naddr pfrio_size2
|
|
#define pfrio_setflag pfrio_size2
|
|
#define pfrio_clrflag pfrio_nadd
|
|
|
|
#ifdef KERNEL
|
|
struct pfioc_table_32 {
|
|
struct pfr_table pfrio_table;
|
|
user32_addr_t pfrio_buffer __attribute__((aligned(8)));
|
|
int pfrio_esize __attribute__((aligned(8)));
|
|
int pfrio_size;
|
|
int pfrio_size2;
|
|
int pfrio_nadd;
|
|
int pfrio_ndel;
|
|
int pfrio_nchange;
|
|
int pfrio_flags;
|
|
u_int32_t pfrio_ticket;
|
|
};
|
|
|
|
struct pfioc_table_64 {
|
|
struct pfr_table pfrio_table;
|
|
user64_addr_t pfrio_buffer __attribute__((aligned(8)));
|
|
int pfrio_esize __attribute__((aligned(8)));
|
|
int pfrio_size;
|
|
int pfrio_size2;
|
|
int pfrio_nadd;
|
|
int pfrio_ndel;
|
|
int pfrio_nchange;
|
|
int pfrio_flags;
|
|
u_int32_t pfrio_ticket;
|
|
};
|
|
#endif /* KERNEL */
|
|
|
|
struct pfioc_iface {
|
|
char pfiio_name[IFNAMSIZ];
|
|
void *pfiio_buffer __attribute__((aligned(8)));
|
|
int pfiio_esize __attribute__((aligned(8)));
|
|
int pfiio_size;
|
|
int pfiio_nzero;
|
|
int pfiio_flags;
|
|
};
|
|
|
|
#ifdef KERNEL
|
|
struct pfioc_iface_32 {
|
|
char pfiio_name[IFNAMSIZ];
|
|
user32_addr_t pfiio_buffer __attribute__((aligned(8)));
|
|
int pfiio_esize __attribute__((aligned(8)));
|
|
int pfiio_size;
|
|
int pfiio_nzero;
|
|
int pfiio_flags;
|
|
};
|
|
|
|
struct pfioc_iface_64 {
|
|
char pfiio_name[IFNAMSIZ];
|
|
user64_addr_t pfiio_buffer __attribute__((aligned(8)));
|
|
int pfiio_esize __attribute__((aligned(8)));
|
|
int pfiio_size;
|
|
int pfiio_nzero;
|
|
int pfiio_flags;
|
|
};
|
|
#endif /* KERNEL */
|
|
|
|
struct pf_ifspeed {
|
|
char ifname[IFNAMSIZ];
|
|
u_int64_t baudrate;
|
|
};
|
|
|
|
/*
|
|
* ioctl operations
|
|
*/
|
|
|
|
#define DIOCSTART _IO ('D', 1)
|
|
#define DIOCSTOP _IO ('D', 2)
|
|
#define DIOCADDRULE _IOWR('D', 4, struct pfioc_rule)
|
|
#define DIOCGETSTARTERS _IOWR('D', 5, struct pfioc_tokens)
|
|
#define DIOCGETRULES _IOWR('D', 6, struct pfioc_rule)
|
|
#define DIOCGETRULE _IOWR('D', 7, struct pfioc_rule)
|
|
#define DIOCSTARTREF _IOR ('D', 8, u_int64_t)
|
|
#define DIOCSTOPREF _IOWR('D', 9, struct pfioc_remove_token)
|
|
/* XXX cut 10 - 17 */
|
|
#define DIOCCLRSTATES _IOWR('D', 18, struct pfioc_state_kill)
|
|
#define DIOCGETSTATE _IOWR('D', 19, struct pfioc_state)
|
|
#define DIOCSETSTATUSIF _IOWR('D', 20, struct pfioc_if)
|
|
#define DIOCGETSTATUS _IOWR('D', 21, struct pf_status)
|
|
#define DIOCCLRSTATUS _IO ('D', 22)
|
|
#define DIOCNATLOOK _IOWR('D', 23, struct pfioc_natlook)
|
|
#define DIOCSETDEBUG _IOWR('D', 24, u_int32_t)
|
|
#define DIOCGETSTATES _IOWR('D', 25, struct pfioc_states)
|
|
#define DIOCCHANGERULE _IOWR('D', 26, struct pfioc_rule)
|
|
#define DIOCINSERTRULE _IOWR('D', 27, struct pfioc_rule)
|
|
#define DIOCDELETERULE _IOWR('D', 28, struct pfioc_rule)
|
|
#define DIOCSETTIMEOUT _IOWR('D', 29, struct pfioc_tm)
|
|
