d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
9734411552
freebsd-skq/sys/contrib/pf/net/pf.c

7610 lines
190 KiB
C
Raw Normal View History

Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
/* $OpenBSD: pf.c,v 1.527 2007/02/22 15:23:23 pyr Exp $ */
Import pf.c, rev 1.559 by markus: allow state reuse for tcp if both sides are in FIN_WAIT_2 and a new SYN arrives; ok dhartmei, henning, feedback aaron
2008-08-04 14:08:55 +00:00
/* add: $OpenBSD: pf.c,v 1.559 2007/09/18 18:45:59 markus Exp $ */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/*
* Copyright (c) 2001 Daniel Hartmeier
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
* Copyright (c) 2002,2003 Henning Brauer
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
* 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.
*
* Effort sponsored in part by the Defense Advanced Research Projects
* Agency (DARPA) and Air Force Research Laboratory, Air Force
* Materiel Command, USAF, under agreement number F30602-01-2-0537.
*
*/
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#include "opt_inet.h"
#include "opt_inet6.h"
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#include "opt_bpf.h"
#include "opt_pf.h"
Fix -Wundef warnings found when compiling i386 LINT, GENERIC and custom kernels.
2005-12-05 11:58:35 +00:00
#ifdef DEV_BPF
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#define NBPFILTER DEV_BPF
Fix -Wundef warnings found when compiling i386 LINT, GENERIC and custom kernels.
2005-12-05 11:58:35 +00:00
#else
#define NBPFILTER 0
#endif
#ifdef DEV_PFLOG
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#define NPFLOG DEV_PFLOG
Fix -Wundef warnings found when compiling i386 LINT, GENERIC and custom kernels.
2005-12-05 11:58:35 +00:00
#else
#define NPFLOG 0
#endif
#ifdef DEV_PFSYNC
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#define NPFSYNC DEV_PFSYNC
Fix -Wundef warnings found when compiling i386 LINT, GENERIC and custom kernels.
2005-12-05 11:58:35 +00:00
#else
#define NPFSYNC 0
#endif
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#include "bpfilter.h"
#include "pflog.h"
#include "pfsync.h"
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/mbuf.h>
#include <sys/filio.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/kernel.h>
#include <sys/time.h>
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#include <sys/sysctl.h>
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#include <sys/endian.h>
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#include <sys/pool.h>
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#include <sys/proc.h>
#ifdef __FreeBSD__
#include <sys/kthread.h>
#include <sys/lock.h>
#include <sys/sx.h>
#else
#include <sys/rwlock.h>
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#include <net/if.h>
#include <net/if_types.h>
#include <net/bpf.h>
#include <net/route.h>
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifndef __FreeBSD__
#include <net/radix_mpath.h>
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#include <netinet/in.h>
#include <netinet/in_var.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/ip_var.h>
#include <netinet/tcp.h>
#include <netinet/tcp_seq.h>
#include <netinet/udp.h>
#include <netinet/ip_icmp.h>
#include <netinet/in_pcb.h>
#include <netinet/tcp_timer.h>
#include <netinet/tcp_var.h>
#include <netinet/udp_var.h>
#include <netinet/icmp_var.h>
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#include <netinet/if_ether.h>
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifndef __FreeBSD__
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#include <dev/rndvar.h>
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#include <net/pfvar.h>
#include <net/if_pflog.h>
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#if NPFSYNC > 0
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#include <net/if_pfsync.h>
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#endif /* NPFSYNC > 0 */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef INET6
#include <netinet/ip6.h>
#include <netinet/in_pcb.h>
#include <netinet/icmp6.h>
#include <netinet6/nd6.h>
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#include <netinet6/ip6_var.h>
#include <netinet6/in6_pcb.h>
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#endif /* INET6 */
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#include <machine/in_cksum.h>
#include <sys/limits.h>
#include <sys/ucred.h>
Complete break-out of sys/sys/mac.h into sys/security/mac/mac_framework.h begun with a repo-copy of mac.h to mac_framework.h. sys/mac.h now contains the userspace and user<->kernel API and definitions, with all in-kernel interfaces moved to mac_framework.h, which is now included across most of the kernel instead. This change is the first step in a larger cleanup and sweep of MAC Framework interfaces in the kernel, and will not be MFC'd. Obtained from: TrustedBSD Project Sponsored by: SPARTA
2006-10-22 11:52:19 +00:00
#include <security/mac/mac_framework.h>
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
extern int ip_optcopy(struct ip *, struct ip *);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
extern int debug_pfugidhack;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#define DPFPRINTF(n, x) if (pf_status.debug >= (n)) printf x
/*
* Global variables
*/
struct pf_altqqueue pf_altqs[2];
struct pf_palist pf_pabuf;
struct pf_altqqueue *pf_altqs_active;
struct pf_altqqueue *pf_altqs_inactive;
struct pf_status pf_status;
u_int32_t ticket_altqs_active;
u_int32_t ticket_altqs_inactive;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
int altqs_inactive_open;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
u_int32_t ticket_pabuf;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct pf_anchor_stackframe {
struct pf_ruleset *rs;
struct pf_rule *r;
struct pf_anchor_node *parent;
struct pf_anchor *child;
} pf_anchor_stack[64];
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
uma_zone_t pf_src_tree_pl, pf_rule_pl;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
uma_zone_t pf_state_pl, pf_altq_pl, pf_pooladdr_pl;
#else
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pool pf_src_tree_pl, pf_rule_pl;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pool pf_state_pl, pf_altq_pl, pf_pooladdr_pl;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
void pf_print_host(struct pf_addr *, u_int16_t, u_int8_t);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
void pf_init_threshold(struct pf_threshold *, u_int32_t,
u_int32_t);
void pf_add_threshold(struct pf_threshold *);
int pf_check_threshold(struct pf_threshold *);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
void pf_change_ap(struct pf_addr *, u_int16_t *,
u_int16_t *, u_int16_t *, struct pf_addr *,
u_int16_t, u_int8_t, sa_family_t);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int pf_modulate_sack(struct mbuf *, int, struct pf_pdesc *,
struct tcphdr *, struct pf_state_peer *);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef INET6
void pf_change_a6(struct pf_addr *, u_int16_t *,
struct pf_addr *, u_int8_t);
#endif /* INET6 */
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);
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#ifdef __FreeBSD__
void pf_send_tcp(struct mbuf *,
const struct pf_rule *, sa_family_t,
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
void pf_send_tcp(const struct pf_rule *, sa_family_t,
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
const struct pf_addr *, const struct pf_addr *,
u_int16_t, u_int16_t, u_int32_t, u_int32_t,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
u_int8_t, u_int16_t, u_int16_t, u_int8_t, int,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
u_int16_t, struct ether_header *, struct ifnet *);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
void pf_send_icmp(struct mbuf *, u_int8_t, u_int8_t,
sa_family_t, struct pf_rule *);
struct pf_rule *pf_match_translation(struct pf_pdesc *, struct mbuf *,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
int, int, struct pfi_kif *,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_addr *, u_int16_t, struct pf_addr *,
u_int16_t, int);
struct pf_rule *pf_get_translation(struct pf_pdesc *, struct mbuf *,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
int, int, struct pfi_kif *, struct pf_src_node **,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_addr *, u_int16_t,
struct pf_addr *, u_int16_t,
struct pf_addr *, u_int16_t *);
int pf_test_tcp(struct pf_rule **, struct pf_state **,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
int, struct pfi_kif *, struct mbuf *, int,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
void *, struct pf_pdesc *, struct pf_rule **,
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct pf_ruleset **, struct ifqueue *,
struct inpcb *);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct pf_ruleset **, struct ifqueue *);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
int pf_test_udp(struct pf_rule **, struct pf_state **,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
int, struct pfi_kif *, struct mbuf *, int,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
void *, struct pf_pdesc *, struct pf_rule **,
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct pf_ruleset **, struct ifqueue *,
struct inpcb *);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct pf_ruleset **, struct ifqueue *);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
int pf_test_icmp(struct pf_rule **, struct pf_state **,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
int, struct pfi_kif *, struct mbuf *, int,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
void *, struct pf_pdesc *, struct pf_rule **,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct pf_ruleset **, struct ifqueue *);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
int pf_test_other(struct pf_rule **, struct pf_state **,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
int, struct pfi_kif *, struct mbuf *, int, void *,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_pdesc *, struct pf_rule **,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct pf_ruleset **, struct ifqueue *);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
int pf_test_fragment(struct pf_rule **, int,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pfi_kif *, struct mbuf *, void *,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_pdesc *, struct pf_rule **,
struct pf_ruleset **);
int pf_test_state_tcp(struct pf_state **, int,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pfi_kif *, struct mbuf *, int,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
void *, struct pf_pdesc *, u_short *);
int pf_test_state_udp(struct pf_state **, int,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pfi_kif *, struct mbuf *, int,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
void *, struct pf_pdesc *);
int pf_test_state_icmp(struct pf_state **, int,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pfi_kif *, struct mbuf *, int,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
void *, struct pf_pdesc *, u_short *);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
int pf_test_state_other(struct pf_state **, int,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pfi_kif *, struct pf_pdesc *);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
int pf_match_tag(struct mbuf *, struct pf_rule *,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_mtag *, int *);
int pf_step_out_of_anchor(int *, struct pf_ruleset **,
int, struct pf_rule **, struct pf_rule **,
int *);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
void pf_hash(struct pf_addr *, struct pf_addr *,
struct pf_poolhashkey *, sa_family_t);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
int pf_map_addr(u_int8_t, struct pf_rule *,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_addr *, struct pf_addr *,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_addr *, struct pf_src_node **);
int pf_get_sport(sa_family_t, u_int8_t, struct pf_rule *,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_addr *, struct pf_addr *, u_int16_t,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_addr *, u_int16_t*, u_int16_t, u_int16_t,
struct pf_src_node **);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
void pf_route(struct mbuf **, struct pf_rule *, int,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct ifnet *, struct pf_state *,
struct pf_pdesc *);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
void pf_route6(struct mbuf **, struct pf_rule *, int,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct ifnet *, struct pf_state *,
struct pf_pdesc *);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
/* XXX: import */
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int pf_socket_lookup(int, struct pf_pdesc *);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
u_int8_t pf_get_wscale(struct mbuf *, int, u_int16_t,
sa_family_t);
u_int16_t pf_get_mss(struct mbuf *, int, u_int16_t,
sa_family_t);
u_int16_t pf_calc_mss(struct pf_addr *, sa_family_t,
u_int16_t);
void pf_set_rt_ifp(struct pf_state *,
struct pf_addr *);
int pf_check_proto_cksum(struct mbuf *, int, int,
u_int8_t, sa_family_t);
int pf_addr_wrap_neq(struct pf_addr_wrap *,
struct pf_addr_wrap *);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_state *pf_find_state_recurse(struct pfi_kif *,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_state_cmp *, u_int8_t);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
int pf_src_connlimit(struct pf_state **);
int pf_check_congestion(struct ifqueue *);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
int in4_cksum(struct mbuf *m, u_int8_t nxt, int off, int len);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
extern int pf_end_threads;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_pool_limit pf_pool_limits[PF_LIMIT_MAX];
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#else
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
extern struct pool pfr_ktable_pl;
extern struct pool pfr_kentry_pl;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_pool_limit pf_pool_limits[PF_LIMIT_MAX] = {
{ &pf_state_pl, PFSTATE_HIWAT },
{ &pf_src_tree_pl, PFSNODE_HIWAT },
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
{ &pf_frent_pl, PFFRAG_FRENT_HIWAT },
{ &pfr_ktable_pl, PFR_KTABLE_HIWAT },
{ &pfr_kentry_pl, PFR_KENTRY_HIWAT }
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
};
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#define STATE_LOOKUP() \
do { \
if (direction == PF_IN) \
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
*state = pf_find_state_recurse( \
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
kif, &key, PF_EXT_GWY); \
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
else \
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
*state = pf_find_state_recurse( \
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
kif, &key, PF_LAN_EXT); \
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (*state == NULL || (*state)->timeout == PFTM_PURGE) \
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_DROP); \
if (direction == PF_OUT && \
(((*state)->rule.ptr->rt == PF_ROUTETO && \
(*state)->rule.ptr->direction == PF_OUT) || \
((*state)->rule.ptr->rt == PF_REPLYTO && \
(*state)->rule.ptr->direction == PF_IN)) && \
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
(*state)->rt_kif != NULL && \
(*state)->rt_kif != kif) \
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_PASS); \
} while (0)
#define STATE_TRANSLATE(s) \
(s)->lan.addr.addr32[0] != (s)->gwy.addr.addr32[0] || \
((s)->af == AF_INET6 && \
((s)->lan.addr.addr32[1] != (s)->gwy.addr.addr32[1] || \
(s)->lan.addr.addr32[2] != (s)->gwy.addr.addr32[2] || \
(s)->lan.addr.addr32[3] != (s)->gwy.addr.addr32[3])) || \
(s)->lan.port != (s)->gwy.port
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#define BOUND_IFACE(r, k) \
((r)->rule_flag & PFRULE_IFBOUND) ? (k) : pfi_all
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#define STATE_INC_COUNTERS(s) \
do { \
s->rule.ptr->states++; \
if (s->anchor.ptr != NULL) \
s->anchor.ptr->states++; \
if (s->nat_rule.ptr != NULL) \
s->nat_rule.ptr->states++; \
} while (0)
#define STATE_DEC_COUNTERS(s) \
do { \
if (s->nat_rule.ptr != NULL) \
s->nat_rule.ptr->states--; \
if (s->anchor.ptr != NULL) \
s->anchor.ptr->states--; \
s->rule.ptr->states--; \
} while (0)
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_src_tree tree_src_tracking;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_state_tree_id tree_id;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_state_queue state_list;
#ifdef __FreeBSD__
static int pf_src_compare(struct pf_src_node *, struct pf_src_node *);
static int pf_state_compare_lan_ext(struct pf_state *, struct pf_state *);
static int pf_state_compare_ext_gwy(struct pf_state *, struct pf_state *);
static int pf_state_compare_id(struct pf_state *, struct pf_state *);
#endif
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
RB_GENERATE(pf_src_tree, pf_src_node, entry, pf_src_compare);
RB_GENERATE(pf_state_tree_lan_ext, pf_state,
u.s.entry_lan_ext, pf_state_compare_lan_ext);
RB_GENERATE(pf_state_tree_ext_gwy, pf_state,
u.s.entry_ext_gwy, pf_state_compare_ext_gwy);
RB_GENERATE(pf_state_tree_id, pf_state,
u.s.entry_id, pf_state_compare_id);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Remove __inline keyword from functions that can't be inlined according to LINT. This fixes LINT compliation for now, but needs to be revised. Changes do not affect the objects. Approved by: bms(mentor)
2004-02-29 16:34:43 +00:00
static int
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
static __inline int
Remove __inline keyword from functions that can't be inlined according to LINT. This fixes LINT compliation for now, but needs to be revised. Changes do not affect the objects. Approved by: bms(mentor)
2004-02-29 16:34:43 +00:00
#endif
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
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);
switch (a->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);
break;
#endif /* INET */
#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);
break;
#endif /* INET6 */
}
return (0);
}
#ifdef __FreeBSD__
static int
#else
static __inline int
#endif
pf_state_compare_lan_ext(struct pf_state *a, struct pf_state *b)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
int diff;
if ((diff = a->proto - b->proto) != 0)
return (diff);
if ((diff = a->af - b->af) != 0)
return (diff);
switch (a->af) {
#ifdef INET
case AF_INET:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->lan.addr.addr32[0] > b->lan.addr.addr32[0])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->lan.addr.addr32[0] < b->lan.addr.addr32[0])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (-1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->ext.addr.addr32[0] > b->ext.addr.addr32[0])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->ext.addr.addr32[0] < b->ext.addr.addr32[0])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (-1);
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->lan.addr.addr32[3] > b->lan.addr.addr32[3])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->lan.addr.addr32[3] < b->lan.addr.addr32[3])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (-1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->ext.addr.addr32[3] > b->ext.addr.addr32[3])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->ext.addr.addr32[3] < b->ext.addr.addr32[3])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (-1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->lan.addr.addr32[2] > b->lan.addr.addr32[2])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->lan.addr.addr32[2] < b->lan.addr.addr32[2])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (-1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->ext.addr.addr32[2] > b->ext.addr.addr32[2])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->ext.addr.addr32[2] < b->ext.addr.addr32[2])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (-1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->lan.addr.addr32[1] > b->lan.addr.addr32[1])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->lan.addr.addr32[1] < b->lan.addr.addr32[1])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (-1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->ext.addr.addr32[1] > b->ext.addr.addr32[1])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->ext.addr.addr32[1] < b->ext.addr.addr32[1])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (-1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->lan.addr.addr32[0] > b->lan.addr.addr32[0])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->lan.addr.addr32[0] < b->lan.addr.addr32[0])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (-1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->ext.addr.addr32[0] > b->ext.addr.addr32[0])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a->ext.addr.addr32[0] < b->ext.addr.addr32[0])
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (-1);
break;
#endif /* INET6 */
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((diff = a->lan.port - b->lan.port) != 0)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (diff);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((diff = a->ext.port - b->ext.port) != 0)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (diff);
return (0);
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#ifdef __FreeBSD__
static int
#else
static __inline int
#endif
pf_state_compare_ext_gwy(struct pf_state *a, struct pf_state *b)
{
int diff;
if ((diff = a->proto - b->proto) != 0)
return (diff);
if ((diff = a->af - b->af) != 0)
return (diff);
switch (a->af) {
#ifdef INET
case AF_INET:
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 (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);
break;
#endif /* INET */
#ifdef INET6
case 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->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->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->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->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->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->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 (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);
break;
#endif /* INET6 */
}
if ((diff = a->ext.port - b->ext.port) != 0)
return (diff);
if ((diff = a->gwy.port - b->gwy.port) != 0)
return (diff);
return (0);
}
#ifdef __FreeBSD__
static int
#else
static __inline int
#endif
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);
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef INET6
void
pf_addrcpy(struct pf_addr *dst, struct pf_addr *src, sa_family_t af)
{
switch (af) {
#ifdef INET
case AF_INET:
dst->addr32[0] = src->addr32[0];
break;
#endif /* INET */
case AF_INET6:
dst->addr32[0] = src->addr32[0];
dst->addr32[1] = src->addr32[1];
dst->addr32[2] = src->addr32[2];
dst->addr32[3] = src->addr32[3];
break;
}
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* INET6 */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_state *
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_find_state_byid(struct pf_state_cmp *key)
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
{
pf_status.fcounters[FCNT_STATE_SEARCH]++;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
return (RB_FIND(pf_state_tree_id, &tree_id, (struct pf_state *)key));
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
struct pf_state *
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_find_state_recurse(struct pfi_kif *kif, struct pf_state_cmp *key, u_int8_t tree)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_state *s;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_status.fcounters[FCNT_STATE_SEARCH]++;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
switch (tree) {
case PF_LAN_EXT:
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if ((s = RB_FIND(pf_state_tree_lan_ext, &kif->pfik_lan_ext,
(struct pf_state *)key)) != NULL)
return (s);
if ((s = RB_FIND(pf_state_tree_lan_ext, &pfi_all->pfik_lan_ext,
(struct pf_state *)key)) != NULL)
return (s);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (NULL);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
case PF_EXT_GWY:
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if ((s = RB_FIND(pf_state_tree_ext_gwy, &kif->pfik_ext_gwy,
(struct pf_state *)key)) != NULL)
return (s);
if ((s = RB_FIND(pf_state_tree_ext_gwy, &pfi_all->pfik_ext_gwy,
(struct pf_state *)key)) != NULL)
return (s);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
return (NULL);
default:
panic("pf_find_state_recurse");
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_state *
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_find_state_all(struct pf_state_cmp *key, u_int8_t tree, int *more)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_state *s, *ss = NULL;
struct pfi_kif *kif;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_status.fcounters[FCNT_STATE_SEARCH]++;
switch (tree) {
case PF_LAN_EXT:
TAILQ_FOREACH(kif, &pfi_statehead, pfik_w_states) {
s = RB_FIND(pf_state_tree_lan_ext,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
&kif->pfik_lan_ext, (struct pf_state *)key);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (s == NULL)
continue;
if (more == NULL)
return (s);
ss = s;
(*more)++;
}
return (ss);
case PF_EXT_GWY:
TAILQ_FOREACH(kif, &pfi_statehead, pfik_w_states) {
s = RB_FIND(pf_state_tree_ext_gwy,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
&kif->pfik_ext_gwy, (struct pf_state *)key);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (s == NULL)
continue;
if (more == NULL)
return (s);
ss = s;
(*more)++;
}
return (ss);
default:
panic("pf_find_state_all");
}
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
void
pf_init_threshold(struct pf_threshold *threshold,
u_int32_t limit, u_int32_t seconds)