#define DIOCGETTIMEOUT _IOWR('D', 30, struct pfioc_tm)
|
|
#define DIOCADDSTATE _IOWR('D', 37, struct pfioc_state)
|
|
#define DIOCCLRRULECTRS _IO ('D', 38)
|
|
#define DIOCGETLIMIT _IOWR('D', 39, struct pfioc_limit)
|
|
#define DIOCSETLIMIT _IOWR('D', 40, struct pfioc_limit)
|
|
#define DIOCKILLSTATES _IOWR('D', 41, struct pfioc_state_kill)
|
|
#define DIOCSTARTALTQ _IO ('D', 42)
|
|
#define DIOCSTOPALTQ _IO ('D', 43)
|
|
#define DIOCADDALTQ _IOWR('D', 45, struct pfioc_altq)
|
|
#define DIOCGETALTQS _IOWR('D', 47, struct pfioc_altq)
|
|
#define DIOCGETALTQ _IOWR('D', 48, struct pfioc_altq)
|
|
#define DIOCCHANGEALTQ _IOWR('D', 49, struct pfioc_altq)
|
|
#define DIOCGETQSTATS _IOWR('D', 50, struct pfioc_qstats)
|
|
#define DIOCBEGINADDRS _IOWR('D', 51, struct pfioc_pooladdr)
|
|
#define DIOCADDADDR _IOWR('D', 52, struct pfioc_pooladdr)
|
|
#define DIOCGETADDRS _IOWR('D', 53, struct pfioc_pooladdr)
|
|
#define DIOCGETADDR _IOWR('D', 54, struct pfioc_pooladdr)
|
|
#define DIOCCHANGEADDR _IOWR('D', 55, struct pfioc_pooladdr)
|
|
/* XXX cut 55 - 57 */
|
|
#define DIOCGETRULESETS _IOWR('D', 58, struct pfioc_ruleset)
|
|
#define DIOCGETRULESET _IOWR('D', 59, struct pfioc_ruleset)
|
|
#define DIOCRCLRTABLES _IOWR('D', 60, struct pfioc_table)
|
|
#define DIOCRADDTABLES _IOWR('D', 61, struct pfioc_table)
|
|
#define DIOCRDELTABLES _IOWR('D', 62, struct pfioc_table)
|
|
#define DIOCRGETTABLES _IOWR('D', 63, struct pfioc_table)
|
|
#define DIOCRGETTSTATS _IOWR('D', 64, struct pfioc_table)
|
|
#define DIOCRCLRTSTATS _IOWR('D', 65, struct pfioc_table)
|
|
#define DIOCRCLRADDRS _IOWR('D', 66, struct pfioc_table)
|
|
#define DIOCRADDADDRS _IOWR('D', 67, struct pfioc_table)
|
|
#define DIOCRDELADDRS _IOWR('D', 68, struct pfioc_table)
|
|
#define DIOCRSETADDRS _IOWR('D', 69, struct pfioc_table)
|
|
#define DIOCRGETADDRS _IOWR('D', 70, struct pfioc_table)
|
|
#define DIOCRGETASTATS _IOWR('D', 71, struct pfioc_table)
|
|
#define DIOCRCLRASTATS _IOWR('D', 72, struct pfioc_table)
|
|
#define DIOCRTSTADDRS _IOWR('D', 73, struct pfioc_table)
|
|
#define DIOCRSETTFLAGS _IOWR('D', 74, struct pfioc_table)
|
|
#define DIOCRINADEFINE _IOWR('D', 77, struct pfioc_table)
|
|
#define DIOCOSFPFLUSH _IO('D', 78)
|
|
#define DIOCOSFPADD _IOWR('D', 79, struct pf_osfp_ioctl)
|
|
#define DIOCOSFPGET _IOWR('D', 80, struct pf_osfp_ioctl)
|
|
#define DIOCXBEGIN _IOWR('D', 81, struct pfioc_trans)
|
|
#define DIOCXCOMMIT _IOWR('D', 82, struct pfioc_trans)
|
|
#define DIOCXROLLBACK _IOWR('D', 83, struct pfioc_trans)
|
|
#define DIOCGETSRCNODES _IOWR('D', 84, struct pfioc_src_nodes)
|
|
#define DIOCCLRSRCNODES _IO('D', 85)
|
|
#define DIOCSETHOSTID _IOWR('D', 86, u_int32_t)
|
|
#define DIOCIGETIFACES _IOWR('D', 87, struct pfioc_iface)
|
|
#define DIOCSETIFFLAG _IOWR('D', 89, struct pfioc_iface)
|
|
#define DIOCCLRIFFLAG _IOWR('D', 90, struct pfioc_iface)
|
|
#define DIOCKILLSRCNODES _IOWR('D', 91, struct pfioc_src_node_kill)
|
|
#define DIOCGIFSPEED _IOWR('D', 92, struct pf_ifspeed)
|
|
|
|
#ifdef KERNEL
|
|
RB_HEAD(pf_src_tree, pf_src_node);
|
|
RB_PROTOTYPE_SC(__private_extern__, pf_src_tree, pf_src_node, entry,
|
|
pf_src_compare);
|
|
extern struct pf_src_tree tree_src_tracking;
|
|
|
|
RB_HEAD(pf_state_tree_id, pf_state);
|
|
RB_PROTOTYPE_SC(__private_extern__, pf_state_tree_id, pf_state,
|
|
entry_id, pf_state_compare_id);
|
|
extern struct pf_state_tree_id tree_id;
|
|
extern struct pf_state_queue state_list;
|
|
|
|
TAILQ_HEAD(pf_poolqueue, pf_pool);
|
|
extern struct pf_poolqueue pf_pools[2];
|
|
extern struct pf_palist pf_pabuf;
|
|
extern u_int32_t ticket_pabuf;
|
|
#if PF_ALTQ
|
|
TAILQ_HEAD(pf_altqqueue, pf_altq);
|
|
extern struct pf_altqqueue pf_altqs[2];
|
|
extern u_int32_t ticket_altqs_active;
|
|
extern u_int32_t ticket_altqs_inactive;
|
|
extern int altqs_inactive_open;
|
|
extern struct pf_altqqueue *pf_altqs_active;
|
|
extern struct pf_altqqueue *pf_altqs_inactive;
|
|
#endif /* PF_ALTQ */
|
|
extern struct pf_poolqueue *pf_pools_active;
|
|
extern struct pf_poolqueue *pf_pools_inactive;
|
|
|
|
__private_extern__ int pf_tbladdr_setup(struct pf_ruleset *,
|
|
struct pf_addr_wrap *);
|
|
__private_extern__ void pf_tbladdr_remove(struct pf_addr_wrap *);
|
|
__private_extern__ void pf_tbladdr_copyout(struct pf_addr_wrap *);
|
|
__private_extern__ void pf_calc_skip_steps(struct pf_rulequeue *);
|
|
__private_extern__ u_int32_t pf_calc_state_key_flowhash(struct pf_state_key *);
|
|
|
|
extern struct pool pf_src_tree_pl, pf_rule_pl;
|
|
extern struct pool pf_state_pl, pf_state_key_pl, pf_pooladdr_pl;
|
|
extern struct pool pf_state_scrub_pl;
|
|
#if PF_ALTQ
|
|
extern struct pool pf_altq_pl;
|
|
#endif /* PF_ALTQ */
|
|
extern struct pool pf_app_state_pl;
|
|
|
|
extern struct thread *pf_purge_thread;
|
|
|
|
__private_extern__ void pfinit(void);
|
|
__private_extern__ void pf_purge_thread_fn(void *, wait_result_t);
|
|
__private_extern__ void pf_purge_expired_src_nodes(void);
|
|
__private_extern__ void pf_purge_expired_states(u_int32_t);
|
|
__private_extern__ void pf_unlink_state(struct pf_state *);
|
|
__private_extern__ void pf_free_state(struct pf_state *);
|
|
__private_extern__ int pf_insert_state(struct pfi_kif *, struct pf_state *);
|
|
__private_extern__ int pf_insert_src_node(struct pf_src_node **,
|
|
struct pf_rule *, struct pf_addr *, sa_family_t);
|
|
__private_extern__ void pf_src_tree_remove_state(struct pf_state *);
|
|
__private_extern__ struct pf_state *pf_find_state_byid(struct pf_state_cmp *);
|
|
__private_extern__ struct pf_state *pf_find_state_all(struct pf_state_key_cmp *,
|
|
u_int, int *);
|
|
__private_extern__ void pf_print_state(struct pf_state *);
|
|
__private_extern__ void pf_print_flags(u_int8_t);
|
|