{
threshold->limit = limit * PF_THRESHOLD_MULT;
threshold->seconds = seconds;
threshold->count = 0;
threshold->last = time_second;
}
void
pf_add_threshold(struct pf_threshold *threshold)
{
u_int32_t t = time_second, diff = t - threshold->last;
if (diff >= threshold->seconds)
threshold->count = 0;
else
threshold->count -= threshold->count * diff /
threshold->seconds;
threshold->count += PF_THRESHOLD_MULT;
threshold->last = t;
}
int
pf_check_threshold(struct pf_threshold *threshold)
{
return (threshold->count > threshold->limit);
}
int
pf_src_connlimit(struct pf_state **state)
{
struct pf_state *s;
int bad = 0;
(*state)->src_node->conn++;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
(*state)->src.tcp_est = 1;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_add_threshold(&(*state)->src_node->conn_rate);
if ((*state)->rule.ptr->max_src_conn &&
(*state)->rule.ptr->max_src_conn <
(*state)->src_node->conn) {
pf_status.lcounters[LCNT_SRCCONN]++;
bad++;
}
if ((*state)->rule.ptr->max_src_conn_rate.limit &&
pf_check_threshold(&(*state)->src_node->conn_rate)) {
pf_status.lcounters[LCNT_SRCCONNRATE]++;
bad++;
}
if (!bad)
return (0);
if ((*state)->rule.ptr->overload_tbl) {
struct pfr_addr p;
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)->af);
}
bzero(&p, sizeof(p));
p.pfra_af = (*state)->af;
switch ((*state)->af) {
#ifdef INET
case AF_INET:
p.pfra_net = 32;
p.pfra_ip4addr = (*state)->src_node->addr.v4;
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
p.pfra_net = 128;
p.pfra_ip6addr = (*state)->src_node->addr.v6;
break;
#endif /* INET6 */
}
pfr_insert_kentry((*state)->rule.ptr->overload_tbl,
&p, time_second);
/* kill existing states if that's required. */
if ((*state)->rule.ptr->flush) {
pf_status.lcounters[LCNT_OVERLOAD_FLUSH]++;
RB_FOREACH(s, pf_state_tree_id, &tree_id) {
/*
* Kill states from this source. (Only those
* from the same rule if PF_FLUSH_GLOBAL is not
* set)
*/
if (s->af == (*state)->af &&
(((*state)->direction == PF_OUT &&
PF_AEQ(&(*state)->src_node->addr,
&s->lan.addr, s->af)) ||
((*state)->direction == PF_IN &&
PF_AEQ(&(*state)->src_node->addr,
&s->ext.addr, s->af))) &&
((*state)->rule.ptr->flush &
PF_FLUSH_GLOBAL ||
(*state)->rule.ptr == s->rule.ptr)) {
s->timeout = PFTM_PURGE;
s->src.state = s->dst.state =
TCPS_CLOSED;
killed++;
}
}
if (pf_status.debug >= PF_DEBUG_MISC)
printf(", %u states killed", killed);
}
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);
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
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;
if (*sn == NULL) {
k.af = af;
PF_ACPY(&k.addr, src, af);
if (rule->rule_flag & PFRULE_RULESRCTRACK ||
rule->rpool.opts & PF_POOL_STICKYADDR)
k.rule.ptr = rule;
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)
(*sn) = pool_get(&pf_src_tree_pl, PR_NOWAIT);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
else
pf_status.lcounters[LCNT_SRCNODES]++;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((*sn) == NULL)
return (-1);
bzero(*sn, sizeof(struct pf_src_node));
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_init_threshold(&(*sn)->conn_rate,
rule->max_src_conn_rate.limit,
rule->max_src_conn_rate.seconds);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
(*sn)->af = af;
if (rule->rule_flag & PFRULE_RULESRCTRACK ||
rule->rpool.opts & PF_POOL_STICKYADDR)
(*sn)->rule.ptr = rule;
else
(*sn)->rule.ptr = NULL;
PF_ACPY(&(*sn)->addr, src, af);
if (RB_INSERT(pf_src_tree,
&tree_src_tracking, *sn) != NULL) {
if (pf_status.debug >= PF_DEBUG_MISC) {
printf("pf: src_tree insert failed: ");
pf_print_host(&(*sn)->addr, 0, af);
printf("\n");
}
pool_put(&pf_src_tree_pl, *sn);
return (-1);
}
(*sn)->creation = time_second;
(*sn)->ruletype = rule->action;
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 &&
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
(*sn)->states >= rule->max_src_states) {
pf_status.lcounters[LCNT_SRCSTATES]++;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
return (-1);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
return (0);
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
int
pf_insert_state(struct pfi_kif *kif, struct pf_state *state)
{
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* Thou MUST NOT insert multiple duplicate keys */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
state->u.s.kif = kif;
if (RB_INSERT(pf_state_tree_lan_ext, &kif->pfik_lan_ext, state)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (pf_status.debug >= PF_DEBUG_MISC) {
printf("pf: state insert failed: tree_lan_ext");
printf(" lan: ");
pf_print_host(&state->lan.addr, state->lan.port,
state->af);
printf(" gwy: ");
pf_print_host(&state->gwy.addr, state->gwy.port,
state->af);
printf(" ext: ");
pf_print_host(&state->ext.addr, state->ext.port,
state->af);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (state->sync_flags & PFSTATE_FROMSYNC)
printf(" (from sync)");
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
printf("\n");
}
return (-1);
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (RB_INSERT(pf_state_tree_ext_gwy, &kif->pfik_ext_gwy, state)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (pf_status.debug >= PF_DEBUG_MISC) {
printf("pf: state insert failed: tree_ext_gwy");
printf(" lan: ");
pf_print_host(&state->lan.addr, state->lan.port,
state->af);
printf(" gwy: ");
pf_print_host(&state->gwy.addr, state->gwy.port,
state->af);
printf(" ext: ");
pf_print_host(&state->ext.addr, state->ext.port,
state->af);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (state->sync_flags & PFSTATE_FROMSYNC)
printf(" (from sync)");
printf("\n");
}
RB_REMOVE(pf_state_tree_lan_ext, &kif->pfik_lan_ext, state);
return (-1);
}
if (state->id == 0 && state->creatorid == 0) {
state->id = htobe64(pf_status.stateid++);
state->creatorid = pf_status.hostid;
}
if (RB_INSERT(pf_state_tree_id, &tree_id, state) != NULL) {
if (pf_status.debug >= PF_DEBUG_MISC) {
#ifdef __FreeBSD__
printf("pf: state insert failed: "
"id: %016llx creatorid: %08x",
(long long)be64toh(state->id),
ntohl(state->creatorid));
#else
printf("pf: state insert failed: "
"id: %016llx creatorid: %08x",
betoh64(state->id), ntohl(state->creatorid));
#endif
if (state->sync_flags & PFSTATE_FROMSYNC)
printf(" (from sync)");
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
printf("\n");
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
RB_REMOVE(pf_state_tree_lan_ext, &kif->pfik_lan_ext, state);
RB_REMOVE(pf_state_tree_ext_gwy, &kif->pfik_ext_gwy, state);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (-1);
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
TAILQ_INSERT_TAIL(&state_list, state, u.s.entry_list);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_status.fcounters[FCNT_STATE_INSERT]++;
pf_status.states++;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pfi_kif_ref(kif, PFI_KIF_REF_STATE);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#if NPFSYNC
pfsync_insert_state(state);
#endif
return (0);
}
void
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_purge_thread(void *v)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int nloops = 0, s;
for (;;) {
tsleep(pf_purge_thread, PWAIT, "pftm", 1 * hz);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
sx_slock(&pf_consistency_lock);
PF_LOCK();
if (pf_end_threads) {
pf_purge_expired_states(pf_status.states);
pf_purge_expired_fragments();
pf_purge_expired_src_nodes(0);
pf_end_threads++;
sx_sunlock(&pf_consistency_lock);
PF_UNLOCK();
wakeup(pf_purge_thread);
Rename the kthread_xxx (e.g. kthread_create()) calls to kproc_xxx as they actually make whole processes. Thos makes way for us to add REAL kthread_create() and friends that actually make theads. it turns out that most of these calls actually end up being moved back to the thread version when it's added. but we need to make this cosmetic change first. I'd LOVE to do this rename in 7.0 so that we can eventually MFC the new kthread_xxx() calls.
2007-10-20 23:23:23 +00:00
kproc_exit(0);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
}
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
s = splsoftnet();
/* process a fraction of the state table every second */
pf_purge_expired_states(1 + (pf_status.states
/ pf_default_rule.timeout[PFTM_INTERVAL]));
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
/* purge other expired types every PFTM_INTERVAL seconds */
if (++nloops >= pf_default_rule.timeout[PFTM_INTERVAL]) {
pf_purge_expired_fragments();
pf_purge_expired_src_nodes(0);
nloops = 0;
}
splx(s);
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
PF_UNLOCK();
sx_sunlock(&pf_consistency_lock);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
u_int32_t
pf_state_expires(const struct pf_state *state)
{
u_int32_t timeout;
u_int32_t start;
u_int32_t end;
u_int32_t states;
/* handle all PFTM_* > PFTM_MAX here */
if (state->timeout == PFTM_PURGE)
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
return (time_second);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (state->timeout == PFTM_UNTIL_PACKET)
return (0);
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
KASSERT(state->timeout != PFTM_UNLINKED,
("pf_state_expires: timeout == PFTM_UNLINKED"));
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
KASSERT((state->timeout < PFTM_MAX),
("pf_state_expires: timeout > PFTM_MAX"));
#else
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
KASSERT(state->timeout != PFTM_UNLINKED);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
KASSERT(state->timeout < PFTM_MAX);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
timeout = state->rule.ptr->timeout[state->timeout];
if (!timeout)
timeout = pf_default_rule.timeout[state->timeout];
start = state->rule.ptr->timeout[PFTM_ADAPTIVE_START];
if (start) {
end = state->rule.ptr->timeout[PFTM_ADAPTIVE_END];
states = state->rule.ptr->states;
} else {
start = pf_default_rule.timeout[PFTM_ADAPTIVE_START];
end = pf_default_rule.timeout[PFTM_ADAPTIVE_END];
states = pf_status.states;
}
if (end && states > start && start < end) {
if (states < end)
return (state->expire + timeout * (end - states) /
(end - start));
else
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
return (time_second);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
return (state->expire + timeout);
}
void
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_purge_expired_src_nodes(int waslocked)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_src_node *cur, *next;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int locked = waslocked;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
for (cur = RB_MIN(pf_src_tree, &tree_src_tracking); cur; cur = next) {
next = RB_NEXT(pf_src_tree, &tree_src_tracking, cur);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (cur->states <= 0 && cur->expire <= time_second) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (! locked) {
#ifdef __FreeBSD__
if (!sx_try_upgrade(&pf_consistency_lock)) {
PF_UNLOCK();
sx_sunlock(&pf_consistency_lock);
sx_xlock(&pf_consistency_lock);
PF_LOCK();
}
#else
rw_enter_write(&pf_consistency_lock);
#endif
next = RB_NEXT(pf_src_tree,
&tree_src_tracking, cur);
locked = 1;
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (cur->rule.ptr != NULL) {
cur->rule.ptr->src_nodes--;
if (cur->rule.ptr->states <= 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]++;
pf_status.src_nodes--;
pool_put(&pf_src_tree_pl, cur);
}
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (locked && !waslocked)
#ifdef __FreeBSD__
sx_downgrade(&pf_consistency_lock);
#else
rw_exit_write(&pf_consistency_lock);
#endif
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
void
pf_src_tree_remove_state(struct pf_state *s)
{
u_int32_t timeout;
if (s->src_node != NULL) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (s->proto == IPPROTO_TCP) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (s->src.tcp_est)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
--s->src_node->conn;
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (--s->src_node->states <= 0) {
timeout = s->rule.ptr->timeout[PFTM_SRC_NODE];
if (!timeout)
timeout =
pf_default_rule.timeout[PFTM_SRC_NODE];
s->src_node->expire = time_second + timeout;
}
}
if (s->nat_src_node != s->src_node && s->nat_src_node != NULL) {
if (--s->nat_src_node->states <= 0) {
timeout = s->rule.ptr->timeout[PFTM_SRC_NODE];
if (!timeout)
timeout =
pf_default_rule.timeout[PFTM_SRC_NODE];
s->nat_src_node->expire = time_second + timeout;
}
}
s->src_node = s->nat_src_node = NULL;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
/* callers should be at splsoftnet */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
void
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_unlink_state(struct pf_state *cur)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
{
Prevent a race condition. As pf_send_tcp() - called for expired synproxy states - has to drop the lock when calling back to ip_output(), the state purge timeout might run and gc the state. This results in a rb-tree inconsistency. With this change we flag expiring states while holding the lock and back off if the flag is already set. Reported by: glebius MFC after: 2 weeks
2005-07-20 18:58:27 +00:00
#ifdef __FreeBSD__
Move PFSTATE_EXPIRING from sync_flags to a new local_flags. sync_flags has special handling when zero. This caused no PFSYNC_ACT_DEL message and thus disfunction of pfflowd and state synchronisation in general. Discovered by: thompsa Good catch by: thompsa MFC after: 7 days
2005-12-20 00:33:33 +00:00
if (cur->local_flags & PFSTATE_EXPIRING)
Prevent a race condition. As pf_send_tcp() - called for expired synproxy states - has to drop the lock when calling back to ip_output(), the state purge timeout might run and gc the state. This results in a rb-tree inconsistency. With this change we flag expiring states while holding the lock and back off if the flag is already set. Reported by: glebius MFC after: 2 weeks
2005-07-20 18:58:27 +00:00
return;
Move PFSTATE_EXPIRING from sync_flags to a new local_flags. sync_flags has special handling when zero. This caused no PFSYNC_ACT_DEL message and thus disfunction of pfflowd and state synchronisation in general. Discovered by: thompsa Good catch by: thompsa MFC after: 7 days
2005-12-20 00:33:33 +00:00
cur->local_flags |= PFSTATE_EXPIRING;
Prevent a race condition. As pf_send_tcp() - called for expired synproxy states - has to drop the lock when calling back to ip_output(), the state purge timeout might run and gc the state. This results in a rb-tree inconsistency. With this change we flag expiring states while holding the lock and back off if the flag is already set. Reported by: glebius MFC after: 2 weeks
2005-07-20 18:58:27 +00:00
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (cur->src.state == PF_TCPS_PROXY_DST) {
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#ifdef __FreeBSD__
pf_send_tcp(NULL, cur->rule.ptr, cur->af,
#else
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_send_tcp(cur->rule.ptr, cur->af,
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#endif
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
&cur->ext.addr, &cur->lan.addr,
cur->ext.port, cur->lan.port,
cur->src.seqhi, cur->src.seqlo + 1,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
TH_RST|TH_ACK, 0, 0, 0, 1, cur->tag, NULL, NULL);
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
RB_REMOVE(pf_state_tree_ext_gwy,
&cur->u.s.kif->pfik_ext_gwy, cur);
RB_REMOVE(pf_state_tree_lan_ext,
&cur->u.s.kif->pfik_lan_ext, cur);
RB_REMOVE(pf_state_tree_id, &tree_id, cur);
#if NPFSYNC
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (cur->creatorid == pf_status.hostid)
pfsync_delete_state(cur);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
cur->timeout = PFTM_UNLINKED;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_src_tree_remove_state(cur);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
}
/* callers should be at splsoftnet and hold the
* write_lock on pf_consistency_lock */
void
pf_free_state(struct pf_state *cur)
{
#if NPFSYNC
if (pfsyncif != NULL &&
(pfsyncif->sc_bulk_send_next == cur ||
pfsyncif->sc_bulk_terminator == cur))
return;
#endif
#ifdef __FreeBSD__
KASSERT(cur->timeout == PFTM_UNLINKED,
("pf_free_state: cur->timeout != PFTM_UNLINKED"));
#else
KASSERT(cur->timeout == PFTM_UNLINKED);
#endif
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (--cur->rule.ptr->states <= 0 &&
cur->rule.ptr->src_nodes <= 0)
pf_rm_rule(NULL, cur->rule.ptr);
if (cur->nat_rule.ptr != NULL)
if (--cur->nat_rule.ptr->states <= 0 &&
cur->nat_rule.ptr->src_nodes <= 0)
pf_rm_rule(NULL, cur->nat_rule.ptr);
if (cur->anchor.ptr != NULL)
if (--cur->anchor.ptr->states <= 0)
pf_rm_rule(NULL, cur->anchor.ptr);
pf_normalize_tcp_cleanup(cur);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pfi_kif_unref(cur->u.s.kif, PFI_KIF_REF_STATE);
TAILQ_REMOVE(&state_list, cur, u.s.entry_list);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (cur->tag)
pf_tag_unref(cur->tag);
pool_put(&pf_state_pl, cur);
pf_status.fcounters[FCNT_STATE_REMOVALS]++;
pf_status.states--;
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
void
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_purge_expired_states(u_int32_t maxcheck)
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
{
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
static struct pf_state *cur = NULL;
struct pf_state *next;
int locked = 0;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
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 */
}
/* get next state, as cur may get deleted */
next = TAILQ_NEXT(cur, u.s.entry_list);
if (cur->timeout == PFTM_UNLINKED) {
/* free unlinked state */
if (! locked) {
#ifdef __FreeBSD__
if (!sx_try_upgrade(&pf_consistency_lock)) {
PF_UNLOCK();
sx_sunlock(&pf_consistency_lock);
sx_xlock(&pf_consistency_lock);
PF_LOCK();
}
#else
rw_enter_write(&pf_consistency_lock);
#endif
locked = 1;
}
pf_free_state(cur);
} else if (pf_state_expires(cur) <= time_second) {
/* unlink and free expired state */
pf_unlink_state(cur);
if (! locked) {
#ifdef __FreeBSD__
if (!sx_try_upgrade(&pf_consistency_lock)) {
PF_UNLOCK();
sx_sunlock(&pf_consistency_lock);
sx_xlock(&pf_consistency_lock);
PF_LOCK();
}
#else
rw_enter_write(&pf_consistency_lock);
#endif
locked = 1;
}
pf_free_state(cur);
}
cur = next;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (locked)
#ifdef __FreeBSD__
sx_downgrade(&pf_consistency_lock);
#else
rw_exit_write(&pf_consistency_lock);
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
int
pf_tbladdr_setup(struct pf_ruleset *rs, struct pf_addr_wrap *aw)
{
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)
{
if (aw->type != PF_ADDR_TABLE || aw->p.tbl == NULL)
return;
pfr_detach_table(aw->p.tbl);
aw->p.tbl = NULL;
}
void
pf_tbladdr_copyout(struct pf_addr_wrap *aw)
{
struct pfr_ktable *kt = aw->p.tbl;
if (aw->type != PF_ADDR_TABLE || kt == NULL)
return;
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;
}
void
pf_print_host(struct pf_addr *addr, u_int16_t p, sa_family_t af)
{
switch (af) {
#ifdef 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);
if (p) {
p = ntohs(p);
printf(":%u", p);
}
break;
}
#endif /* INET */
#ifdef INET6
case AF_INET6: {
u_int16_t b;
u_int8_t i, curstart = 255, curend = 0,
maxstart = 0, maxend = 0;
for (i = 0; i < 8; i++) {
if (!addr->addr16[i]) {
if (curstart == 255)
curstart = i;
else
curend = i;
} else {
if (curstart) {
if ((curend - curstart) >
(maxend - maxstart)) {
maxstart = curstart;
maxend = curend;
curstart = 255;
}
}
}
}
for (i = 0; i < 8; i++) {
if (i >= maxstart && i <= maxend) {
if (maxend != 7) {
if (i == maxstart)
printf(":");
} else {
if (i == maxend)
printf(":");
}
} else {
b = ntohs(addr->addr16[i]);
printf("%x", b);
if (i < 7)
printf(":");
}
}
if (p) {
p = ntohs(p);
printf("[%u]", p);
}
break;
}
#endif /* INET6 */
}
}
void
pf_print_state(struct pf_state *s)
{
switch (s->proto) {
case IPPROTO_TCP:
printf("TCP ");
break;
case IPPROTO_UDP:
printf("UDP ");
break;
case IPPROTO_ICMP:
printf("ICMP ");
break;
case IPPROTO_ICMPV6:
printf("ICMPV6 ");
break;
default:
printf("%u ", s->proto);
break;
}
pf_print_host(&s->lan.addr, s->lan.port, s->af);
printf(" ");
pf_print_host(&s->gwy.addr, s->gwy.port, s->af);
printf(" ");
pf_print_host(&s->ext.addr, s->ext.port, s->af);
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)
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)
printf(" wscale=%u", s->dst.wscale & PF_WSCALE_MASK);
printf("]");
printf(" %u:%u", s->src.state, s->dst.state);
}
void
pf_print_flags(u_int8_t f)
{
if (f)
printf(" ");
if (f & TH_FIN)
printf("F");
if (f & TH_SYN)
printf("S");
if (f & TH_RST)
printf("R");
if (f & TH_PUSH)
printf("P");
if (f & TH_ACK)
printf("A");
if (f & TH_URG)
printf("U");
if (f & TH_ECE)
printf("E");
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); \
} \
} while (0)
void
pf_calc_skip_steps(struct pf_rulequeue *rules)
{
struct pf_rule *cur, *prev, *head[PF_SKIP_COUNT];
int i;
cur = TAILQ_FIRST(rules);
prev = cur;
for (i = 0; i < PF_SKIP_COUNT; ++i)
head[i] = cur;
while (cur != NULL) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (cur->kif != prev->kif || cur->ifnot != prev->ifnot)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
PF_SET_SKIP_STEPS(PF_SKIP_IFP);
if (cur->direction != prev->direction)
PF_SET_SKIP_STEPS(PF_SKIP_DIR);
if (cur->af != prev->af)
PF_SET_SKIP_STEPS(PF_SKIP_AF);
if (cur->proto != prev->proto)
PF_SET_SKIP_STEPS(PF_SKIP_PROTO);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (cur->src.neg != prev->src.neg ||
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_addr_wrap_neq(&cur->src.addr, &prev->src.addr))
PF_SET_SKIP_STEPS(PF_SKIP_SRC_ADDR);
if (cur->src.port[0] != prev->src.port[0] ||
cur->src.port[1] != prev->src.port[1] ||
cur->src.port_op != prev->src.port_op)
PF_SET_SKIP_STEPS(PF_SKIP_SRC_PORT);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (cur->dst.neg != prev->dst.neg ||
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_addr_wrap_neq(&cur->dst.addr, &prev->dst.addr))
PF_SET_SKIP_STEPS(PF_SKIP_DST_ADDR);
if (cur->dst.port[0] != prev->dst.port[0] ||
cur->dst.port[1] != prev->dst.port[1] ||
cur->dst.port_op != prev->dst.port_op)
PF_SET_SKIP_STEPS(PF_SKIP_DST_PORT);
prev = cur;
cur = TAILQ_NEXT(cur, entries);
}
for (i = 0; i < PF_SKIP_COUNT; ++i)
PF_SET_SKIP_STEPS(i);
}
int
pf_addr_wrap_neq(struct pf_addr_wrap *aw1, struct pf_addr_wrap *aw2)
{
if (aw1->type != aw2->type)
return (1);
switch (aw1->type) {
case PF_ADDR_ADDRMASK:
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:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
return (aw1->p.dyn->pfid_kt != aw2->p.dyn->pfid_kt);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
case PF_ADDR_NOROUTE:
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
case PF_ADDR_URPFFAILED:
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (0);
case PF_ADDR_TABLE:
return (aw1->p.tbl != aw2->p.tbl);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
case PF_ADDR_RTLABEL:
return (aw1->v.rtlabel != aw2->v.rtlabel);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
default:
printf("invalid address type: %d\n", aw1->type);
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 (0x0000);
l = cksum + old - new;
l = (l >> 16) + (l & 65535);
l = l & 65535;
if (udp && !l)
return (0xFFFF);
return (l);
}
void
pf_change_ap(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)
{
struct pf_addr ao;
u_int16_t po = *p;
PF_ACPY(&ao, a, af);
PF_ACPY(a, an, af);
*p = pn;
switch (af) {
#ifdef 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;
*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;
#endif /* INET */
#ifdef INET6
case AF_INET6:
*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;
#endif /* INET6 */
}
}
/* Changes a u_int32_t. Uses a void * so there are no align restrictions */
void
pf_change_a(void *a, u_int16_t *c, u_int32_t an, u_int8_t u)
{
u_int32_t ao;
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);
}
#ifdef INET6
void
pf_change_a6(struct pf_addr *a, u_int16_t *c, struct pf_addr *an, u_int8_t u)
{
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),
ao.addr16[6], an->addr16[6], u),
ao.addr16[7], an->addr16[7], u);
}
#endif /* INET6 */
void
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;
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;
Shut up a couple of annoying gcc warnings. Do not enclose the fixes with #ifdefs in order to loop it back to OpenBSD after the next import. There are a some implicit asserts involved which might be better spelled out explicitly (af == AF_INET ...) Approved by: bms(mentor)
2004-03-30 18:28:30 +00:00
u_int32_t opc = 0; /* make the compiler happy */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (pc != NULL)
opc = *pc;
*ip = np;
if (pc != NULL)
*pc = pf_cksum_fixup(*pc, oip, *ip, u);
*ic = pf_cksum_fixup(*ic, oip, *ip, 0);
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) {
#ifdef INET
case AF_INET: {
u_int32_t oh2c = *h2c;
*h2c = pf_cksum_fixup(pf_cksum_fixup(*h2c,
oia.addr16[0], ia->addr16[0], 0),
oia.addr16[1], ia->addr16[1], 0);
*ic = pf_cksum_fixup(pf_cksum_fixup(*ic,
oia.addr16[0], ia->addr16[0], 0),
oia.addr16[1], ia->addr16[1], 0);
*ic = pf_cksum_fixup(*ic, oh2c, *h2c, 0);
break;
}
#endif /* INET */
#ifdef 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),
oia.addr16[6], ia->addr16[6], u),
oia.addr16[7], ia->addr16[7], u);
break;
#endif /* INET6 */
}
/* Change outer ip address, fix outer ip or icmpv6 checksum. */
PF_ACPY(oa, na, af);
switch (af) {
#ifdef INET
case AF_INET:
*hc = pf_cksum_fixup(pf_cksum_fixup(*hc,
ooa.addr16[0], oa->addr16[0], 0),
ooa.addr16[1], oa->addr16[1], 0);
break;
#endif /* INET */
#ifdef 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),
ooa.addr16[6], oa->addr16[6], u),
ooa.addr16[7], oa->addr16[7], u);
break;
#endif /* INET6 */
}
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
/*
* Need to modulate the sequence numbers in the TCP SACK option
* (credits to Krzysztof Pfaff for report and patch)
*/
int
pf_modulate_sack(struct mbuf *m, int off, struct pf_pdesc *pd,
struct tcphdr *th, struct pf_state_peer *dst)
{
int hlen = (th->th_off << 2) - sizeof(*th), thoptlen = hlen;
#ifdef __FreeBSD__
u_int8_t opts[TCP_MAXOLEN], *opt = opts;
#else
u_int8_t opts[MAX_TCPOPTLEN], *opt = opts;
#endif
int copyback = 0, i, olen;
struct sackblk sack;
#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;
while (hlen >= TCPOLEN_SACKLEN) {
olen = opt[1];
switch (*opt) {
case TCPOPT_EOL: /* FALLTHROUGH */
case TCPOPT_NOP:
opt++;
hlen--;
break;
case TCPOPT_SACK:
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));
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));
}
copyback = 1;
}
/* FALLTHROUGH */
default:
if (olen < 2)
olen = 2;
hlen -= olen;
opt += olen;
}
}
if (copyback)
#ifdef __FreeBSD__
m_copyback(m, off + sizeof(*th), thoptlen, (caddr_t)opts);
#else
m_copyback(m, off + sizeof(*th), thoptlen, opts);
#endif
return (copyback);
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
void
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#ifdef __FreeBSD__
pf_send_tcp(struct mbuf *replyto, const struct pf_rule *r, sa_family_t af,
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_send_tcp(const struct pf_rule *r, sa_family_t af,
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
const struct pf_addr *saddr, const struct pf_addr *daddr,
u_int16_t sport, u_int16_t dport, u_int32_t seq, u_int32_t ack,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
u_int8_t flags, u_int16_t win, u_int16_t mss, u_int8_t ttl, int tag,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
u_int16_t rtag, struct ether_header *eh, struct ifnet *ifp)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
struct mbuf *m;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int len, tlen;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef INET
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct ip *h;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#endif /* INET */
#ifdef INET6
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct ip6_hdr *h6;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#endif /* INET6 */
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct tcphdr *th;
char *opt;
struct pf_mtag *pf_mtag;
#ifdef __FreeBSD__
KASSERT(
#ifdef INET
af == AF_INET
#else
0
#endif
||
#ifdef INET6
af == AF_INET6
#else
0
#endif
, ("Unsupported AF %d", af));
len = 0;
th = NULL;
#ifdef INET
h = NULL;
#endif
#ifdef INET6
h6 = NULL;
#endif
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* maximum segment size tcp option */
tlen = sizeof(struct tcphdr);
if (mss)
tlen += 4;
switch (af) {
#ifdef INET
case AF_INET:
len = sizeof(struct ip) + tlen;
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
len = sizeof(struct ip6_hdr) + tlen;
break;
#endif /* INET6 */
}
/* create outgoing mbuf */
Define semantic of M_SKIP_FIREWALL more precisely, i.e. also pass associated icmp_error() packets. While here retire PACKET_TAG_PF_GENERATED (which served the same purpose) and use M_SKIP_FIREWALL in pf as well. This should speed up things a bit as we get rid of the tag allocations. Discussed with: juli
2004-07-17 05:10:06 +00:00
m = m_gethdr(M_DONTWAIT, MT_HEADER);
if (m == NULL)
return;
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#ifdef __FreeBSD__
#ifdef MAC
if (replyto)
Continue to move from generic network entry points in the TrustedBSD MAC Framework by moving from mac_mbuf_create_netlayer() to more specific entry points for specific network services: - mac_netinet_firewall_reply() to be used when replying to in-bound TCP segments in pf and ipfw (etc). - Rename mac_netinet_icmp_reply() to mac_netinet_icmp_replyinplace() and add mac_netinet_icmp_reply(), reflecting that in some cases we overwrite a label in place, but in others we apply the label to a new mbuf. Obtained from: TrustedBSD Project
2007-10-28 17:12:48 +00:00
mac_netinet_firewall_reply(replyto, m);
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
else
Rename 'mac_mbuf_create_from_firewall' to 'mac_netinet_firewall_send' as we move towards netinet as a pseudo-object for the MAC Framework. Rename 'mac_create_mbuf_linklayer' to 'mac_mbuf_create_linklayer' to reflect general object-first ordering preference. Sponsored by: SPARTA (original patches against Mac OS X) Obtained from: TrustedBSD Project, Apple Computer
2007-10-26 13:18:38 +00:00
mac_netinet_firewall_send(m);
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#else
(void)replyto;
#endif
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if ((pf_mtag = pf_get_mtag(m)) == NULL) {
m_freem(m);
return;
}
if (tag)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#ifdef __FreeBSD__
m->m_flags |= M_SKIP_FIREWALL;
Define semantic of M_SKIP_FIREWALL more precisely, i.e. also pass associated icmp_error() packets. While here retire PACKET_TAG_PF_GENERATED (which served the same purpose) and use M_SKIP_FIREWALL in pf as well. This should speed up things a bit as we get rid of the tag allocations. Discussed with: juli
2004-07-17 05:10:06 +00:00
#else
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_mtag->flags |= PF_TAG_GENERATED;
Define semantic of M_SKIP_FIREWALL more precisely, i.e. also pass associated icmp_error() packets. While here retire PACKET_TAG_PF_GENERATED (which served the same purpose) and use M_SKIP_FIREWALL in pf as well. This should speed up things a bit as we get rid of the tag allocations. Discussed with: juli
2004-07-17 05:10:06 +00:00
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_mtag->tag = rtag;
if (r != NULL && r->rtableid >= 0)
Add code to allow the system to handle multiple routing tables. This particular implementation is designed to be fully backwards compatible and to be MFC-able to 7.x (and 6.x) Currently the only protocol that can make use of the multiple tables is IPv4 Similar functionality exists in OpenBSD and Linux. From my notes: ----- One thing where FreeBSD has been falling behind, and which by chance I have some time to work on is "policy based routing", which allows different packet streams to be routed by more than just the destination address. Constraints: ------------ I want to make some form of this available in the 6.x tree (and by extension 7.x) , but FreeBSD in general needs it so I might as well do it in -current and back port the portions I need. One of the ways that this can be done is to have the ability to instantiate multiple kernel routing tables (which I will now refer to as "Forwarding Information Bases" or "FIBs" for political correctness reasons). Which FIB a particular packet uses to make the next hop decision can be decided by a number of mechanisms. The policies these mechanisms implement are the "Policies" referred to in "Policy based routing". One of the constraints I have if I try to back port this work to 6.x is that it must be implemented as a EXTENSION to the existing ABIs in 6.x so that third party applications do not need to be recompiled in timespan of the branch. This first version will not have some of the bells and whistles that will come with later versions. It will, for example, be limited to 16 tables in the first commit. Implementation method, Compatible version. (part 1) ------------------------------- For this reason I have implemented a "sufficient subset" of a multiple routing table solution in Perforce, and back-ported it to 6.x. (also in Perforce though not always caught up with what I have done in -current/P4). The subset allows a number of FIBs to be defined at compile time (8 is sufficient for my purposes in 6.x) and implements the changes needed to allow IPV4 to use them. I have not done the changes for ipv6 simply because I do not need it, and I do not have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it. Other protocol families are left untouched and should there be users with proprietary protocol families, they should continue to work and be oblivious to the existence of the extra FIBs. To understand how this is done, one must know that the current FIB code starts everything off with a single dimensional array of pointers to FIB head structures (One per protocol family), each of which in turn points to the trie of routes available to that family. The basic change in the ABI compatible version of the change is to extent that array to be a 2 dimensional array, so that instead of protocol family X looking at rt_tables[X] for the table it needs, it looks at rt_tables[Y][X] when for all protocol families except ipv4 Y is always 0. Code that is unaware of the change always just sees the first row of the table, which of course looks just like the one dimensional array that existed before. The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign() are all maintained, but refer only to the first row of the array, so that existing callers in proprietary protocols can continue to do the "right thing". Some new entry points are added, for the exclusive use of ipv4 code called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(), which have an extra argument which refers the code to the correct row. In addition, there are some new entry points (currently called rtalloc_fib() and friends) that check the Address family being looked up and call either rtalloc() (and friends) if the protocol is not IPv4 forcing the action to row 0 or to the appropriate row if it IS IPv4 (and that info is available). These are for calling from code that is not specific to any particular protocol. The way these are implemented would change in the non ABI preserving code to be added later. One feature of the first version of the code is that for ipv4, the interface routes show up automatically on all the FIBs, so that no matter what FIB you select you always have the basic direct attached hosts available to you. (rtinit() does this automatically). You CAN delete an interface route from one FIB should you want to but by default it's there. ARP information is also available in each FIB. It's assumed that the same machine would have the same MAC address, regardless of which FIB you are using to get to it. This brings us as to how the correct FIB is selected for an outgoing IPV4 packet. Firstly, all packets have a FIB associated with them. if nothing has been done to change it, it will be FIB 0. The FIB is changed in the following ways. Packets fall into one of a number of classes. 1/ locally generated packets, coming from a socket/PCB. Such packets select a FIB from a number associated with the socket/PCB. This in turn is inherited from the process, but can be changed by a socket option. The process in turn inherits it on fork. I have written a utility call setfib that acts a bit like nice.. setfib -3 ping target.example.com # will use fib 3 for ping. It is an obvious extension to make it a property of a jail but I have not done so. It can be achieved by combining the setfib and jail commands. 2/ packets received on an interface for forwarding. By default these packets would use table 0, (or possibly a number settable in a sysctl(not yet)). but prior to routing the firewall can inspect them (see below). (possibly in the future you may be able to associate a FIB with packets received on an interface.. An ifconfig arg, but not yet.) 3/ packets inspected by a packet classifier, which can arbitrarily associate a fib with it on a packet by packet basis. A fib assigned to a packet by a packet classifier (such as ipfw) would over-ride a fib associated by a more default source. (such as cases 1 or 2). 4/ a tcp listen socket associated with a fib will generate accept sockets that are associated with that same fib. 5/ Packets generated in response to some other packet (e.g. reset or icmp packets). These should use the FIB associated with the packet being reponded to. 6/ Packets generated during encapsulation. gif, tun and other tunnel interfaces will encapsulate using the FIB that was in effect withthe proces that set up the tunnel. thus setfib 1 ifconfig gif0 [tunnel instructions] will set the fib for the tunnel to use to be fib 1. Routing messages would be associated with their process, and thus select one FIB or another. messages from the kernel would be associated with the fib they refer to and would only be received by a routing socket associated with that fib. (not yet implemented) In addition Netstat has been edited to be able to cope with the fact that the array is now 2 dimensional. (It looks in system memory using libkvm (!)). Old versions of netstat see only the first FIB. In addition two sysctls are added to give: a) the number of FIBs compiled in (active) b) the default FIB of the calling process. Early testing experience: ------------------------- Basically our (IronPort's) appliance does this functionality already using ipfw fwd but that method has some drawbacks. For example, It can't fully simulate a routing table because it can't influence the socket's choice of local address when a connect() is done. Testing during the generating of these changes has been remarkably smooth so far. Multiple tables have co-existed with no notable side effects, and packets have been routes accordingly. ipfw has grown 2 new keywords: setfib N ip from anay to any count ip from any to any fib N In pf there seems to be a requirement to be able to give symbolic names to the fibs but I do not have that capacity. I am not sure if it is required. SCTP has interestingly enough built in support for this, called VRFs in Cisco parlance. it will be interesting to see how that handles it when it suddenly actually does something. Where to next: -------------------- After committing the ABI compatible version and MFCing it, I'd like to proceed in a forward direction in -current. this will result in some roto-tilling in the routing code. Firstly: the current code's idea of having a separate tree per protocol family, all of the same format, and pointed to by the 1 dimensional array is a bit silly. Especially when one considers that there is code that makes assumptions about every protocol having the same internal structures there. Some protocols don't WANT that sort of structure. (for example the whole idea of a netmask is foreign to appletalk). This needs to be made opaque to the external code. My suggested first change is to add routing method pointers to the 'domain' structure, along with information pointing the data. instead of having an array of pointers to uniform structures, there would be an array pointing to the 'domain' structures for each protocol address domain (protocol family), and the methods this reached would be called. The methods would have an argument that gives FIB number, but the protocol would be free to ignore it. When the ABI can be changed it raises the possibilty of the addition of a fib entry into the "struct route". Currently, the structure contains the sockaddr of the desination, and the resulting fib entry. To make this work fully, one could add a fib number so that given an address and a fib, one can find the third element, the fib entry. Interaction with the ARP layer/ LL layer would need to be revisited as well. Qing Li has been working on this already. This work was sponsored by Ironport Systems/Cisco Reviewed by: several including rwatson, bz and mlair (parts each) Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
#ifdef __FreeBSD__
{
M_SETFIB(m, r->rtableid);
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_mtag->rtableid = r->rtableid;
Add code to allow the system to handle multiple routing tables. This particular implementation is designed to be fully backwards compatible and to be MFC-able to 7.x (and 6.x) Currently the only protocol that can make use of the multiple tables is IPv4 Similar functionality exists in OpenBSD and Linux. From my notes: ----- One thing where FreeBSD has been falling behind, and which by chance I have some time to work on is "policy based routing", which allows different packet streams to be routed by more than just the destination address. Constraints: ------------ I want to make some form of this available in the 6.x tree (and by extension 7.x) , but FreeBSD in general needs it so I might as well do it in -current and back port the portions I need. One of the ways that this can be done is to have the ability to instantiate multiple kernel routing tables (which I will now refer to as "Forwarding Information Bases" or "FIBs" for political correctness reasons). Which FIB a particular packet uses to make the next hop decision can be decided by a number of mechanisms. The policies these mechanisms implement are the "Policies" referred to in "Policy based routing". One of the constraints I have if I try to back port this work to 6.x is that it must be implemented as a EXTENSION to the existing ABIs in 6.x so that third party applications do not need to be recompiled in timespan of the branch. This first version will not have some of the bells and whistles that will come with later versions. It will, for example, be limited to 16 tables in the first commit. Implementation method, Compatible version. (part 1) ------------------------------- For this reason I have implemented a "sufficient subset" of a multiple routing table solution in Perforce, and back-ported it to 6.x. (also in Perforce though not always caught up with what I have done in -current/P4). The subset allows a number of FIBs to be defined at compile time (8 is sufficient for my purposes in 6.x) and implements the changes needed to allow IPV4 to use them. I have not done the changes for ipv6 simply because I do not need it, and I do not have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it. Other protocol families are left untouched and should there be users with proprietary protocol families, they should continue to work and be oblivious to the existence of the extra FIBs. To understand how this is done, one must know that the current FIB code starts everything off with a single dimensional array of pointers to FIB head structures (One per protocol family), each of which in turn points to the trie of routes available to that family. The basic change in the ABI compatible version of the change is to extent that array to be a 2 dimensional array, so that instead of protocol family X looking at rt_tables[X] for the table it needs, it looks at rt_tables[Y][X] when for all protocol families except ipv4 Y is always 0. Code that is unaware of the change always just sees the first row of the table, which of course looks just like the one dimensional array that existed before. The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign() are all maintained, but refer only to the first row of the array, so that existing callers in proprietary protocols can continue to do the "right thing". Some new entry points are added, for the exclusive use of ipv4 code called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(), which have an extra argument which refers the code to the correct row. In addition, there are some new entry points (currently called rtalloc_fib() and friends) that check the Address family being looked up and call either rtalloc() (and friends) if the protocol is not IPv4 forcing the action to row 0 or to the appropriate row if it IS IPv4 (and that info is available). These are for calling from code that is not specific to any particular protocol. The way these are implemented would change in the non ABI preserving code to be added later. One feature of the first version of the code is that for ipv4, the interface routes show up automatically on all the FIBs, so that no matter what FIB you select you always have the basic direct attached hosts available to you. (rtinit() does this automatically). You CAN delete an interface route from one FIB should you want to but by default it's there. ARP information is also available in each FIB. It's assumed that the same machine would have the same MAC address, regardless of which FIB you are using to get to it. This brings us as to how the correct FIB is selected for an outgoing IPV4 packet. Firstly, all packets have a FIB associated with them. if nothing has been done to change it, it will be FIB 0. The FIB is changed in the following ways. Packets fall into one of a number of classes. 1/ locally generated packets, coming from a socket/PCB. Such packets select a FIB from a number associated with the socket/PCB. This in turn is inherited from the process, but can be changed by a socket option. The process in turn inherits it on fork. I have written a utility call setfib that acts a bit like nice.. setfib -3 ping target.example.com # will use fib 3 for ping. It is an obvious extension to make it a property of a jail but I have not done so. It can be achieved by combining the setfib and jail commands. 2/ packets received on an interface for forwarding. By default these packets would use table 0, (or possibly a number settable in a sysctl(not yet)). but prior to routing the firewall can inspect them (see below). (possibly in the future you may be able to associate a FIB with packets received on an interface.. An ifconfig arg, but not yet.) 3/ packets inspected by a packet classifier, which can arbitrarily associate a fib with it on a packet by packet basis. A fib assigned to a packet by a packet classifier (such as ipfw) would over-ride a fib associated by a more default source. (such as cases 1 or 2). 4/ a tcp listen socket associated with a fib will generate accept sockets that are associated with that same fib. 5/ Packets generated in response to some other packet (e.g. reset or icmp packets). These should use the FIB associated with the packet being reponded to. 6/ Packets generated during encapsulation. gif, tun and other tunnel interfaces will encapsulate using the FIB that was in effect withthe proces that set up the tunnel. thus setfib 1 ifconfig gif0 [tunnel instructions] will set the fib for the tunnel to use to be fib 1. Routing messages would be associated with their process, and thus select one FIB or another. messages from the kernel would be associated with the fib they refer to and would only be received by a routing socket associated with that fib. (not yet implemented) In addition Netstat has been edited to be able to cope with the fact that the array is now 2 dimensional. (It looks in system memory using libkvm (!)). Old versions of netstat see only the first FIB. In addition two sysctls are added to give: a) the number of FIBs compiled in (active) b) the default FIB of the calling process. Early testing experience: ------------------------- Basically our (IronPort's) appliance does this functionality already using ipfw fwd but that method has some drawbacks. For example, It can't fully simulate a routing table because it can't influence the socket's choice of local address when a connect() is done. Testing during the generating of these changes has been remarkably smooth so far. Multiple tables have co-existed with no notable side effects, and packets have been routes accordingly. ipfw has grown 2 new keywords: setfib N ip from anay to any count ip from any to any fib N In pf there seems to be a requirement to be able to give symbolic names to the fibs but I do not have that capacity. I am not sure if it is required. SCTP has interestingly enough built in support for this, called VRFs in Cisco parlance. it will be interesting to see how that handles it when it suddenly actually does something. Where to next: -------------------- After committing the ABI compatible version and MFCing it, I'd like to proceed in a forward direction in -current. this will result in some roto-tilling in the routing code. Firstly: the current code's idea of having a separate tree per protocol family, all of the same format, and pointed to by the 1 dimensional array is a bit silly. Especially when one considers that there is code that makes assumptions about every protocol having the same internal structures there. Some protocols don't WANT that sort of structure. (for example the whole idea of a netmask is foreign to appletalk). This needs to be made opaque to the external code. My suggested first change is to add routing method pointers to the 'domain' structure, along with information pointing the data. instead of having an array of pointers to uniform structures, there would be an array pointing to the 'domain' structures for each protocol address domain (protocol family), and the methods this reached would be called. The methods would have an argument that gives FIB number, but the protocol would be free to ignore it. When the ABI can be changed it raises the possibilty of the addition of a fib entry into the "struct route". Currently, the structure contains the sockaddr of the desination, and the resulting fib entry. To make this work fully, one could add a fib number so that given an address and a fib, one can find the third element, the fib entry. Interaction with the ARP layer/ LL layer would need to be revisited as well. Qing Li has been working on this already. This work was sponsored by Ironport Systems/Cisco Reviewed by: several including rwatson, bz and mlair (parts each) Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
#ifdef __FreeBSD__
}
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef ALTQ
if (r != NULL && r->qid) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_mtag->qid = r->qid;
/* add hints for ecn */
pf_mtag->af = af;
pf_mtag->hdr = mtod(m, struct ip *);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* ALTQ */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
m->m_data += max_linkhdr;
m->m_pkthdr.len = m->m_len = len;
m->m_pkthdr.rcvif = NULL;
bzero(m->m_data, len);
switch (af) {
#ifdef INET
case AF_INET:
h = mtod(m, struct ip *);
/* 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;
th = (struct tcphdr *)((caddr_t)h + sizeof(struct ip));
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
h6 = mtod(m, struct ip6_hdr *);
/* 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));