__private_extern__ u_int16_t pf_cksum_fixup(u_int16_t, u_int16_t, u_int16_t,
|
|
u_int8_t);
|
|
|
|
extern struct ifnet *sync_ifp;
|
|
extern struct pf_rule pf_default_rule;
|
|
__private_extern__ void pf_addrcpy(struct pf_addr *, struct pf_addr *,
|
|
u_int8_t);
|
|
__private_extern__ void pf_rm_rule(struct pf_rulequeue *, struct pf_rule *);
|
|
|
|
struct ip_fw_args;
|
|
#if INET
|
|
__private_extern__ int pf_test(int, struct ifnet *, struct mbuf **,
|
|
struct ether_header *, struct ip_fw_args *);
|
|
#endif /* INET */
|
|
|
|
#if INET6
|
|
__private_extern__ int pf_test6(int, struct ifnet *, struct mbuf **,
|
|
struct ether_header *, struct ip_fw_args *);
|
|
__private_extern__ void pf_poolmask(struct pf_addr *, struct pf_addr *,
|
|
struct pf_addr *, struct pf_addr *, u_int8_t);
|
|
__private_extern__ void pf_addr_inc(struct pf_addr *, sa_family_t);
|
|
#endif /* INET6 */
|
|
|
|
__private_extern__ struct mbuf *pf_lazy_makewritable(struct pf_pdesc *,
|
|
struct mbuf *, int);
|
|
__private_extern__ void *pf_pull_hdr(struct mbuf *, int, void *, int,
|
|
u_short *, u_short *, sa_family_t);
|
|
__private_extern__ void pf_change_a(void *, u_int16_t *, u_int32_t, u_int8_t);
|
|
__private_extern__ int pflog_packet(struct pfi_kif *, struct mbuf *,
|
|
sa_family_t, u_int8_t, u_int8_t, struct pf_rule *, struct pf_rule *,
|
|
struct pf_ruleset *, struct pf_pdesc *);
|
|
__private_extern__ int pf_match_addr(u_int8_t, struct pf_addr *,
|
|
struct pf_addr *, struct pf_addr *, sa_family_t);
|
|
__private_extern__ int pf_match_addr_range(struct pf_addr *, struct pf_addr *,
|
|
struct pf_addr *, sa_family_t);
|
|
__private_extern__ int pf_match(u_int8_t, u_int32_t, u_int32_t, u_int32_t);
|
|
__private_extern__ int pf_match_port(u_int8_t, u_int16_t, u_int16_t, u_int16_t);
|
|
__private_extern__ int pf_match_xport(u_int8_t, u_int8_t, union pf_rule_xport *,
|
|
union pf_state_xport *);
|
|
__private_extern__ int pf_match_uid(u_int8_t, uid_t, uid_t, uid_t);
|
|
__private_extern__ int pf_match_gid(u_int8_t, gid_t, gid_t, gid_t);
|
|
|
|
__private_extern__ void pf_normalize_init(void);
|
|
__private_extern__ int pf_normalize_isempty(void);
|
|
__private_extern__ int pf_normalize_ip(struct mbuf **, int, struct pfi_kif *,
|
|
u_short *, struct pf_pdesc *);
|
|
__private_extern__ int pf_normalize_ip6(struct mbuf **, int, struct pfi_kif *,
|
|
u_short *, struct pf_pdesc *);
|
|
__private_extern__ int pf_normalize_tcp(int, struct pfi_kif *, struct mbuf *,
|
|
int, int, void *, struct pf_pdesc *);
|
|
__private_extern__ void pf_normalize_tcp_cleanup(struct pf_state *);
|
|
__private_extern__ int pf_normalize_tcp_init(struct mbuf *, int,
|
|
struct pf_pdesc *, struct tcphdr *, struct pf_state_peer *,
|
|
struct pf_state_peer *);
|
|
__private_extern__ int pf_normalize_tcp_stateful(struct mbuf *, int,
|
|
struct pf_pdesc *, u_short *, struct tcphdr *, struct pf_state *,