th = (struct tcphdr *)((caddr_t)h6 + sizeof(struct ip6_hdr));
break;
#endif /* INET6 */
}
/* TCP header */
th->th_sport = sport;
th->th_dport = dport;
th->th_seq = htonl(seq);
th->th_ack = htonl(ack);
th->th_off = tlen >> 2;
th->th_flags = flags;
th->th_win = htons(win);
if (mss) {
opt = (char *)(th + 1);
opt[0] = TCPOPT_MAXSEG;
opt[1] = 4;
HTONS(mss);
bcopy((caddr_t)&mss, (caddr_t)(opt + 2), 2);
}
switch (af) {
#ifdef INET
case AF_INET:
/* TCP checksum */
th->th_sum = in_cksum(m, len);
/* Finish the IP header */
h->ip_v = 4;
h->ip_hl = sizeof(*h) >> 2;
h->ip_tos = IPTOS_LOWDELAY;
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Commit step 1 of the vimage project, (network stack) virtualization work done by Marko Zec (zec@). This is the first in a series of commits over the course of the next few weeks. Mark all uses of global variables to be virtualized with a V_ prefix. Use macros to map them back to their global names for now, so this is a NOP change only. We hope to have caught at least 85-90% of what is needed so we do not invalidate a lot of outstanding patches again. Obtained from: //depot/projects/vimage-commit2/... Reviewed by: brooks, des, ed, mav, julian, jamie, kris, rwatson, zec, ... (various people I forgot, different versions) md5 (with a bit of help) Sponsored by: NLnet Foundation, The FreeBSD Foundation X-MFC after: never V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
h->ip_off = V_path_mtu_discovery ? IP_DF : 0;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
h->ip_len = len;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
h->ip_off = htons(ip_mtudisc ? IP_DF : 0);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
h->ip_len = htons(len);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#endif
Commit step 1 of the vimage project, (network stack) virtualization work done by Marko Zec (zec@). This is the first in a series of commits over the course of the next few weeks. Mark all uses of global variables to be virtualized with a V_ prefix. Use macros to map them back to their global names for now, so this is a NOP change only. We hope to have caught at least 85-90% of what is needed so we do not invalidate a lot of outstanding patches again. Obtained from: //depot/projects/vimage-commit2/... Reviewed by: brooks, des, ed, mav, julian, jamie, kris, rwatson, zec, ... (various people I forgot, different versions) md5 (with a bit of help) Sponsored by: NLnet Foundation, The FreeBSD Foundation X-MFC after: never V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
h->ip_ttl = ttl ? ttl : V_ip_defttl;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
h->ip_sum = 0;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (eh == NULL) {
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
PF_UNLOCK();
ip_output(m, (void *)NULL, (void *)NULL, 0,
(void *)NULL, (void *)NULL);
PF_LOCK();
#else /* ! __FreeBSD__ */
ip_output(m, (void *)NULL, (void *)NULL, 0,
(void *)NULL, (void *)NULL);
#endif
} else {
struct route ro;
struct rtentry rt;
struct ether_header *e = (void *)ro.ro_dst.sa_data;
if (ifp == NULL) {
m_freem(m);
return;
}
rt.rt_ifp = ifp;
ro.ro_rt = &rt;
ro.ro_dst.sa_len = sizeof(ro.ro_dst);
ro.ro_dst.sa_family = pseudo_AF_HDRCMPLT;
bcopy(eh->ether_dhost, e->ether_shost, ETHER_ADDR_LEN);
bcopy(eh->ether_shost, e->ether_dhost, ETHER_ADDR_LEN);
e->ether_type = eh->ether_type;
#ifdef __FreeBSD__
PF_UNLOCK();
/* XXX_IMPORT: later */
ip_output(m, (void *)NULL, &ro, 0,
(void *)NULL, (void *)NULL);
PF_LOCK();
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#else /* ! __FreeBSD__ */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
ip_output(m, (void *)NULL, &ro, IP_ROUTETOETHER,
(void *)NULL, (void *)NULL);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
/* TCP checksum */
th->th_sum = in6_cksum(m, IPPROTO_TCP,
sizeof(struct ip6_hdr), tlen);
h6->ip6_vfc |= IPV6_VERSION;
h6->ip6_hlim = IPV6_DEFHLIM;
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_UNLOCK();
ip6_output(m, NULL, NULL, 0, NULL, NULL, NULL);
PF_LOCK();
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
ip6_output(m, NULL, NULL, 0, NULL, NULL);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET6 */
}
}
void
pf_send_icmp(struct mbuf *m, u_int8_t type, u_int8_t code, sa_family_t af,
struct pf_rule *r)
{
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_mtag *pf_mtag;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct mbuf *m0;
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
struct ip *ip;
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
m0 = m_copypacket(m, M_DONTWAIT);
Define semantic of M_SKIP_FIREWALL more precisely, i.e. also pass associated icmp_error() packets. While here retire PACKET_TAG_PF_GENERATED (which served the same purpose) and use M_SKIP_FIREWALL in pf as well. This should speed up things a bit as we get rid of the tag allocations. Discussed with: juli
2004-07-17 05:10:06 +00:00
if (m0 == NULL)
return;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
m0 = m_copy(m, 0, M_COPYALL);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#endif
if ((pf_mtag = pf_get_mtag(m0)) == NULL)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifdef __FreeBSD__
/* XXX: revisit */
m0->m_flags |= M_SKIP_FIREWALL;
#else
pf_mtag->flags |= PF_TAG_GENERATED;
Define semantic of M_SKIP_FIREWALL more precisely, i.e. also pass associated icmp_error() packets. While here retire PACKET_TAG_PF_GENERATED (which served the same purpose) and use M_SKIP_FIREWALL in pf as well. This should speed up things a bit as we get rid of the tag allocations. Discussed with: juli
2004-07-17 05:10:06 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r->rtableid >= 0)
Add code to allow the system to handle multiple routing tables. This particular implementation is designed to be fully backwards compatible and to be MFC-able to 7.x (and 6.x) Currently the only protocol that can make use of the multiple tables is IPv4 Similar functionality exists in OpenBSD and Linux. From my notes: ----- One thing where FreeBSD has been falling behind, and which by chance I have some time to work on is "policy based routing", which allows different packet streams to be routed by more than just the destination address. Constraints: ------------ I want to make some form of this available in the 6.x tree (and by extension 7.x) , but FreeBSD in general needs it so I might as well do it in -current and back port the portions I need. One of the ways that this can be done is to have the ability to instantiate multiple kernel routing tables (which I will now refer to as "Forwarding Information Bases" or "FIBs" for political correctness reasons). Which FIB a particular packet uses to make the next hop decision can be decided by a number of mechanisms. The policies these mechanisms implement are the "Policies" referred to in "Policy based routing". One of the constraints I have if I try to back port this work to 6.x is that it must be implemented as a EXTENSION to the existing ABIs in 6.x so that third party applications do not need to be recompiled in timespan of the branch. This first version will not have some of the bells and whistles that will come with later versions. It will, for example, be limited to 16 tables in the first commit. Implementation method, Compatible version. (part 1) ------------------------------- For this reason I have implemented a "sufficient subset" of a multiple routing table solution in Perforce, and back-ported it to 6.x. (also in Perforce though not always caught up with what I have done in -current/P4). The subset allows a number of FIBs to be defined at compile time (8 is sufficient for my purposes in 6.x) and implements the changes needed to allow IPV4 to use them. I have not done the changes for ipv6 simply because I do not need it, and I do not have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it. Other protocol families are left untouched and should there be users with proprietary protocol families, they should continue to work and be oblivious to the existence of the extra FIBs. To understand how this is done, one must know that the current FIB code starts everything off with a single dimensional array of pointers to FIB head structures (One per protocol family), each of which in turn points to the trie of routes available to that family. The basic change in the ABI compatible version of the change is to extent that array to be a 2 dimensional array, so that instead of protocol family X looking at rt_tables[X] for the table it needs, it looks at rt_tables[Y][X] when for all protocol families except ipv4 Y is always 0. Code that is unaware of the change always just sees the first row of the table, which of course looks just like the one dimensional array that existed before. The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign() are all maintained, but refer only to the first row of the array, so that existing callers in proprietary protocols can continue to do the "right thing". Some new entry points are added, for the exclusive use of ipv4 code called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(), which have an extra argument which refers the code to the correct row. In addition, there are some new entry points (currently called rtalloc_fib() and friends) that check the Address family being looked up and call either rtalloc() (and friends) if the protocol is not IPv4 forcing the action to row 0 or to the appropriate row if it IS IPv4 (and that info is available). These are for calling from code that is not specific to any particular protocol. The way these are implemented would change in the non ABI preserving code to be added later. One feature of the first version of the code is that for ipv4, the interface routes show up automatically on all the FIBs, so that no matter what FIB you select you always have the basic direct attached hosts available to you. (rtinit() does this automatically). You CAN delete an interface route from one FIB should you want to but by default it's there. ARP information is also available in each FIB. It's assumed that the same machine would have the same MAC address, regardless of which FIB you are using to get to it. This brings us as to how the correct FIB is selected for an outgoing IPV4 packet. Firstly, all packets have a FIB associated with them. if nothing has been done to change it, it will be FIB 0. The FIB is changed in the following ways. Packets fall into one of a number of classes. 1/ locally generated packets, coming from a socket/PCB. Such packets select a FIB from a number associated with the socket/PCB. This in turn is inherited from the process, but can be changed by a socket option. The process in turn inherits it on fork. I have written a utility call setfib that acts a bit like nice.. setfib -3 ping target.example.com # will use fib 3 for ping. It is an obvious extension to make it a property of a jail but I have not done so. It can be achieved by combining the setfib and jail commands. 2/ packets received on an interface for forwarding. By default these packets would use table 0, (or possibly a number settable in a sysctl(not yet)). but prior to routing the firewall can inspect them (see below). (possibly in the future you may be able to associate a FIB with packets received on an interface.. An ifconfig arg, but not yet.) 3/ packets inspected by a packet classifier, which can arbitrarily associate a fib with it on a packet by packet basis. A fib assigned to a packet by a packet classifier (such as ipfw) would over-ride a fib associated by a more default source. (such as cases 1 or 2). 4/ a tcp listen socket associated with a fib will generate accept sockets that are associated with that same fib. 5/ Packets generated in response to some other packet (e.g. reset or icmp packets). These should use the FIB associated with the packet being reponded to. 6/ Packets generated during encapsulation. gif, tun and other tunnel interfaces will encapsulate using the FIB that was in effect withthe proces that set up the tunnel. thus setfib 1 ifconfig gif0 [tunnel instructions] will set the fib for the tunnel to use to be fib 1. Routing messages would be associated with their process, and thus select one FIB or another. messages from the kernel would be associated with the fib they refer to and would only be received by a routing socket associated with that fib. (not yet implemented) In addition Netstat has been edited to be able to cope with the fact that the array is now 2 dimensional. (It looks in system memory using libkvm (!)). Old versions of netstat see only the first FIB. In addition two sysctls are added to give: a) the number of FIBs compiled in (active) b) the default FIB of the calling process. Early testing experience: ------------------------- Basically our (IronPort's) appliance does this functionality already using ipfw fwd but that method has some drawbacks. For example, It can't fully simulate a routing table because it can't influence the socket's choice of local address when a connect() is done. Testing during the generating of these changes has been remarkably smooth so far. Multiple tables have co-existed with no notable side effects, and packets have been routes accordingly. ipfw has grown 2 new keywords: setfib N ip from anay to any count ip from any to any fib N In pf there seems to be a requirement to be able to give symbolic names to the fibs but I do not have that capacity. I am not sure if it is required. SCTP has interestingly enough built in support for this, called VRFs in Cisco parlance. it will be interesting to see how that handles it when it suddenly actually does something. Where to next: -------------------- After committing the ABI compatible version and MFCing it, I'd like to proceed in a forward direction in -current. this will result in some roto-tilling in the routing code. Firstly: the current code's idea of having a separate tree per protocol family, all of the same format, and pointed to by the 1 dimensional array is a bit silly. Especially when one considers that there is code that makes assumptions about every protocol having the same internal structures there. Some protocols don't WANT that sort of structure. (for example the whole idea of a netmask is foreign to appletalk). This needs to be made opaque to the external code. My suggested first change is to add routing method pointers to the 'domain' structure, along with information pointing the data. instead of having an array of pointers to uniform structures, there would be an array pointing to the 'domain' structures for each protocol address domain (protocol family), and the methods this reached would be called. The methods would have an argument that gives FIB number, but the protocol would be free to ignore it. When the ABI can be changed it raises the possibilty of the addition of a fib entry into the "struct route". Currently, the structure contains the sockaddr of the desination, and the resulting fib entry. To make this work fully, one could add a fib number so that given an address and a fib, one can find the third element, the fib entry. Interaction with the ARP layer/ LL layer would need to be revisited as well. Qing Li has been working on this already. This work was sponsored by Ironport Systems/Cisco Reviewed by: several including rwatson, bz and mlair (parts each) Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
#ifdef __FreeBSD__
{
M_SETFIB(m0, r->rtableid);
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_mtag->rtableid = r->rtableid;
Add code to allow the system to handle multiple routing tables. This particular implementation is designed to be fully backwards compatible and to be MFC-able to 7.x (and 6.x) Currently the only protocol that can make use of the multiple tables is IPv4 Similar functionality exists in OpenBSD and Linux. From my notes: ----- One thing where FreeBSD has been falling behind, and which by chance I have some time to work on is "policy based routing", which allows different packet streams to be routed by more than just the destination address. Constraints: ------------ I want to make some form of this available in the 6.x tree (and by extension 7.x) , but FreeBSD in general needs it so I might as well do it in -current and back port the portions I need. One of the ways that this can be done is to have the ability to instantiate multiple kernel routing tables (which I will now refer to as "Forwarding Information Bases" or "FIBs" for political correctness reasons). Which FIB a particular packet uses to make the next hop decision can be decided by a number of mechanisms. The policies these mechanisms implement are the "Policies" referred to in "Policy based routing". One of the constraints I have if I try to back port this work to 6.x is that it must be implemented as a EXTENSION to the existing ABIs in 6.x so that third party applications do not need to be recompiled in timespan of the branch. This first version will not have some of the bells and whistles that will come with later versions. It will, for example, be limited to 16 tables in the first commit. Implementation method, Compatible version. (part 1) ------------------------------- For this reason I have implemented a "sufficient subset" of a multiple routing table solution in Perforce, and back-ported it to 6.x. (also in Perforce though not always caught up with what I have done in -current/P4). The subset allows a number of FIBs to be defined at compile time (8 is sufficient for my purposes in 6.x) and implements the changes needed to allow IPV4 to use them. I have not done the changes for ipv6 simply because I do not need it, and I do not have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it. Other protocol families are left untouched and should there be users with proprietary protocol families, they should continue to work and be oblivious to the existence of the extra FIBs. To understand how this is done, one must know that the current FIB code starts everything off with a single dimensional array of pointers to FIB head structures (One per protocol family), each of which in turn points to the trie of routes available to that family. The basic change in the ABI compatible version of the change is to extent that array to be a 2 dimensional array, so that instead of protocol family X looking at rt_tables[X] for the table it needs, it looks at rt_tables[Y][X] when for all protocol families except ipv4 Y is always 0. Code that is unaware of the change always just sees the first row of the table, which of course looks just like the one dimensional array that existed before. The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign() are all maintained, but refer only to the first row of the array, so that existing callers in proprietary protocols can continue to do the "right thing". Some new entry points are added, for the exclusive use of ipv4 code called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(), which have an extra argument which refers the code to the correct row. In addition, there are some new entry points (currently called rtalloc_fib() and friends) that check the Address family being looked up and call either rtalloc() (and friends) if the protocol is not IPv4 forcing the action to row 0 or to the appropriate row if it IS IPv4 (and that info is available). These are for calling from code that is not specific to any particular protocol. The way these are implemented would change in the non ABI preserving code to be added later. One feature of the first version of the code is that for ipv4, the interface routes show up automatically on all the FIBs, so that no matter what FIB you select you always have the basic direct attached hosts available to you. (rtinit() does this automatically). You CAN delete an interface route from one FIB should you want to but by default it's there. ARP information is also available in each FIB. It's assumed that the same machine would have the same MAC address, regardless of which FIB you are using to get to it. This brings us as to how the correct FIB is selected for an outgoing IPV4 packet. Firstly, all packets have a FIB associated with them. if nothing has been done to change it, it will be FIB 0. The FIB is changed in the following ways. Packets fall into one of a number of classes. 1/ locally generated packets, coming from a socket/PCB. Such packets select a FIB from a number associated with the socket/PCB. This in turn is inherited from the process, but can be changed by a socket option. The process in turn inherits it on fork. I have written a utility call setfib that acts a bit like nice.. setfib -3 ping target.example.com # will use fib 3 for ping. It is an obvious extension to make it a property of a jail but I have not done so. It can be achieved by combining the setfib and jail commands. 2/ packets received on an interface for forwarding. By default these packets would use table 0, (or possibly a number settable in a sysctl(not yet)). but prior to routing the firewall can inspect them (see below). (possibly in the future you may be able to associate a FIB with packets received on an interface.. An ifconfig arg, but not yet.) 3/ packets inspected by a packet classifier, which can arbitrarily associate a fib with it on a packet by packet basis. A fib assigned to a packet by a packet classifier (such as ipfw) would over-ride a fib associated by a more default source. (such as cases 1 or 2). 4/ a tcp listen socket associated with a fib will generate accept sockets that are associated with that same fib. 5/ Packets generated in response to some other packet (e.g. reset or icmp packets). These should use the FIB associated with the packet being reponded to. 6/ Packets generated during encapsulation. gif, tun and other tunnel interfaces will encapsulate using the FIB that was in effect withthe proces that set up the tunnel. thus setfib 1 ifconfig gif0 [tunnel instructions] will set the fib for the tunnel to use to be fib 1. Routing messages would be associated with their process, and thus select one FIB or another. messages from the kernel would be associated with the fib they refer to and would only be received by a routing socket associated with that fib. (not yet implemented) In addition Netstat has been edited to be able to cope with the fact that the array is now 2 dimensional. (It looks in system memory using libkvm (!)). Old versions of netstat see only the first FIB. In addition two sysctls are added to give: a) the number of FIBs compiled in (active) b) the default FIB of the calling process. Early testing experience: ------------------------- Basically our (IronPort's) appliance does this functionality already using ipfw fwd but that method has some drawbacks. For example, It can't fully simulate a routing table because it can't influence the socket's choice of local address when a connect() is done. Testing during the generating of these changes has been remarkably smooth so far. Multiple tables have co-existed with no notable side effects, and packets have been routes accordingly. ipfw has grown 2 new keywords: setfib N ip from anay to any count ip from any to any fib N In pf there seems to be a requirement to be able to give symbolic names to the fibs but I do not have that capacity. I am not sure if it is required. SCTP has interestingly enough built in support for this, called VRFs in Cisco parlance. it will be interesting to see how that handles it when it suddenly actually does something. Where to next: -------------------- After committing the ABI compatible version and MFCing it, I'd like to proceed in a forward direction in -current. this will result in some roto-tilling in the routing code. Firstly: the current code's idea of having a separate tree per protocol family, all of the same format, and pointed to by the 1 dimensional array is a bit silly. Especially when one considers that there is code that makes assumptions about every protocol having the same internal structures there. Some protocols don't WANT that sort of structure. (for example the whole idea of a netmask is foreign to appletalk). This needs to be made opaque to the external code. My suggested first change is to add routing method pointers to the 'domain' structure, along with information pointing the data. instead of having an array of pointers to uniform structures, there would be an array pointing to the 'domain' structures for each protocol address domain (protocol family), and the methods this reached would be called. The methods would have an argument that gives FIB number, but the protocol would be free to ignore it. When the ABI can be changed it raises the possibilty of the addition of a fib entry into the "struct route". Currently, the structure contains the sockaddr of the desination, and the resulting fib entry. To make this work fully, one could add a fib number so that given an address and a fib, one can find the third element, the fib entry. Interaction with the ARP layer/ LL layer would need to be revisited as well. Qing Li has been working on this already. This work was sponsored by Ironport Systems/Cisco Reviewed by: several including rwatson, bz and mlair (parts each) Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
#ifdef __FreeBSD__
}
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef ALTQ
if (r->qid) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_mtag->qid = r->qid;
/* add hints for ecn */
pf_mtag->af = af;
pf_mtag->hdr = mtod(m0, struct ip *);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* ALTQ */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
switch (af) {
#ifdef INET
case AF_INET:
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
/* icmp_error() expects host byte ordering */
ip = mtod(m0, struct ip *);
NTOHS(ip->ip_len);
NTOHS(ip->ip_off);
PF_UNLOCK();
Pass icmp_error() the MTU argument directly instead of an interface pointer. This simplifies a couple of uses and removes some XXX workarounds.