|
|
struct pf_state_peer *, struct pf_state_peer *, int *);
|
|
__private_extern__ u_int64_t pf_state_expires(const struct pf_state *);
|
|
__private_extern__ void pf_purge_expired_fragments(void);
|
|
__private_extern__ int pf_routable(struct pf_addr *addr, sa_family_t af,
|
|
struct pfi_kif *);
|
|
__private_extern__ int pf_rtlabel_match(struct pf_addr *, sa_family_t,
|
|
struct pf_addr_wrap *);
|
|
__private_extern__ int pf_socket_lookup(int, struct pf_pdesc *);
|
|
__private_extern__ struct pf_state_key *pf_alloc_state_key(struct pf_state *,
|
|
struct pf_state_key *);
|
|
__private_extern__ void pfr_initialize(void);
|
|
__private_extern__ int pfr_match_addr(struct pfr_ktable *, struct pf_addr *,
|
|
sa_family_t);
|
|
__private_extern__ void pfr_update_stats(struct pfr_ktable *, struct pf_addr *,
|
|
sa_family_t, u_int64_t, int, int, int);
|
|
__private_extern__ int pfr_pool_get(struct pfr_ktable *, int *,
|
|
struct pf_addr *, struct pf_addr **, struct pf_addr **, sa_family_t);
|
|
__private_extern__ void pfr_dynaddr_update(struct pfr_ktable *,
|
|
struct pfi_dynaddr *);
|
|
__private_extern__ void pfr_table_copyin_cleanup(struct pfr_table *);
|
|
__private_extern__ struct pfr_ktable *pfr_attach_table(struct pf_ruleset *,
|
|
char *);
|
|
__private_extern__ void pfr_detach_table(struct pfr_ktable *);
|
|
__private_extern__ int pfr_clr_tables(struct pfr_table *, int *, int);
|
|
__private_extern__ int pfr_add_tables(user_addr_t, int, int *, int);
|
|
__private_extern__ int pfr_del_tables(user_addr_t, int, int *, int);
|
|
__private_extern__ int pfr_get_tables(struct pfr_table *, user_addr_t,
|
|
int *, int);
|
|
__private_extern__ int pfr_get_tstats(struct pfr_table *, user_addr_t,
|
|
int *, int);
|
|
__private_extern__ int pfr_clr_tstats(user_addr_t, int, int *, int);
|
|
__private_extern__ int pfr_set_tflags(user_addr_t, int, int, int, int *,
|
|
int *, int);
|
|
__private_extern__ int pfr_clr_addrs(struct pfr_table *, int *, int);
|
|
__private_extern__ int pfr_insert_kentry(struct pfr_ktable *, struct pfr_addr *,
|
|
u_int64_t);
|
|
__private_extern__ int pfr_add_addrs(struct pfr_table *, user_addr_t,
|
|
int, int *, int);
|
|
__private_extern__ int pfr_del_addrs(struct pfr_table *, user_addr_t,
|
|
int, int *, int);
|
|
__private_extern__ int pfr_set_addrs(struct pfr_table *, user_addr_t,
|
|
int, int *, int *, int *, int *, int, u_int32_t);
|
|
__private_extern__ int pfr_get_addrs(struct pfr_table *, user_addr_t,
|
|
int *, int);
|
|
__private_extern__ int pfr_get_astats(struct pfr_table *, user_addr_t,
|
|
int *, int);
|
|
__private_extern__ int pfr_clr_astats(struct pfr_table *, user_addr_t,
|
|
int, int *, int);
|
|
__private_extern__ int pfr_tst_addrs(struct pfr_table *, user_addr_t,
|
|
int, int *, int);
|
|
__private_extern__ int pfr_ina_begin(struct pfr_table *, u_int32_t *, int *,
|