2005-05-04 13:09:19 +00:00
icmp_error(m0, type, code, 0, 0);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_LOCK();
Combine rev. 1.29 and 1.30 to something that will make sense for future imports.
2005-05-04 15:55:29 +00:00
#else
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
icmp_error(m0, type, code, 0, 0);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_UNLOCK();
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
icmp6_error(m0, type, code, 0);
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_LOCK();
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET6 */
}
}
/*
* Return 1 if the addresses a and b match (with mask m), otherwise return 0.
* If n is 0, they match if they are equal. If n is != 0, they match if they
* are different.
*/
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;
switch (af) {
#ifdef INET
case AF_INET:
if ((a->addr32[0] & m->addr32[0]) ==
(b->addr32[0] & m->addr32[0]))
match++;
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
if (((a->addr32[0] & m->addr32[0]) ==
(b->addr32[0] & m->addr32[0])) &&
((a->addr32[1] & m->addr32[1]) ==
(b->addr32[1] & m->addr32[1])) &&
((a->addr32[2] & m->addr32[2]) ==
(b->addr32[2] & m->addr32[2])) &&
((a->addr32[3] & m->addr32[3]) ==
(b->addr32[3] & m->addr32[3])))
match++;
break;
#endif /* INET6 */
}
if (match) {
if (n)
return (0);
else
return (1);
} else {
if (n)
return (1);
else
return (0);
}
}
int
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));
case PF_OP_XRG:
return ((p < a1) || (p > a2));
case PF_OP_RRG:
return ((p >= a1) && (p <= a2));
case PF_OP_EQ:
return (p == a1);
case PF_OP_NE:
return (p != a1);
case PF_OP_LT:
return (p < a1);
case PF_OP_LE:
return (p <= a1);
case PF_OP_GT:
return (p > a1);
case PF_OP_GE:
return (p >= a1);
}
return (0); /* never reached */
}
int
pf_match_port(u_int8_t op, u_int16_t a1, u_int16_t a2, u_int16_t p)
{
NTOHS(a1);
NTOHS(a2);
NTOHS(p);
return (pf_match(op, a1, a2, p));
}
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));
}
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));
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifndef __FreeBSD__
struct pf_mtag *
pf_find_mtag(struct mbuf *m)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
struct m_tag *mtag;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if ((mtag = m_tag_find(m, PACKET_TAG_PF, NULL)) == NULL)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (NULL);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
return ((struct pf_mtag *)(mtag + 1));
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_mtag *
pf_get_mtag(struct mbuf *m)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct m_tag *mtag;
if ((mtag = m_tag_find(m, PACKET_TAG_PF, NULL)) == NULL) {
mtag = m_tag_get(PACKET_TAG_PF, sizeof(struct pf_mtag),
M_NOWAIT);
if (mtag == NULL)
return (NULL);
bzero(mtag + 1, sizeof(struct pf_mtag));
m_tag_prepend(m, mtag);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
return ((struct pf_mtag *)(mtag + 1));
}
#endif
int
pf_match_tag(struct mbuf *m, struct pf_rule *r, struct pf_mtag *pf_mtag,
int *tag)
{
if (*tag == -1)
*tag = pf_mtag->tag;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return ((!r->match_tag_not && r->match_tag == *tag) ||
(r->match_tag_not && r->match_tag != *tag));
}
int
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_tag_packet(struct mbuf *m, struct pf_mtag *pf_mtag, int tag, int rtableid)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (tag <= 0 && rtableid < 0)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (0);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pf_mtag == NULL)
if ((pf_mtag = pf_get_mtag(m)) == NULL)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (tag > 0)
pf_mtag->tag = tag;
if (rtableid >= 0)
Add code to allow the system to handle multiple routing tables. This particular implementation is designed to be fully backwards compatible and to be MFC-able to 7.x (and 6.x) Currently the only protocol that can make use of the multiple tables is IPv4 Similar functionality exists in OpenBSD and Linux. From my notes: ----- One thing where FreeBSD has been falling behind, and which by chance I have some time to work on is "policy based routing", which allows different packet streams to be routed by more than just the destination address. Constraints: ------------ I want to make some form of this available in the 6.x tree (and by extension 7.x) , but FreeBSD in general needs it so I might as well do it in -current and back port the portions I need. One of the ways that this can be done is to have the ability to instantiate multiple kernel routing tables (which I will now refer to as "Forwarding Information Bases" or "FIBs" for political correctness reasons). Which FIB a particular packet uses to make the next hop decision can be decided by a number of mechanisms. The policies these mechanisms implement are the "Policies" referred to in "Policy based routing". One of the constraints I have if I try to back port this work to 6.x is that it must be implemented as a EXTENSION to the existing ABIs in 6.x so that third party applications do not need to be recompiled in timespan of the branch. This first version will not have some of the bells and whistles that will come with later versions. It will, for example, be limited to 16 tables in the first commit. Implementation method, Compatible version. (part 1) ------------------------------- For this reason I have implemented a "sufficient subset" of a multiple routing table solution in Perforce, and back-ported it to 6.x. (also in Perforce though not always caught up with what I have done in -current/P4). The subset allows a number of FIBs to be defined at compile time (8 is sufficient for my purposes in 6.x) and implements the changes needed to allow IPV4 to use them. I have not done the changes for ipv6 simply because I do not need it, and I do not have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it. Other protocol families are left untouched and should there be users with proprietary protocol families, they should continue to work and be oblivious to the existence of the extra FIBs. To understand how this is done, one must know that the current FIB code starts everything off with a single dimensional array of pointers to FIB head structures (One per protocol family), each of which in turn points to the trie of routes available to that family. The basic change in the ABI compatible version of the change is to extent that array to be a 2 dimensional array, so that instead of protocol family X looking at rt_tables[X] for the table it needs, it looks at rt_tables[Y][X] when for all protocol families except ipv4 Y is always 0. Code that is unaware of the change always just sees the first row of the table, which of course looks just like the one dimensional array that existed before. The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign() are all maintained, but refer only to the first row of the array, so that existing callers in proprietary protocols can continue to do the "right thing". Some new entry points are added, for the exclusive use of ipv4 code called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(), which have an extra argument which refers the code to the correct row. In addition, there are some new entry points (currently called rtalloc_fib() and friends) that check the Address family being looked up and call either rtalloc() (and friends) if the protocol is not IPv4 forcing the action to row 0 or to the appropriate row if it IS IPv4 (and that info is available). These are for calling from code that is not specific to any particular protocol. The way these are implemented would change in the non ABI preserving code to be added later. One feature of the first version of the code is that for ipv4, the interface routes show up automatically on all the FIBs, so that no matter what FIB you select you always have the basic direct attached hosts available to you. (rtinit() does this automatically). You CAN delete an interface route from one FIB should you want to but by default it's there. ARP information is also available in each FIB. It's assumed that the same machine would have the same MAC address, regardless of which FIB you are using to get to it. This brings us as to how the correct FIB is selected for an outgoing IPV4 packet. Firstly, all packets have a FIB associated with them. if nothing has been done to change it, it will be FIB 0. The FIB is changed in the following ways. Packets fall into one of a number of classes. 1/ locally generated packets, coming from a socket/PCB. Such packets select a FIB from a number associated with the socket/PCB. This in turn is inherited from the process, but can be changed by a socket option. The process in turn inherits it on fork. I have written a utility call setfib that acts a bit like nice.. setfib -3 ping target.example.com # will use fib 3 for ping. It is an obvious extension to make it a property of a jail but I have not done so. It can be achieved by combining the setfib and jail commands. 2/ packets received on an interface for forwarding. By default these packets would use table 0, (or possibly a number settable in a sysctl(not yet)). but prior to routing the firewall can inspect them (see below). (possibly in the future you may be able to associate a FIB with packets received on an interface.. An ifconfig arg, but not yet.) 3/ packets inspected by a packet classifier, which can arbitrarily associate a fib with it on a packet by packet basis. A fib assigned to a packet by a packet classifier (such as ipfw) would over-ride a fib associated by a more default source. (such as cases 1 or 2). 4/ a tcp listen socket associated with a fib will generate accept sockets that are associated with that same fib. 5/ Packets generated in response to some other packet (e.g. reset or icmp packets). These should use the FIB associated with the packet being reponded to. 6/ Packets generated during encapsulation. gif, tun and other tunnel interfaces will encapsulate using the FIB that was in effect withthe proces that set up the tunnel. thus setfib 1 ifconfig gif0 [tunnel instructions] will set the fib for the tunnel to use to be fib 1. Routing messages would be associated with their process, and thus select one FIB or another. messages from the kernel would be associated with the fib they refer to and would only be received by a routing socket associated with that fib. (not yet implemented) In addition Netstat has been edited to be able to cope with the fact that the array is now 2 dimensional. (It looks in system memory using libkvm (!)). Old versions of netstat see only the first FIB. In addition two sysctls are added to give: a) the number of FIBs compiled in (active) b) the default FIB of the calling process. Early testing experience: ------------------------- Basically our (IronPort's) appliance does this functionality already using ipfw fwd but that method has some drawbacks. For example, It can't fully simulate a routing table because it can't influence the socket's choice of local address when a connect() is done. Testing during the generating of these changes has been remarkably smooth so far. Multiple tables have co-existed with no notable side effects, and packets have been routes accordingly. ipfw has grown 2 new keywords: setfib N ip from anay to any count ip from any to any fib N In pf there seems to be a requirement to be able to give symbolic names to the fibs but I do not have that capacity. I am not sure if it is required. SCTP has interestingly enough built in support for this, called VRFs in Cisco parlance. it will be interesting to see how that handles it when it suddenly actually does something. Where to next: -------------------- After committing the ABI compatible version and MFCing it, I'd like to proceed in a forward direction in -current. this will result in some roto-tilling in the routing code. Firstly: the current code's idea of having a separate tree per protocol family, all of the same format, and pointed to by the 1 dimensional array is a bit silly. Especially when one considers that there is code that makes assumptions about every protocol having the same internal structures there. Some protocols don't WANT that sort of structure. (for example the whole idea of a netmask is foreign to appletalk). This needs to be made opaque to the external code. My suggested first change is to add routing method pointers to the 'domain' structure, along with information pointing the data. instead of having an array of pointers to uniform structures, there would be an array pointing to the 'domain' structures for each protocol address domain (protocol family), and the methods this reached would be called. The methods would have an argument that gives FIB number, but the protocol would be free to ignore it. When the ABI can be changed it raises the possibilty of the addition of a fib entry into the "struct route". Currently, the structure contains the sockaddr of the desination, and the resulting fib entry. To make this work fully, one could add a fib number so that given an address and a fib, one can find the third element, the fib entry. Interaction with the ARP layer/ LL layer would need to be revisited as well. Qing Li has been working on this already. This work was sponsored by Ironport Systems/Cisco Reviewed by: several including rwatson, bz and mlair (parts each) Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
#ifdef __FreeBSD__
{
M_SETFIB(m, rtableid);
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_mtag->rtableid = rtableid;
Add code to allow the system to handle multiple routing tables. This particular implementation is designed to be fully backwards compatible and to be MFC-able to 7.x (and 6.x) Currently the only protocol that can make use of the multiple tables is IPv4 Similar functionality exists in OpenBSD and Linux. From my notes: ----- One thing where FreeBSD has been falling behind, and which by chance I have some time to work on is "policy based routing", which allows different packet streams to be routed by more than just the destination address. Constraints: ------------ I want to make some form of this available in the 6.x tree (and by extension 7.x) , but FreeBSD in general needs it so I might as well do it in -current and back port the portions I need. One of the ways that this can be done is to have the ability to instantiate multiple kernel routing tables (which I will now refer to as "Forwarding Information Bases" or "FIBs" for political correctness reasons). Which FIB a particular packet uses to make the next hop decision can be decided by a number of mechanisms. The policies these mechanisms implement are the "Policies" referred to in "Policy based routing". One of the constraints I have if I try to back port this work to 6.x is that it must be implemented as a EXTENSION to the existing ABIs in 6.x so that third party applications do not need to be recompiled in timespan of the branch. This first version will not have some of the bells and whistles that will come with later versions. It will, for example, be limited to 16 tables in the first commit. Implementation method, Compatible version. (part 1) ------------------------------- For this reason I have implemented a "sufficient subset" of a multiple routing table solution in Perforce, and back-ported it to 6.x. (also in Perforce though not always caught up with what I have done in -current/P4). The subset allows a number of FIBs to be defined at compile time (8 is sufficient for my purposes in 6.x) and implements the changes needed to allow IPV4 to use them. I have not done the changes for ipv6 simply because I do not need it, and I do not have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it. Other protocol families are left untouched and should there be users with proprietary protocol families, they should continue to work and be oblivious to the existence of the extra FIBs. To understand how this is done, one must know that the current FIB code starts everything off with a single dimensional array of pointers to FIB head structures (One per protocol family), each of which in turn points to the trie of routes available to that family. The basic change in the ABI compatible version of the change is to extent that array to be a 2 dimensional array, so that instead of protocol family X looking at rt_tables[X] for the table it needs, it looks at rt_tables[Y][X] when for all protocol families except ipv4 Y is always 0. Code that is unaware of the change always just sees the first row of the table, which of course looks just like the one dimensional array that existed before. The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign() are all maintained, but refer only to the first row of the array, so that existing callers in proprietary protocols can continue to do the "right thing". Some new entry points are added, for the exclusive use of ipv4 code called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(), which have an extra argument which refers the code to the correct row. In addition, there are some new entry points (currently called rtalloc_fib() and friends) that check the Address family being looked up and call either rtalloc() (and friends) if the protocol is not IPv4 forcing the action to row 0 or to the appropriate row if it IS IPv4 (and that info is available). These are for calling from code that is not specific to any particular protocol. The way these are implemented would change in the non ABI preserving code to be added later. One feature of the first version of the code is that for ipv4, the interface routes show up automatically on all the FIBs, so that no matter what FIB you select you always have the basic direct attached hosts available to you. (rtinit() does this automatically). You CAN delete an interface route from one FIB should you want to but by default it's there. ARP information is also available in each FIB. It's assumed that the same machine would have the same MAC address, regardless of which FIB you are using to get to it. This brings us as to how the correct FIB is selected for an outgoing IPV4 packet. Firstly, all packets have a FIB associated with them. if nothing has been done to change it, it will be FIB 0. The FIB is changed in the following ways. Packets fall into one of a number of classes. 1/ locally generated packets, coming from a socket/PCB. Such packets select a FIB from a number associated with the socket/PCB. This in turn is inherited from the process, but can be changed by a socket option. The process in turn inherits it on fork. I have written a utility call setfib that acts a bit like nice.. setfib -3 ping target.example.com # will use fib 3 for ping. It is an obvious extension to make it a property of a jail but I have not done so. It can be achieved by combining the setfib and jail commands. 2/ packets received on an interface for forwarding. By default these packets would use table 0, (or possibly a number settable in a sysctl(not yet)). but prior to routing the firewall can inspect them (see below). (possibly in the future you may be able to associate a FIB with packets received on an interface.. An ifconfig arg, but not yet.) 3/ packets inspected by a packet classifier, which can arbitrarily associate a fib with it on a packet by packet basis. A fib assigned to a packet by a packet classifier (such as ipfw) would over-ride a fib associated by a more default source. (such as cases 1 or 2). 4/ a tcp listen socket associated with a fib will generate accept sockets that are associated with that same fib. 5/ Packets generated in response to some other packet (e.g. reset or icmp packets). These should use the FIB associated with the packet being reponded to. 6/ Packets generated during encapsulation. gif, tun and other tunnel interfaces will encapsulate using the FIB that was in effect withthe proces that set up the tunnel. thus setfib 1 ifconfig gif0 [tunnel instructions] will set the fib for the tunnel to use to be fib 1. Routing messages would be associated with their process, and thus select one FIB or another. messages from the kernel would be associated with the fib they refer to and would only be received by a routing socket associated with that fib. (not yet implemented) In addition Netstat has been edited to be able to cope with the fact that the array is now 2 dimensional. (It looks in system memory using libkvm (!)). Old versions of netstat see only the first FIB. In addition two sysctls are added to give: a) the number of FIBs compiled in (active) b) the default FIB of the calling process. Early testing experience: ------------------------- Basically our (IronPort's) appliance does this functionality already using ipfw fwd but that method has some drawbacks. For example, It can't fully simulate a routing table because it can't influence the socket's choice of local address when a connect() is done. Testing during the generating of these changes has been remarkably smooth so far. Multiple tables have co-existed with no notable side effects, and packets have been routes accordingly. ipfw has grown 2 new keywords: setfib N ip from anay to any count ip from any to any fib N In pf there seems to be a requirement to be able to give symbolic names to the fibs but I do not have that capacity. I am not sure if it is required. SCTP has interestingly enough built in support for this, called VRFs in Cisco parlance. it will be interesting to see how that handles it when it suddenly actually does something. Where to next: -------------------- After committing the ABI compatible version and MFCing it, I'd like to proceed in a forward direction in -current. this will result in some roto-tilling in the routing code. Firstly: the current code's idea of having a separate tree per protocol family, all of the same format, and pointed to by the 1 dimensional array is a bit silly. Especially when one considers that there is code that makes assumptions about every protocol having the same internal structures there. Some protocols don't WANT that sort of structure. (for example the whole idea of a netmask is foreign to appletalk). This needs to be made opaque to the external code. My suggested first change is to add routing method pointers to the 'domain' structure, along with information pointing the data. instead of having an array of pointers to uniform structures, there would be an array pointing to the 'domain' structures for each protocol address domain (protocol family), and the methods this reached would be called. The methods would have an argument that gives FIB number, but the protocol would be free to ignore it. When the ABI can be changed it raises the possibilty of the addition of a fib entry into the "struct route". Currently, the structure contains the sockaddr of the desination, and the resulting fib entry. To make this work fully, one could add a fib number so that given an address and a fib, one can find the third element, the fib entry. Interaction with the ARP layer/ LL layer would need to be revisited as well. Qing Li has been working on this already. This work was sponsored by Ironport Systems/Cisco Reviewed by: several including rwatson, bz and mlair (parts each) Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