|
int);
|
|
__private_extern__ int pfr_ina_rollback(struct pfr_table *, u_int32_t, int *,
|
|
int);
|
|
__private_extern__ int pfr_ina_commit(struct pfr_table *, u_int32_t, int *,
|
|
int *, int);
|
|
__private_extern__ int pfr_ina_define(struct pfr_table *, user_addr_t,
|
|
int, int *, int *, u_int32_t, int);
|
|
|
|
extern struct pfi_kif *pfi_all;
|
|
|
|
__private_extern__ void pfi_initialize(void);
|
|
__private_extern__ struct pfi_kif *pfi_kif_get(const char *);
|
|
__private_extern__ void pfi_kif_ref(struct pfi_kif *, enum pfi_kif_refs);
|
|
__private_extern__ void pfi_kif_unref(struct pfi_kif *, enum pfi_kif_refs);
|
|
__private_extern__ int pfi_kif_match(struct pfi_kif *, struct pfi_kif *);
|
|
__private_extern__ void pfi_attach_ifnet(struct ifnet *);
|
|
__private_extern__ void pfi_detach_ifnet(struct ifnet *);
|
|
__private_extern__ int pfi_match_addr(struct pfi_dynaddr *, struct pf_addr *,
|
|
sa_family_t);
|
|
__private_extern__ int pfi_dynaddr_setup(struct pf_addr_wrap *, sa_family_t);
|
|
__private_extern__ void pfi_dynaddr_remove(struct pf_addr_wrap *);
|
|
__private_extern__ void pfi_dynaddr_copyout(struct pf_addr_wrap *);
|
|
__private_extern__ void pfi_update_status(const char *, struct pf_status *);
|
|
__private_extern__ int pfi_get_ifaces(const char *, user_addr_t, int *);
|
|
__private_extern__ int pfi_set_flags(const char *, int);
|
|
__private_extern__ int pfi_clear_flags(const char *, int);
|
|
|
|
__private_extern__ u_int16_t pf_tagname2tag(char *);
|
|
__private_extern__ void pf_tag2tagname(u_int16_t, char *);
|
|
__private_extern__ void pf_tag_ref(u_int16_t);
|
|
__private_extern__ void pf_tag_unref(u_int16_t);
|
|
__private_extern__ int pf_tag_packet(struct mbuf *, struct pf_mtag *,
|
|
int, unsigned int, struct pf_pdesc *);
|
|
__private_extern__ void pf_step_into_anchor(int *, struct pf_ruleset **, int,
|
|
struct pf_rule **, struct pf_rule **, int *);
|
|
__private_extern__ int pf_step_out_of_anchor(int *, struct pf_ruleset **, int,
|
|
struct pf_rule **, struct pf_rule **, int *);
|
|
__private_extern__ u_int32_t pf_qname2qid(char *);
|
|
__private_extern__ void pf_qid2qname(u_int32_t, char *);
|
|
__private_extern__ void pf_qid_unref(u_int32_t);
|
|
|
|
extern struct pf_status pf_status;
|
|
extern struct pool pf_frent_pl, pf_frag_pl;
|
|
|
|
struct pf_pool_limit {
|
|
void *pp;
|
|
unsigned limit;
|
|
};
|
|
extern struct pf_pool_limit pf_pool_limits[PF_LIMIT_MAX];
|
|
|
|
__private_extern__ int pf_af_hook(struct ifnet *, struct mbuf **,
|
|
struct mbuf **, unsigned int, int, struct ip_fw_args *);
|
|
__private_extern__ int pf_ifaddr_hook(struct ifnet *);
|
|
__private_extern__ void pf_ifnet_hook(struct ifnet *, int);
|
|
|
|
/*
|
|
* The following are defined with "private extern" storage class for
|
|
* kernel, and "extern" for user-space.