#ifdef __FreeBSD__
}
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (0);
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
static void
pf_step_into_anchor(int *depth, struct pf_ruleset **rs, int n,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_rule **r, struct pf_rule **a, int *match)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
{
struct pf_anchor_stackframe *f;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
(*r)->anchor->match = 0;
if (match)
*match = 0;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (*depth >= sizeof(pf_anchor_stack) /
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)
*a = *r;
f = pf_anchor_stack + (*depth)++;
f->rs = *rs;
f->r = *r;
if ((*r)->anchor_wildcard) {
f->parent = &(*r)->anchor->children;
if ((f->child = RB_MIN(pf_anchor_node, f->parent)) ==
NULL) {
*r = NULL;
return;
}
*rs = &f->child->ruleset;
} else {
f->parent = NULL;
f->child = NULL;
*rs = &(*r)->anchor->ruleset;
}
*r = TAILQ_FIRST((*rs)->rules[n].active.ptr);
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_step_out_of_anchor(int *depth, struct pf_ruleset **rs, int n,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_rule **r, struct pf_rule **a, int *match)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
{
struct pf_anchor_stackframe *f;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int quick = 0;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
do {
if (*depth <= 0)
break;
f = pf_anchor_stack + *depth - 1;
if (f->parent != NULL && f->child != NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (f->child->match ||
(match != NULL && *match)) {
f->r->anchor->match = 1;
*match = 0;
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
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)
continue;
else
break;
}
}
(*depth)--;
if (*depth == 0 && a != NULL)
*a = NULL;
*rs = f->rs;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (f->r->anchor->match || (match != NULL && *match))
quick = f->r->quick;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
*r = TAILQ_NEXT(f->r, entries);
} while (*r == NULL);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
return (quick);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef INET6
void
pf_poolmask(struct pf_addr *naddr, struct pf_addr *raddr,
struct pf_addr *rmask, struct pf_addr *saddr, sa_family_t af)
{
switch (af) {
#ifdef INET
case AF_INET:
naddr->addr32[0] = (raddr->addr32[0] & rmask->addr32[0]) |
((rmask->addr32[0] ^ 0xffffffff ) & saddr->addr32[0]);
break;
#endif /* INET */
case AF_INET6:
naddr->addr32[0] = (raddr->addr32[0] & rmask->addr32[0]) |
((rmask->addr32[0] ^ 0xffffffff ) & saddr->addr32[0]);
naddr->addr32[1] = (raddr->addr32[1] & rmask->addr32[1]) |
((rmask->addr32[1] ^ 0xffffffff ) & saddr->addr32[1]);
naddr->addr32[2] = (raddr->addr32[2] & rmask->addr32[2]) |
((rmask->addr32[2] ^ 0xffffffff ) & saddr->addr32[2]);
naddr->addr32[3] = (raddr->addr32[3] & rmask->addr32[3]) |
((rmask->addr32[3] ^ 0xffffffff ) & saddr->addr32[3]);
break;
}
}
void
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_addr_inc(struct pf_addr *addr, sa_family_t af)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
switch (af) {
#ifdef INET
case AF_INET:
addr->addr32[0] = htonl(ntohl(addr->addr32[0]) + 1);
break;
#endif /* INET */
case AF_INET6:
if (addr->addr32[3] == 0xffffffff) {
addr->addr32[3] = 0;
if (addr->addr32[2] == 0xffffffff) {
addr->addr32[2] = 0;
if (addr->addr32[1] == 0xffffffff) {
addr->addr32[1] = 0;
addr->addr32[0] =
htonl(ntohl(addr->addr32[0]) + 1);
} else
addr->addr32[1] =
htonl(ntohl(addr->addr32[1]) + 1);
} else
addr->addr32[2] =
htonl(ntohl(addr->addr32[2]) + 1);
} 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); \
} while (0)
/*
* hash function based on bridge_hash in if_bridge.c
*/
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];
switch (af) {
#ifdef INET
case AF_INET:
a += inaddr->addr32[0];
b += key->key32[1];
mix(a, b, c);
hash->addr32[0] = c + key->key32[2];
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
a += inaddr->addr32[0];
b += inaddr->addr32[2];
mix(a, b, c);
hash->addr32[0] = c;
a += inaddr->addr32[1];
b += inaddr->addr32[3];
c += key->key32[1];
mix(a, b, c);
hash->addr32[1] = c;
a += inaddr->addr32[2];
b += inaddr->addr32[1];
c += key->key32[2];
mix(a, b, c);
hash->addr32[2] = c;
a += inaddr->addr32[3];
b += inaddr->addr32[0];
c += key->key32[3];
mix(a, b, c);
hash->addr32[3] = c;
break;
#endif /* INET6 */
}
}
int
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
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)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
unsigned char hash[16];
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
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;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_pooladdr *acur = rpool->cur;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
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)
k.rule.ptr = r;
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 (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);
printf("\n");
}
return (0);
}
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (rpool->cur->addr.type == PF_ADDR_NOROUTE)
return (1);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (rpool->cur->addr.type == PF_ADDR_DYNIFTL) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
switch (af) {
#ifdef INET
case AF_INET:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (rpool->cur->addr.p.dyn->pfid_acnt4 < 1 &&
(rpool->opts & PF_POOL_TYPEMASK) !=
PF_POOL_ROUNDROBIN)
return (1);
raddr = &rpool->cur->addr.p.dyn->pfid_addr4;
rmask = &rpool->cur->addr.p.dyn->pfid_mask4;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (rpool->cur->addr.p.dyn->pfid_acnt6 < 1 &&
(rpool->opts & PF_POOL_TYPEMASK) !=
PF_POOL_ROUNDROBIN)
return (1);
raddr = &rpool->cur->addr.p.dyn->pfid_addr6;
rmask = &rpool->cur->addr.p.dyn->pfid_mask6;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
break;
#endif /* INET6 */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
} else if (rpool->cur->addr.type == PF_ADDR_TABLE) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
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;
}
switch (rpool->opts & PF_POOL_TYPEMASK) {
case PF_POOL_NONE:
PF_ACPY(naddr, raddr, af);
break;
case PF_POOL_BITMASK:
PF_POOLMASK(naddr, raddr, rmask, saddr, af);
break;
case PF_POOL_RANDOM:
if (init_addr != NULL && PF_AZERO(init_addr, af)) {
switch (af) {
#ifdef INET
case AF_INET:
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
rpool->counter.addr32[0] = htonl(arc4random());
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
if (rmask->addr32[3] != 0xffffffff)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
rpool->counter.addr32[3] =
htonl(arc4random());
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
else
break;
if (rmask->addr32[2] != 0xffffffff)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
rpool->counter.addr32[2] =
htonl(arc4random());
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
else
break;
if (rmask->addr32[1] != 0xffffffff)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
rpool->counter.addr32[1] =
htonl(arc4random());
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
else
break;
if (rmask->addr32[0] != 0xffffffff)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
rpool->counter.addr32[0] =
htonl(arc4random());
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET6 */
}
PF_POOLMASK(naddr, raddr, rmask, &rpool->counter, af);
PF_ACPY(init_addr, naddr, af);
} else {
PF_AINC(&rpool->counter, af);
PF_POOLMASK(naddr, raddr, rmask, &rpool->counter, af);
}
break;
case PF_POOL_SRCHASH:
pf_hash(saddr, (struct pf_addr *)&hash, &rpool->key, af);
PF_POOLMASK(naddr, raddr, rmask, (struct pf_addr *)&hash, af);
break;
case PF_POOL_ROUNDROBIN:
if (rpool->cur->addr.type == PF_ADDR_TABLE) {
if (!pfr_pool_get(rpool->cur->addr.p.tbl,
&rpool->tblidx, &rpool->counter,
&raddr, &rmask, af))
goto get_addr;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
} else if (rpool->cur->addr.type == PF_ADDR_DYNIFTL) {
if (!pfr_pool_get(rpool->cur->addr.p.dyn->pfid_kt,
&rpool->tblidx, &rpool->counter,
&raddr, &rmask, af))
goto get_addr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
} else if (pf_match_addr(0, raddr, rmask, &rpool->counter, af))
goto get_addr;
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)) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
/* table contains no address of type 'af' */
if (rpool->cur != acur)
goto try_next;
return (1);
}
} else if (rpool->cur->addr.type == PF_ADDR_DYNIFTL) {
rpool->tblidx = -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' */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (rpool->cur != acur)
goto try_next;
return (1);
}
} else {
raddr = &rpool->cur->addr.v.a.addr;
rmask = &rpool->cur->addr.v.a.mask;
PF_ACPY(&rpool->counter, raddr, af);
}
get_addr:
PF_ACPY(naddr, &rpool->counter, af);
Initialise init_addr in pf_map_addr() in the PF_POOL_ROUNDROBIN, prevents a possible endless loop in pf_get_sport() with 'static-port' ICMP state entries use the ICMP ID as port for the unique state key. When checking for a usable key, construct the key in the same way. Otherwise, a colliding key might be missed or a state insertion might be refused even though it could be inserted. The second case triggers the endless loop, possibly allowing a NATed LAN client to lock up the kernel. PR: kern/74930 Reported and tested by: Hugo Silva, Srebrenko Sehic MFC after: 3 days
2004-12-19 19:43:04 +00:00
if (init_addr != NULL && PF_AZERO(init_addr, af))
PF_ACPY(init_addr, naddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
PF_AINC(&rpool->counter, af);
break;
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (*sn != NULL)
PF_ACPY(&(*sn)->raddr, naddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (pf_status.debug >= PF_DEBUG_MISC &&
(rpool->opts & PF_POOL_TYPEMASK) != PF_POOL_NONE) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
printf("pf_map_addr: selected address ");
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_print_host(naddr, 0, af);
printf("\n");
}
return (0);
}
int
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_get_sport(sa_family_t af, u_int8_t proto, struct pf_rule *r,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_addr *saddr, struct pf_addr *daddr, u_int16_t dport,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_addr *naddr, u_int16_t *nport, u_int16_t low, u_int16_t high,
struct pf_src_node **sn)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_state_cmp key;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_addr init_addr;
u_int16_t cut;
bzero(&init_addr, sizeof(init_addr));
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (pf_map_addr(af, r, saddr, naddr, &init_addr, sn))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
if (proto == IPPROTO_ICMP) {
low = 1;
high = 65535;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
do {
key.af = af;
key.proto = proto;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
PF_ACPY(&key.ext.addr, daddr, key.af);
PF_ACPY(&key.gwy.addr, naddr, key.af);
key.ext.port = dport;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/*
* port search; start random, step;
* similar 2 portloop in in_pcbbind
*/
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
if (!(proto == IPPROTO_TCP || proto == IPPROTO_UDP ||
proto == IPPROTO_ICMP)) {
Initialise init_addr in pf_map_addr() in the PF_POOL_ROUNDROBIN, prevents a possible endless loop in pf_get_sport() with 'static-port' ICMP state entries use the ICMP ID as port for the unique state key. When checking for a usable key, construct the key in the same way. Otherwise, a colliding key might be missed or a state insertion might be refused even though it could be inserted. The second case triggers the endless loop, possibly allowing a NATed LAN client to lock up the kernel. PR: kern/74930 Reported and tested by: Hugo Silva, Srebrenko Sehic MFC after: 3 days
2004-12-19 19:43:04 +00:00
key.gwy.port = dport;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (pf_find_state_all(&key, PF_EXT_GWY, NULL) == NULL)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (0);
} else if (low == 0 && high == 0) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
key.gwy.port = *nport;
if (pf_find_state_all(&key, PF_EXT_GWY, NULL) == NULL)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (0);
} else if (low == high) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
key.gwy.port = htons(low);
if (pf_find_state_all(&key, PF_EXT_GWY, NULL) == NULL) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
*nport = htons(low);
return (0);
}
} else {
u_int16_t tmp;
if (low > high) {
tmp = low;
low = high;
high = tmp;
}
/* low < high */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
cut = htonl(arc4random()) % (1 + high - low) + low;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* low <= cut <= high */
for (tmp = cut; tmp <= high; ++(tmp)) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
key.gwy.port = htons(tmp);
if (pf_find_state_all(&key, PF_EXT_GWY, NULL) ==
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
NULL) {
*nport = htons(tmp);
return (0);
}
}
for (tmp = cut - 1; tmp >= low; --(tmp)) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
key.gwy.port = htons(tmp);
if (pf_find_state_all(&key, PF_EXT_GWY, NULL) ==
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
NULL) {
*nport = htons(tmp);
return (0);
}
}
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
switch (r->rpool.opts & PF_POOL_TYPEMASK) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
case PF_POOL_RANDOM:
case PF_POOL_ROUNDROBIN:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (pf_map_addr(af, r, saddr, naddr, &init_addr, sn))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
break;
case PF_POOL_NONE:
case PF_POOL_SRCHASH:
case PF_POOL_BITMASK:
default:
return (1);
}
} while (! PF_AEQ(&init_addr, naddr, af) );
return (1); /* none available */
}
struct pf_rule *
pf_match_translation(struct pf_pdesc *pd, struct mbuf *m, int off,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
int direction, struct pfi_kif *kif, struct pf_addr *saddr, u_int16_t sport,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_addr *daddr, u_int16_t dport, int rs_num)
{
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct pf_rule *r, *rm = NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_ruleset *ruleset = NULL;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
int tag = -1;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int rtableid = -1;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
int asd = 0;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
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;
if (r->action == PF_BINAT && direction == PF_IN) {
src = &r->dst;
if (r->rpool.cur != NULL)
xdst = &r->rpool.cur->addr;
} else {
src = &r->src;
dst = &r->dst;
}
r->evaluations++;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pfi_kif_match(r->kif, kif) == r->ifnot)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_IFP].ptr;
else if (r->direction && r->direction != direction)
r = r->skip[PF_SKIP_DIR].ptr;
else if (r->af && r->af != pd->af)
r = r->skip[PF_SKIP_AF].ptr;
else if (r->proto && r->proto != pd->proto)
r = r->skip[PF_SKIP_PROTO].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (PF_MISMATCHAW(&src->addr, saddr, pd->af,
src->neg, kif))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[src == &r->src ? PF_SKIP_SRC_ADDR :
PF_SKIP_DST_ADDR].ptr;
else if (src->port_op && !pf_match_port(src->port_op,
src->port[0], src->port[1], sport))
r = r->skip[src == &r->src ? PF_SKIP_SRC_PORT :
PF_SKIP_DST_PORT].ptr;
else if (dst != NULL &&
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
PF_MISMATCHAW(&dst->addr, daddr, pd->af, dst->neg, NULL))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_DST_ADDR].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (xdst != NULL && PF_MISMATCHAW(xdst, daddr, pd->af,
0, NULL))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
else if (dst != NULL && dst->port_op &&
!pf_match_port(dst->port_op, dst->port[0],
dst->port[1], dport))
r = r->skip[PF_SKIP_DST_PORT].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->match_tag && !pf_match_tag(m, r, pd->pf_mtag, &tag))
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
r = TAILQ_NEXT(r, entries);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
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)))
r = TAILQ_NEXT(r, entries);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
else {
if (r->tag)
tag = r->tag;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r->rtableid >= 0)
rtableid = r->rtableid;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (r->anchor == NULL) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
rm = r;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
} else
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_step_into_anchor(&asd, &ruleset, rs_num,
&r, NULL, NULL);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
if (r == NULL)
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_step_out_of_anchor(&asd, &ruleset, rs_num, &r,
NULL, NULL);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pf_tag_packet(m, pd->pf_mtag, tag, rtableid))
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
return (NULL);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (rm != NULL && (rm->action == PF_NONAT ||
rm->action == PF_NORDR || rm->action == PF_NOBINAT))
return (NULL);
return (rm);
}
struct pf_rule *
pf_get_translation(struct pf_pdesc *pd, struct mbuf *m, int off, int direction,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pfi_kif *kif, struct pf_src_node **sn,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_addr *saddr, u_int16_t sport,
struct pf_addr *daddr, u_int16_t dport,
struct pf_addr *naddr, u_int16_t *nport)
{
struct pf_rule *r = NULL;
if (direction == PF_OUT) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
r = pf_match_translation(pd, m, off, direction, kif, saddr,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
sport, daddr, dport, PF_RULESET_BINAT);
if (r == NULL)
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
r = pf_match_translation(pd, m, off, direction, kif,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
saddr, sport, daddr, dport, PF_RULESET_NAT);
} else {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
r = pf_match_translation(pd, m, off, direction, kif, saddr,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
sport, daddr, dport, PF_RULESET_RDR);
if (r == NULL)
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
r = pf_match_translation(pd, m, off, direction, kif,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
saddr, sport, daddr, dport, PF_RULESET_BINAT);
}
if (r != NULL) {
switch (r->action) {
case PF_NONAT:
case PF_NOBINAT:
case PF_NORDR:
return (NULL);
case PF_NAT:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (pf_get_sport(pd->af, pd->proto, r, saddr,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
daddr, dport, naddr, nport, r->rpool.proxy_port[0],
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
r->rpool.proxy_port[1], sn)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
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);
}
break;
case PF_BINAT:
switch (direction) {
case PF_OUT:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (r->rpool.cur->addr.type == PF_ADDR_DYNIFTL){
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
switch (pd->af) {
#ifdef INET
case AF_INET:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (r->rpool.cur->addr.p.dyn->
pfid_acnt4 < 1)
return (NULL);
PF_POOLMASK(naddr,
&r->rpool.cur->addr.p.dyn->
pfid_addr4,
&r->rpool.cur->addr.p.dyn->
pfid_mask4,
saddr, AF_INET);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (r->rpool.cur->addr.p.dyn->
pfid_acnt6 < 1)
return (NULL);
PF_POOLMASK(naddr,
&r->rpool.cur->addr.p.dyn->
pfid_addr6,
&r->rpool.cur->addr.p.dyn->
pfid_mask6,
saddr, AF_INET6);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
break;
#endif /* INET6 */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
} else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
PF_POOLMASK(naddr,
&r->rpool.cur->addr.v.a.addr,
&r->rpool.cur->addr.v.a.mask,
saddr, pd->af);
break;
case PF_IN:
fix a bug that leads to a crash when binat rules of the form 'binat from ... to ... -> (if)' are used, where the interface is dynamic. Discovered by: kos(at)bastard(dot)net Analyzed by: Pyun YongHyeon Approved by: mlaier (mentor) MFC after: 1 week
2004-11-24 00:43:34 +00:00
if (r->src.addr.type == PF_ADDR_DYNIFTL) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
switch (pd->af) {
#ifdef INET
case AF_INET:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (r->src.addr.p.dyn->
pfid_acnt4 < 1)
return (NULL);
PF_POOLMASK(naddr,
&r->src.addr.p.dyn->
pfid_addr4,
&r->src.addr.p.dyn->
pfid_mask4,
daddr, AF_INET);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (r->src.addr.p.dyn->
pfid_acnt6 < 1)
return (NULL);
PF_POOLMASK(naddr,
&r->src.addr.p.dyn->
pfid_addr6,
&r->src.addr.p.dyn->
pfid_mask6,
daddr, AF_INET6);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
break;
#endif /* INET6 */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
} else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
PF_POOLMASK(naddr,
&r->src.addr.v.a.addr,
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
&r->src.addr.v.a.mask, daddr,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pd->af);
break;
}
break;
case PF_RDR: {
Use the packet's address family instead of the rule's when selecting a replacement address for an rdr rule. Some rdr rules have no address family (when the replacement is a table and no other criterion implies one AF). In this case, pf would fail to select a replacement address and drop the packet due to translation failure. Found by: Gustavo A. Baratto
2005-01-20 18:07:35 +00:00
if (pf_map_addr(pd->af, r, saddr, naddr, NULL, sn))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (NULL);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
if ((r->rpool.opts & PF_POOL_TYPEMASK) ==
PF_POOL_BITMASK)
PF_POOLMASK(naddr, naddr,
&r->rpool.cur->addr.v.a.mask, daddr,
pd->af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (r->rpool.proxy_port[1]) {
u_int32_t tmp_nport;
tmp_nport = ((ntohs(dport) -
ntohs(r->dst.port[0])) %
(r->rpool.proxy_port[1] -
r->rpool.proxy_port[0] + 1)) +
r->rpool.proxy_port[0];
/* wrap around if necessary */
if (tmp_nport > 65535)
tmp_nport -= 65535;
*nport = htons((u_int16_t)tmp_nport);
} else if (r->rpool.proxy_port[0])
*nport = htons(r->rpool.proxy_port[0]);
break;
}
default:
return (NULL);
}
}
return (r);
}
int
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_socket_lookup(int direction, struct pf_pdesc *pd, struct inpcb *inp_arg)
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_socket_lookup(int direction, struct pf_pdesc *pd)
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
struct pf_addr *saddr, *daddr;
u_int16_t sport, dport;
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
struct inpcbinfo *pi;
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct inpcbtable *tb;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct inpcb *inp;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pd == NULL)
return (-1);
pd->lookup.uid = UID_MAX;
pd->lookup.gid = GID_MAX;
pd->lookup.pid = NO_PID; /* XXX: revisit */
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
if (inp_arg != NULL) {
Teach pf and ipfw to use read locks in inpcbs write than write locks when reading credential data from sockets. Teach pf to unlock the pcbinfo more quickly once it has acquired an inpcb lock, as the inpcb lock is sufficient to protect the reference. Assert locks, rather than read locks or write locks, on inpcbs in subroutines--this is necessary as the inpcb may be passed down with a write lock from the protocol, or may be passed down with a read lock from the firewall lookup routine, and either is sufficient. MFC after: 3 months
2008-04-20 00:21:54 +00:00
INP_LOCK_ASSERT(inp_arg);
Cache so_cred as inp_cred in the inpcb. This means that inp_cred is always there, even after the socket has gone away. It also means that it is constant for the lifetime of the inp. Both facts lead to simpler code and possibly less locking. Suggested by: rwatson Reviewed by: rwatson MFC after: 6 weeks X-MFC Note: use a inp_pspare for inp_cred
2008-10-04 15:06:34 +00:00
pd->lookup.uid = inp_arg->inp_cred->cr_uid;
pd->lookup.gid = inp_arg->inp_cred->cr_groups[0];
return (1);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
}
#endif
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
switch (pd->proto) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
case IPPROTO_TCP:
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pd->hdr.tcp == NULL)
return (-1);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
sport = pd->hdr.tcp->th_sport;
dport = pd->hdr.tcp->th_dport;
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Commit step 1 of the vimage project, (network stack) virtualization work done by Marko Zec (zec@). This is the first in a series of commits over the course of the next few weeks. Mark all uses of global variables to be virtualized with a V_ prefix. Use macros to map them back to their global names for now, so this is a NOP change only. We hope to have caught at least 85-90% of what is needed so we do not invalidate a lot of outstanding patches again. Obtained from: //depot/projects/vimage-commit2/... Reviewed by: brooks, des, ed, mav, julian, jamie, kris, rwatson, zec, ... (various people I forgot, different versions) md5 (with a bit of help) Sponsored by: NLnet Foundation, The FreeBSD Foundation X-MFC after: never V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
pi = &V_tcbinfo;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
tb = &tcbtable;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
case IPPROTO_UDP:
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pd->hdr.udp == NULL)
return (-1);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
sport = pd->hdr.udp->uh_sport;
dport = pd->hdr.udp->uh_dport;
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Commit step 1 of the vimage project, (network stack) virtualization work done by Marko Zec (zec@). This is the first in a series of commits over the course of the next few weeks. Mark all uses of global variables to be virtualized with a V_ prefix. Use macros to map them back to their global names for now, so this is a NOP change only. We hope to have caught at least 85-90% of what is needed so we do not invalidate a lot of outstanding patches again. Obtained from: //depot/projects/vimage-commit2/... Reviewed by: brooks, des, ed, mav, julian, jamie, kris, rwatson, zec, ... (various people I forgot, different versions) md5 (with a bit of help) Sponsored by: NLnet Foundation, The FreeBSD Foundation X-MFC after: never V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
pi = &V_udbinfo;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
tb = &udbtable;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
default:
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
return (-1);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
if (direction == PF_IN) {
saddr = pd->src;
daddr = pd->dst;
} else {
u_int16_t p;
p = sport;
sport = dport;
dport = p;
saddr = pd->dst;
daddr = pd->src;
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
switch (pd->af) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#ifdef INET
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
case AF_INET:
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
INP_INFO_RLOCK(pi); /* XXX LOR */
inp = in_pcblookup_hash(pi, saddr->v4, sport, daddr->v4,
dport, 0, NULL);
if (inp == NULL) {
inp = in_pcblookup_hash(pi, saddr->v4, sport,
daddr->v4, dport, INPLOOKUP_WILDCARD, NULL);
if(inp == NULL) {
INP_INFO_RUNLOCK(pi);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
return (-1);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
}
}
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
inp = in_pcbhashlookup(tb, saddr->v4, sport, daddr->v4, dport);
if (inp == NULL) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
inp = in_pcblookup_listen(tb, daddr->v4, dport, 0);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (inp == NULL)
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
return (-1);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* INET */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef INET6
case AF_INET6:
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
INP_INFO_RLOCK(pi);
inp = in6_pcblookup_hash(pi, &saddr->v6, sport,
&daddr->v6, dport, 0, NULL);
if (inp == NULL) {
inp = in6_pcblookup_hash(pi, &saddr->v6, sport,
&daddr->v6, dport, INPLOOKUP_WILDCARD, NULL);
if (inp == NULL) {
INP_INFO_RUNLOCK(pi);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
return (-1);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
}
}
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
inp = in6_pcbhashlookup(tb, &saddr->v6, sport, &daddr->v6,
dport);
if (inp == NULL) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
inp = in6_pcblookup_listen(tb, &daddr->v6, dport, 0);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (inp == NULL)
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
return (-1);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET6 */
default:
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
return (-1);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Cache so_cred as inp_cred in the inpcb. This means that inp_cred is always there, even after the socket has gone away. It also means that it is constant for the lifetime of the inp. Both facts lead to simpler code and possibly less locking. Suggested by: rwatson Reviewed by: rwatson MFC after: 6 weeks X-MFC Note: use a inp_pspare for inp_cred
2008-10-04 15:06:34 +00:00
pd->lookup.uid = inp->inp_cred->cr_uid;
pd->lookup.gid = inp->inp_cred->cr_groups[0];
Teach pf and ipfw to use read locks in inpcbs write than write locks when reading credential data from sockets. Teach pf to unlock the pcbinfo more quickly once it has acquired an inpcb lock, as the inpcb lock is sufficient to protect the reference. Assert locks, rather than read locks or write locks, on inpcbs in subroutines--this is necessary as the inpcb may be passed down with a write lock from the protocol, or may be passed down with a read lock from the firewall lookup routine, and either is sufficient. MFC after: 3 months
2008-04-20 00:21:54 +00:00
INP_INFO_RUNLOCK(pi);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#else
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pd->lookup.uid = inp->inp_socket->so_euid;
pd->lookup.gid = inp->inp_socket->so_egid;
pd->lookup.pid = inp->inp_socket->so_cpid;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
}
u_int8_t
pf_get_wscale(struct mbuf *m, 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 <= 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);
while (hlen >= 3) {
switch (*opt) {
case TCPOPT_EOL:
case TCPOPT_NOP:
++opt;
--hlen;
break;
case TCPOPT_WINDOW:
wscale = opt[2];
if (wscale > TCP_MAX_WINSHIFT)
wscale = TCP_MAX_WINSHIFT;
wscale |= PF_WSCALE_FLAG;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
/* FALLTHROUGH */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
default:
optlen = opt[1];
if (optlen < 2)
optlen = 2;
hlen -= optlen;
opt += optlen;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
break;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
}
return (wscale);
}
u_int16_t
pf_get_mss(struct mbuf *m, int off, u_int16_t th_off, sa_family_t af)
{
int hlen;
u_int8_t hdr[60];
u_int8_t *opt, optlen;
Commit step 1 of the vimage project, (network stack) virtualization work done by Marko Zec (zec@). This is the first in a series of commits over the course of the next few weeks. Mark all uses of global variables to be virtualized with a V_ prefix. Use macros to map them back to their global names for now, so this is a NOP change only. We hope to have caught at least 85-90% of what is needed so we do not invalidate a lot of outstanding patches again. Obtained from: //depot/projects/vimage-commit2/... Reviewed by: brooks, des, ed, mav, julian, jamie, kris, rwatson, zec, ... (various people I forgot, different versions) md5 (with a bit of help) Sponsored by: NLnet Foundation, The FreeBSD Foundation X-MFC after: never V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
u_int16_t mss = V_tcp_mssdflt;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
hlen = th_off << 2; /* hlen <= sizeof(hdr) */
if (hlen <= 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);
while (hlen >= TCPOLEN_MAXSEG) {
switch (*opt) {
case TCPOPT_EOL:
case TCPOPT_NOP:
++opt;
--hlen;
break;
case TCPOPT_MAXSEG:
bcopy((caddr_t)(opt + 2), (caddr_t)&mss, 2);
Fix mss byte order, only affects synproxy code path. Submitted by: John L. Scarfone via OpenBSD Reviewed by: mlaier Obtained from: OpenBSD, rev. 1.483 MFC after: 2 days
2005-04-13 21:05:55 +00:00
NTOHS(mss);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
/* FALLTHROUGH */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
default:
optlen = opt[1];
if (optlen < 2)
optlen = 2;
hlen -= optlen;
opt += optlen;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
break;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
}
return (mss);
}
u_int16_t
pf_calc_mss(struct pf_addr *addr, sa_family_t af, u_int16_t offer)
{
#ifdef INET
struct sockaddr_in *dst;
struct route ro;
#endif /* INET */
#ifdef INET6
struct sockaddr_in6 *dst6;
struct route_in6 ro6;
#endif /* INET6 */
struct rtentry *rt = NULL;
Shut up a couple of annoying gcc warnings. Do not enclose the fixes with #ifdefs in order to loop it back to OpenBSD after the next import. There are a some implicit asserts involved which might be better spelled out explicitly (af == AF_INET ...) Approved by: bms(mentor)
2004-03-30 18:28:30 +00:00
int hlen = 0; /* make the compiler happy */
Commit step 1 of the vimage project, (network stack) virtualization work done by Marko Zec (zec@). This is the first in a series of commits over the course of the next few weeks. Mark all uses of global variables to be virtualized with a V_ prefix. Use macros to map them back to their global names for now, so this is a NOP change only. We hope to have caught at least 85-90% of what is needed so we do not invalidate a lot of outstanding patches again. Obtained from: //depot/projects/vimage-commit2/... Reviewed by: brooks, des, ed, mav, julian, jamie, kris, rwatson, zec, ... (various people I forgot, different versions) md5 (with a bit of help) Sponsored by: NLnet Foundation, The FreeBSD Foundation X-MFC after: never V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
u_int16_t mss = V_tcp_mssdflt;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
switch (af) {
#ifdef INET
case AF_INET:
hlen = sizeof(struct ip);
bzero(&ro, sizeof(ro));
dst = (struct sockaddr_in *)&ro.ro_dst;
dst->sin_family = AF_INET;
dst->sin_len = sizeof(*dst);
dst->sin_addr = addr->v4;
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#ifdef RTF_PRCLONING
rtalloc_ign(&ro, (RTF_CLONING | RTF_PRCLONING));
#else /* !RTF_PRCLONING */
This main goals of this project are: 1. separating L2 tables (ARP, NDP) from the L3 routing tables 2. removing as much locking dependencies among these layers as possible to allow for some parallelism in the search operations 3. simplify the logic in the routing code, The most notable end result is the obsolescent of the route cloning (RTF_CLONING) concept, which translated into code reduction in both IPv4 ARP and IPv6 NDP related modules, and size reduction in struct rtentry{}. The change in design obsoletes the semantics of RTF_CLONING, RTF_WASCLONE and RTF_LLINFO routing flags. The userland applications such as "arp" and "ndp" have been modified to reflect those changes. The output from "netstat -r" shows only the routing entries. Quite a few developers have contributed to this project in the past: Glebius Smirnoff, Luigi Rizzo, Alessandro Cerri, and Andre Oppermann. And most recently: - Kip Macy revised the locking code completely, thus completing the last piece of the puzzle, Kip has also been conducting active functional testing - Sam Leffler has helped me improving/refactoring the code, and provided valuable reviews - Julian Elischer setup the perforce tree for me and has helped me maintaining that branch before the svn conversion
2008-12-15 06:10:57 +00:00
in_rtalloc_ign(&ro, 0, 0);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
#else /* ! __FreeBSD__ */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
rtalloc_noclone(&ro, NO_CLONING);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
rt = ro.ro_rt;
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
hlen = sizeof(struct ip6_hdr);
bzero(&ro6, sizeof(ro6));
dst6 = (struct sockaddr_in6 *)&ro6.ro_dst;
dst6->sin6_family = AF_INET6;
dst6->sin6_len = sizeof(*dst6);
dst6->sin6_addr = addr->v6;
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#ifdef RTF_PRCLONING
rtalloc_ign((struct route *)&ro6,
(RTF_CLONING | RTF_PRCLONING));
#else /* !RTF_PRCLONING */
This main goals of this project are: 1. separating L2 tables (ARP, NDP) from the L3 routing tables 2. removing as much locking dependencies among these layers as possible to allow for some parallelism in the search operations 3. simplify the logic in the routing code, The most notable end result is the obsolescent of the route cloning (RTF_CLONING) concept, which translated into code reduction in both IPv4 ARP and IPv6 NDP related modules, and size reduction in struct rtentry{}. The change in design obsoletes the semantics of RTF_CLONING, RTF_WASCLONE and RTF_LLINFO routing flags. The userland applications such as "arp" and "ndp" have been modified to reflect those changes. The output from "netstat -r" shows only the routing entries. Quite a few developers have contributed to this project in the past: Glebius Smirnoff, Luigi Rizzo, Alessandro Cerri, and Andre Oppermann. And most recently: - Kip Macy revised the locking code completely, thus completing the last piece of the puzzle, Kip has also been conducting active functional testing - Sam Leffler has helped me improving/refactoring the code, and provided valuable reviews - Julian Elischer setup the perforce tree for me and has helped me maintaining that branch before the svn conversion
2008-12-15 06:10:57 +00:00
rtalloc_ign((struct route *)&ro6, 0);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
#else /* ! __FreeBSD__ */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
rtalloc_noclone((struct route *)&ro6, NO_CLONING);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
rt = ro6.ro_rt;
break;
#endif /* INET6 */
}
if (rt && rt->rt_ifp) {
mss = rt->rt_ifp->if_mtu - hlen - sizeof(struct tcphdr);
Commit step 1 of the vimage project, (network stack) virtualization work done by Marko Zec (zec@). This is the first in a series of commits over the course of the next few weeks. Mark all uses of global variables to be virtualized with a V_ prefix. Use macros to map them back to their global names for now, so this is a NOP change only. We hope to have caught at least 85-90% of what is needed so we do not invalidate a lot of outstanding patches again. Obtained from: //depot/projects/vimage-commit2/... Reviewed by: brooks, des, ed, mav, julian, jamie, kris, rwatson, zec, ... (various people I forgot, different versions) md5 (with a bit of help) Sponsored by: NLnet Foundation, The FreeBSD Foundation X-MFC after: never V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
mss = max(V_tcp_mssdflt, mss);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
RTFREE(rt);
}
mss = min(mss, offer);
mss = max(mss, 64); /* sanity - at least max opt space */
return (mss);
}
void
pf_set_rt_ifp(struct pf_state *s, struct pf_addr *saddr)
{
struct pf_rule *r = s->rule.ptr;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
s->rt_kif = NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (!r->rt || r->rt == PF_FASTROUTE)
return;
switch (s->af) {
#ifdef INET
case AF_INET:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_map_addr(AF_INET, r, saddr, &s->rt_addr, NULL,
&s->nat_src_node);
s->rt_kif = r->rpool.cur->kif;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_map_addr(AF_INET6, r, saddr, &s->rt_addr, NULL,
&s->nat_src_node);
s->rt_kif = r->rpool.cur->kif;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET6 */
}
}
int
pf_test_tcp(struct pf_rule **rm, struct pf_state **sm, int direction,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pfi_kif *kif, struct mbuf *m, int off, void *h,
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
struct pf_pdesc *pd, struct pf_rule **am, struct pf_ruleset **rsm,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct ifqueue *ifq, struct inpcb *inp)
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct pf_pdesc *pd, struct pf_rule **am, struct pf_ruleset **rsm,
struct ifqueue *ifq)
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_rule *nr = NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_addr *saddr = pd->src, *daddr = pd->dst;
struct tcphdr *th = pd->hdr.tcp;
u_int16_t bport, nport = 0;
sa_family_t af = pd->af;
struct pf_rule *r, *a = NULL;
struct pf_ruleset *ruleset = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_src_node *nsn = NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
u_short reason;
int rewrite = 0;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int tag = -1, rtableid = -1;
Commit step 1 of the vimage project, (network stack) virtualization work done by Marko Zec (zec@). This is the first in a series of commits over the course of the next few weeks. Mark all uses of global variables to be virtualized with a V_ prefix. Use macros to map them back to their global names for now, so this is a NOP change only. We hope to have caught at least 85-90% of what is needed so we do not invalidate a lot of outstanding patches again. Obtained from: //depot/projects/vimage-commit2/... Reviewed by: brooks, des, ed, mav, julian, jamie, kris, rwatson, zec, ... (various people I forgot, different versions) md5 (with a bit of help) Sponsored by: NLnet Foundation, The FreeBSD Foundation X-MFC after: never V_Commit_Message_Reviewed_By: more people than the patch
2008-08-17 23:27:27 +00:00
u_int16_t mss = V_tcp_mssdflt;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
int asd = 0;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int match = 0;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (pf_check_congestion(ifq)) {
REASON_SET(&reason, PFRES_CONGEST);
return (PF_DROP);
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifdef __FreeBSD__
if (inp != NULL)
pd->lookup.done = pf_socket_lookup(direction, pd, inp);
else if (debug_pfugidhack) {
PF_UNLOCK();
DPFPRINTF(PF_DEBUG_MISC, ("pf: unlocked lookup\n"));
pd->lookup.done = pf_socket_lookup(direction, pd, inp);
PF_LOCK();
}
Work around a long standing LOR with user/group rules by doing the socket lookup early. This has some performance implications and should not be enabled by default, but might help greatly in certain setups. After some more testing this could be turned into a sysctl. Tested by: avatar LOR ids: 17, 24, 32, 46, 191 (conceptual) MFC after: 6 weeks
2006-12-29 13:59:03 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_FIRST(pf_main_ruleset.rules[PF_RULESET_FILTER].active.ptr);
if (direction == PF_OUT) {
bport = nport = th->th_sport;
/* check outgoing packet for BINAT/NAT */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((nr = pf_get_translation(pd, m, off, PF_OUT, kif, &nsn,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
saddr, th->th_sport, daddr, th->th_dport,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
&pd->naddr, &nport)) != NULL) {
PF_ACPY(&pd->baddr, saddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_change_ap(saddr, &th->th_sport, pd->ip_sum,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
&th->th_sum, &pd->naddr, nport, 0, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
rewrite++;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr->natpass)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pd->nat_rule = nr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
} else {
bport = nport = th->th_dport;
/* check incoming packet for BINAT/RDR */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((nr = pf_get_translation(pd, m, off, PF_IN, kif, &nsn,
saddr, th->th_sport, daddr, th->th_dport,
&pd->naddr, &nport)) != NULL) {