|
|
*/
|
|
extern struct pf_anchor_global pf_anchors;
|
|
extern struct pf_anchor pf_main_anchor;
|
|
#define pf_main_ruleset pf_main_anchor.ruleset
|
|
|
|
extern int pf_is_enabled;
|
|
extern int16_t pf_nat64_configured;
|
|
#define PF_IS_ENABLED (pf_is_enabled != 0)
|
|
extern u_int32_t pf_hash_seed;
|
|
|
|
#if PF_ALTQ
|
|
extern u_int32_t altq_allowed;
|
|
#endif /* PF_ALTQ */
|
|
|
|
/* these ruleset functions can be linked into userland programs (pfctl) */
|
|
__private_extern__ int pf_get_ruleset_number(u_int8_t);
|
|
__private_extern__ void pf_init_ruleset(struct pf_ruleset *);
|
|
__private_extern__ int pf_anchor_setup(struct pf_rule *,
|
|
const struct pf_ruleset *, const char *);
|
|
__private_extern__ int pf_anchor_copyout(const struct pf_ruleset *,
|
|
const struct pf_rule *, struct pfioc_rule *);
|
|
__private_extern__ void pf_anchor_remove(struct pf_rule *);
|
|
__private_extern__ void pf_remove_if_empty_ruleset(struct pf_ruleset *);
|
|
__private_extern__ struct pf_anchor *pf_find_anchor(const char *);
|
|
__private_extern__ struct pf_ruleset *pf_find_ruleset(const char *);
|
|
__private_extern__ struct pf_ruleset *pf_find_ruleset_with_owner(const char *,
|
|
const char *, int, int *);
|
|
__private_extern__ struct pf_ruleset *pf_find_or_create_ruleset(const char *);
|
|
__private_extern__ void pf_rs_initialize(void);
|
|
|
|
__private_extern__ int pf_osfp_add(struct pf_osfp_ioctl *);
|
|
__private_extern__ struct pf_osfp_enlist *pf_osfp_fingerprint(struct pf_pdesc *,
|
|
struct mbuf *, int, const struct tcphdr *);
|
|
__private_extern__ struct pf_osfp_enlist *pf_osfp_fingerprint_hdr(
|
|
const struct ip *, const struct ip6_hdr *, const struct tcphdr *);
|
|
__private_extern__ void pf_osfp_flush(void);
|
|
__private_extern__ int pf_osfp_get(struct pf_osfp_ioctl *);
|
|
__private_extern__ void pf_osfp_initialize(void);
|
|
__private_extern__ int pf_osfp_match(struct pf_osfp_enlist *, pf_osfp_t);
|
|
__private_extern__ struct pf_os_fingerprint *pf_osfp_validate(void);
|
|
__private_extern__ struct pf_mtag *pf_find_mtag(struct mbuf *);
|
|
__private_extern__ struct pf_mtag *pf_get_mtag(struct mbuf *);
|
|
#else /* !KERNEL */
|
|
extern struct pf_anchor_global pf_anchors;
|
|
extern struct pf_anchor pf_main_anchor;
|
|
#define pf_main_ruleset pf_main_anchor.ruleset
|
|
|
|
/* these ruleset functions can be linked into userland programs (pfctl) */
|
|
extern int pf_get_ruleset_number(u_int8_t);
|
|
extern void pf_init_ruleset(struct pf_ruleset *);
|
|
extern int pf_anchor_setup(struct pf_rule *, const struct pf_ruleset *,
|
|
const char *);
|
|
extern int pf_anchor_copyout(const struct pf_ruleset *, const struct pf_rule *,
|
|
struct pfioc_rule *);
|
|
extern void pf_anchor_remove(struct pf_rule *);
|
|
extern void pf_remove_if_empty_ruleset(struct pf_ruleset *);
|
|
extern struct pf_anchor *pf_find_anchor(const char *);
|
|
extern struct pf_ruleset *pf_find_ruleset(const char *);
|
|
extern struct pf_ruleset *pf_find_ruleset_with_owner(const char *,
|
|
const char *, int, int *);
|
|
extern struct pf_ruleset *pf_find_or_create_ruleset(const char *);
|
|
extern void pf_rs_initialize(void);
|
|
#endif /* !KERNEL */
|
|
|
|
#ifdef __cplusplus
|
|
}
|
|
#endif
|
|
#endif /* PF || !KERNEL */
|
|
#endif /* PRIVATE */
|
|
#endif /* _NET_PFVAR_H_ */
|