PF_ACPY(&pd->baddr, daddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_change_ap(daddr, &th->th_dport, pd->ip_sum,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
&th->th_sum, &pd->naddr, nport, 0, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
rewrite++;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr->natpass)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pd->nat_rule = nr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
}
while (r != NULL) {
r->evaluations++;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pfi_kif_match(r->kif, kif) == r->ifnot)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_IFP].ptr;
else if (r->direction && r->direction != direction)
r = r->skip[PF_SKIP_DIR].ptr;
else if (r->af && r->af != af)
r = r->skip[PF_SKIP_AF].ptr;
else if (r->proto && r->proto != IPPROTO_TCP)
r = r->skip[PF_SKIP_PROTO].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (PF_MISMATCHAW(&r->src.addr, saddr, af,
r->src.neg, kif))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_SRC_ADDR].ptr;
else if (r->src.port_op && !pf_match_port(r->src.port_op,
r->src.port[0], r->src.port[1], th->th_sport))
r = r->skip[PF_SKIP_SRC_PORT].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (PF_MISMATCHAW(&r->dst.addr, daddr, af,
r->dst.neg, NULL))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_DST_ADDR].ptr;
else if (r->dst.port_op && !pf_match_port(r->dst.port_op,
r->dst.port[0], r->dst.port[1], th->th_dport))
r = r->skip[PF_SKIP_DST_PORT].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->tos && !(r->tos == pd->tos))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
else if (r->rule_flag & PFRULE_FRAGMENT)
r = TAILQ_NEXT(r, entries);
else if ((r->flagset & th->th_flags) != r->flags)
r = TAILQ_NEXT(r, entries);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->uid.op && (pd->lookup.done || (pd->lookup.done =
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_socket_lookup(direction, pd, inp), 1)) &&
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_socket_lookup(direction, pd), 1)) &&
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
!pf_match_uid(r->uid.op, r->uid.uid[0], r->uid.uid[1],
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pd->lookup.uid))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->gid.op && (pd->lookup.done || (pd->lookup.done =
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_socket_lookup(direction, pd, inp), 1)) &&
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_socket_lookup(direction, pd), 1)) &&
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
!pf_match_gid(r->gid.op, r->gid.gid[0], r->gid.gid[1],
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pd->lookup.gid))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
else if (r->prob && r->prob <= arc4random())
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->match_tag && !pf_match_tag(m, r, pd->pf_mtag, &tag))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
else if (r->os_fingerprint != PF_OSFP_ANY && !pf_osfp_match(
pf_osfp_fingerprint(pd, m, off, th), r->os_fingerprint))
r = TAILQ_NEXT(r, entries);
else {
if (r->tag)
tag = r->tag;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r->rtableid >= 0)
rtableid = r->rtableid;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (r->anchor == NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
match = 1;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
*rm = r;
*am = a;
*rsm = ruleset;
if ((*rm)->quick)
break;
r = TAILQ_NEXT(r, entries);
} else
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_step_into_anchor(&asd, &ruleset,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
PF_RULESET_FILTER, &r, &a, &match);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r == NULL && pf_step_out_of_anchor(&asd, &ruleset,
PF_RULESET_FILTER, &r, &a, &match))
break;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
r = *rm;
a = *am;
ruleset = *rsm;
REASON_SET(&reason, PFRES_MATCH);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r->log || (nr != NULL && nr->natpass && nr->log)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (rewrite)
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
m_copyback(m, off, sizeof(*th), (caddr_t)th);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#else
m_copyback(m, off, sizeof(*th), th);
#endif
PFLOG_PACKET(kif, h, m, af, direction, reason, r->log ? r : nr,
a, ruleset, pd);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
if ((r->action == PF_DROP) &&
((r->rule_flag & PFRULE_RETURNRST) ||
(r->rule_flag & PFRULE_RETURNICMP) ||
(r->rule_flag & PFRULE_RETURN))) {
/* undo NAT changes, if they have taken place */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr != NULL) {
if (direction == PF_OUT) {
pf_change_ap(saddr, &th->th_sport, pd->ip_sum,
&th->th_sum, &pd->baddr, bport, 0, af);
rewrite++;
} else {
pf_change_ap(daddr, &th->th_dport, pd->ip_sum,
&th->th_sum, &pd->baddr, bport, 0, af);
rewrite++;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
if (((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;
if (th->th_flags & TH_SYN)
ack++;
if (th->th_flags & TH_FIN)
ack++;
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#ifdef __FreeBSD__
pf_send_tcp(m, r, af, pd->dst,
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_send_tcp(r, af, pd->dst,
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pd->src, th->th_dport, th->th_sport,
ntohl(th->th_ack), ack, TH_RST|TH_ACK, 0, 0,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
r->return_ttl, 1, 0, pd->eh, kif->pfik_ifp);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
} else if ((af == AF_INET) && r->return_icmp)
pf_send_icmp(m, r->return_icmp >> 8,
r->return_icmp & 255, af, r);
else if ((af == AF_INET6) && r->return_icmp6)
pf_send_icmp(m, r->return_icmp6 >> 8,
r->return_icmp6 & 255, af, r);
}
if (r->action == PF_DROP)
return (PF_DROP);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pf_tag_packet(m, pd->pf_mtag, tag, rtableid)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
REASON_SET(&reason, PFRES_MEMORY);
return (PF_DROP);
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (r->keep_state || nr != NULL ||
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
(pd->flags & PFDESC_TCP_NORM)) {
/* create new state */
u_int16_t len;
struct pf_state *s = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_src_node *sn = NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
len = pd->tot_len - off - (th->th_off << 2);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
/* check maximums */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (r->max_states && (r->states >= r->max_states)) {
pf_status.lcounters[LCNT_STATES]++;
REASON_SET(&reason, PFRES_MAXSTATES);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto cleanup;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
/* src node for filter rule */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((r->rule_flag & PFRULE_SRCTRACK ||
r->rpool.opts & PF_POOL_STICKYADDR) &&
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_insert_src_node(&sn, r, saddr, af) != 0) {
REASON_SET(&reason, PFRES_SRCLIMIT);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto cleanup;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
/* src node for translation rule */
if (nr != NULL && (nr->rpool.opts & PF_POOL_STICKYADDR) &&
((direction == PF_OUT &&
pf_insert_src_node(&nsn, nr, &pd->baddr, af) != 0) ||
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
(pf_insert_src_node(&nsn, nr, saddr, af) != 0))) {
REASON_SET(&reason, PFRES_SRCLIMIT);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto cleanup;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
s = pool_get(&pf_state_pl, PR_NOWAIT);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (s == NULL) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(&reason, PFRES_MEMORY);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
cleanup:
if (sn != NULL && sn->states == 0 && sn->expire == 0) {
RB_REMOVE(pf_src_tree, &tree_src_tracking, sn);
pf_status.scounters[SCNT_SRC_NODE_REMOVALS]++;
pf_status.src_nodes--;
pool_put(&pf_src_tree_pl, sn);
}
if (nsn != sn && nsn != NULL && nsn->states == 0 &&
nsn->expire == 0) {
RB_REMOVE(pf_src_tree, &tree_src_tracking, nsn);
pf_status.scounters[SCNT_SRC_NODE_REMOVALS]++;
pf_status.src_nodes--;
pool_put(&pf_src_tree_pl, nsn);
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_DROP);
}
bzero(s, sizeof(*s));
s->rule.ptr = r;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
s->nat_rule.ptr = nr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->anchor.ptr = a;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
STATE_INC_COUNTERS(s);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->allow_opts = r->allow_opts;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
s->log = r->log & PF_LOG_ALL;
if (nr != NULL)
s->log |= nr->log & PF_LOG_ALL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->proto = IPPROTO_TCP;
s->direction = direction;
s->af = af;
if (direction == PF_OUT) {
PF_ACPY(&s->gwy.addr, saddr, af);
s->gwy.port = th->th_sport; /* sport */
PF_ACPY(&s->ext.addr, daddr, af);
s->ext.port = th->th_dport;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr != NULL) {
PF_ACPY(&s->lan.addr, &pd->baddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->lan.port = bport;
} else {
PF_ACPY(&s->lan.addr, &s->gwy.addr, af);
s->lan.port = s->gwy.port;
}
} else {
PF_ACPY(&s->lan.addr, daddr, af);
s->lan.port = th->th_dport;
PF_ACPY(&s->ext.addr, saddr, af);
s->ext.port = th->th_sport;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr != NULL) {
PF_ACPY(&s->gwy.addr, &pd->baddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->gwy.port = bport;
} else {
PF_ACPY(&s->gwy.addr, &s->lan.addr, af);
s->gwy.port = s->lan.port;
}
}
s->src.seqlo = ntohl(th->th_seq);
s->src.seqhi = s->src.seqlo + len + 1;
if ((th->th_flags & (TH_SYN|TH_ACK)) == TH_SYN &&
r->keep_state == PF_STATE_MODULATE) {
/* Generate sequence number modulator */
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifdef __FreeBSD__
while ((s->src.seqdiff =
pf_new_isn(s) - s->src.seqlo) == 0)
;
#else
while ((s->src.seqdiff =
tcp_rndiss_next() - s->src.seqlo) == 0)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_change_a(&th->th_seq, &th->th_sum,
htonl(s->src.seqlo + s->src.seqdiff), 0);
rewrite = 1;
} else
s->src.seqdiff = 0;
if (th->th_flags & TH_SYN) {
s->src.seqhi++;
s->src.wscale = pf_get_wscale(m, off, th->th_off, af);
}
s->src.max_win = MAX(ntohs(th->th_win), 1);
if (s->src.wscale & PF_WSCALE_MASK) {
/* Remove scale factor from initial window */
int win = s->src.max_win;
win += 1 << (s->src.wscale & PF_WSCALE_MASK);
s->src.max_win = (win - 1) >>
(s->src.wscale & PF_WSCALE_MASK);
}
if (th->th_flags & TH_FIN)
s->src.seqhi++;
s->dst.seqhi = 1;
s->dst.max_win = 1;
s->src.state = TCPS_SYN_SENT;
s->dst.state = TCPS_CLOSED;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
s->creation = time_second;
s->expire = time_second;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->timeout = PFTM_TCP_FIRST_PACKET;
pf_set_rt_ifp(s, saddr);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (sn != NULL) {
s->src_node = sn;
s->src_node->states++;
}
if (nsn != NULL) {
PF_ACPY(&nsn->raddr, &pd->naddr, af);
s->nat_src_node = nsn;
s->nat_src_node->states++;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if ((pd->flags & PFDESC_TCP_NORM) && pf_normalize_tcp_init(m,
off, pd, th, &s->src, &s->dst)) {
REASON_SET(&reason, PFRES_MEMORY);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_src_tree_remove_state(s);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
STATE_DEC_COUNTERS(s);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pool_put(&pf_state_pl, s);
return (PF_DROP);
}
if ((pd->flags & PFDESC_TCP_NORM) && s->src.scrub &&
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_normalize_tcp_stateful(m, off, pd, &reason, th, s,
&s->src, &s->dst, &rewrite)) {
/* This really shouldn't happen!!! */
DPFPRINTF(PF_DEBUG_URGENT,
("pf_normalize_tcp_stateful failed on first pkt"));
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_normalize_tcp_cleanup(s);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_src_tree_remove_state(s);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
STATE_DEC_COUNTERS(s);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pool_put(&pf_state_pl, s);
return (PF_DROP);
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (pf_insert_state(BOUND_IFACE(r, kif), s)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_normalize_tcp_cleanup(s);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(&reason, PFRES_STATEINS);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_src_tree_remove_state(s);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
STATE_DEC_COUNTERS(s);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pool_put(&pf_state_pl, s);
return (PF_DROP);
} else
*sm = s;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (tag > 0) {
pf_tag_ref(tag);
s->tag = tag;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if ((th->th_flags & (TH_SYN|TH_ACK)) == TH_SYN &&
r->keep_state == PF_STATE_SYNPROXY) {
s->src.state = PF_TCPS_PROXY_SRC;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr != NULL) {
if (direction == PF_OUT) {
pf_change_ap(saddr, &th->th_sport,
pd->ip_sum, &th->th_sum, &pd->baddr,
bport, 0, af);
} else {
pf_change_ap(daddr, &th->th_dport,
pd->ip_sum, &th->th_sum, &pd->baddr,
bport, 0, af);
}
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
s->src.seqhi = htonl(arc4random());
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* Find mss option */
mss = pf_get_mss(m, off, th->th_off, af);
mss = pf_calc_mss(saddr, af, mss);
mss = pf_calc_mss(daddr, af, mss);
s->src.mss = mss;
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#ifdef __FreeBSD__
pf_send_tcp(NULL, r, af, daddr, saddr, th->th_dport,
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_send_tcp(r, af, daddr, saddr, th->th_dport,
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#endif
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
th->th_sport, s->src.seqhi, ntohl(th->th_seq) + 1,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
TH_SYN|TH_ACK, 0, s->src.mss, 0, 1, 0, NULL, NULL);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(&reason, PFRES_SYNPROXY);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_SYNPROXY_DROP);
}
}
/* copy back packet headers if we performed NAT operations */
if (rewrite)
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
m_copyback(m, off, sizeof(*th), (caddr_t)th);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_PASS);
}
int
pf_test_udp(struct pf_rule **rm, struct pf_state **sm, int direction,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pfi_kif *kif, struct mbuf *m, int off, void *h,
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
struct pf_pdesc *pd, struct pf_rule **am, struct pf_ruleset **rsm,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct ifqueue *ifq, struct inpcb *inp)
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct pf_pdesc *pd, struct pf_rule **am, struct pf_ruleset **rsm,
struct ifqueue *ifq)
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_rule *nr = NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_addr *saddr = pd->src, *daddr = pd->dst;
struct udphdr *uh = pd->hdr.udp;
u_int16_t bport, nport = 0;
sa_family_t af = pd->af;
struct pf_rule *r, *a = NULL;
struct pf_ruleset *ruleset = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_src_node *nsn = NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
u_short reason;
int rewrite = 0;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int tag = -1, rtableid = -1;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
int asd = 0;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int match = 0;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (pf_check_congestion(ifq)) {
REASON_SET(&reason, PFRES_CONGEST);
return (PF_DROP);
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifdef __FreeBSD__
if (inp != NULL)
pd->lookup.done = pf_socket_lookup(direction, pd, inp);
else if (debug_pfugidhack) {
PF_UNLOCK();
DPFPRINTF(PF_DEBUG_MISC, ("pf: unlocked lookup\n"));
pd->lookup.done = pf_socket_lookup(direction, pd, inp);
PF_LOCK();
}
Work around a long standing LOR with user/group rules by doing the socket lookup early. This has some performance implications and should not be enabled by default, but might help greatly in certain setups. After some more testing this could be turned into a sysctl. Tested by: avatar LOR ids: 17, 24, 32, 46, 191 (conceptual) MFC after: 6 weeks
2006-12-29 13:59:03 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_FIRST(pf_main_ruleset.rules[PF_RULESET_FILTER].active.ptr);
if (direction == PF_OUT) {
bport = nport = uh->uh_sport;
/* check outgoing packet for BINAT/NAT */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((nr = pf_get_translation(pd, m, off, PF_OUT, kif, &nsn,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
saddr, uh->uh_sport, daddr, uh->uh_dport,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
&pd->naddr, &nport)) != NULL) {
PF_ACPY(&pd->baddr, saddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_change_ap(saddr, &uh->uh_sport, pd->ip_sum,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
&uh->uh_sum, &pd->naddr, nport, 1, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
rewrite++;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr->natpass)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pd->nat_rule = nr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
} else {
bport = nport = uh->uh_dport;
/* check incoming packet for BINAT/RDR */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((nr = pf_get_translation(pd, m, off, PF_IN, kif, &nsn,
saddr, uh->uh_sport, daddr, uh->uh_dport, &pd->naddr,
&nport)) != NULL) {
PF_ACPY(&pd->baddr, daddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_change_ap(daddr, &uh->uh_dport, pd->ip_sum,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
&uh->uh_sum, &pd->naddr, nport, 1, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
rewrite++;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr->natpass)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pd->nat_rule = nr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
}
while (r != NULL) {
r->evaluations++;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pfi_kif_match(r->kif, kif) == r->ifnot)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_IFP].ptr;
else if (r->direction && r->direction != direction)
r = r->skip[PF_SKIP_DIR].ptr;
else if (r->af && r->af != af)
r = r->skip[PF_SKIP_AF].ptr;
else if (r->proto && r->proto != IPPROTO_UDP)
r = r->skip[PF_SKIP_PROTO].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (PF_MISMATCHAW(&r->src.addr, saddr, af,
r->src.neg, kif))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_SRC_ADDR].ptr;
else if (r->src.port_op && !pf_match_port(r->src.port_op,
r->src.port[0], r->src.port[1], uh->uh_sport))
r = r->skip[PF_SKIP_SRC_PORT].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (PF_MISMATCHAW(&r->dst.addr, daddr, af,
r->dst.neg, NULL))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_DST_ADDR].ptr;
else if (r->dst.port_op && !pf_match_port(r->dst.port_op,
r->dst.port[0], r->dst.port[1], uh->uh_dport))
r = r->skip[PF_SKIP_DST_PORT].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->tos && !(r->tos == pd->tos))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
else if (r->rule_flag & PFRULE_FRAGMENT)
r = TAILQ_NEXT(r, entries);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->uid.op && (pd->lookup.done || (pd->lookup.done =
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_socket_lookup(direction, pd, inp), 1)) &&
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_socket_lookup(direction, pd), 1)) &&
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
!pf_match_uid(r->uid.op, r->uid.uid[0], r->uid.uid[1],
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pd->lookup.uid))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->gid.op && (pd->lookup.done || (pd->lookup.done =
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_socket_lookup(direction, pd, inp), 1)) &&
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_socket_lookup(direction, pd), 1)) &&
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
!pf_match_gid(r->gid.op, r->gid.gid[0], r->gid.gid[1],
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pd->lookup.gid))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
else if (r->prob && r->prob <= arc4random())
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->match_tag && !pf_match_tag(m, r, pd->pf_mtag, &tag))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
else if (r->os_fingerprint != PF_OSFP_ANY)
r = TAILQ_NEXT(r, entries);
else {
if (r->tag)
tag = r->tag;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r->rtableid >= 0)
rtableid = r->rtableid;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (r->anchor == NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
match = 1;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
*rm = r;
*am = a;
*rsm = ruleset;
if ((*rm)->quick)
break;
r = TAILQ_NEXT(r, entries);
} else
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_step_into_anchor(&asd, &ruleset,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
PF_RULESET_FILTER, &r, &a, &match);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r == NULL && pf_step_out_of_anchor(&asd, &ruleset,
PF_RULESET_FILTER, &r, &a, &match))
break;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
r = *rm;
a = *am;
ruleset = *rsm;
REASON_SET(&reason, PFRES_MATCH);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r->log || (nr != NULL && nr->natpass && nr->log)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (rewrite)
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
m_copyback(m, off, sizeof(*uh), (caddr_t)uh);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#else
m_copyback(m, off, sizeof(*uh), uh);
#endif
PFLOG_PACKET(kif, h, m, af, direction, reason, r->log ? r : nr,
a, ruleset, pd);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
if ((r->action == PF_DROP) &&
((r->rule_flag & PFRULE_RETURNICMP) ||
(r->rule_flag & PFRULE_RETURN))) {
/* undo NAT changes, if they have taken place */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr != NULL) {
if (direction == PF_OUT) {
pf_change_ap(saddr, &uh->uh_sport, pd->ip_sum,
&uh->uh_sum, &pd->baddr, bport, 1, af);
rewrite++;
} else {
pf_change_ap(daddr, &uh->uh_dport, pd->ip_sum,
&uh->uh_sum, &pd->baddr, bport, 1, af);
rewrite++;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
if ((af == AF_INET) && r->return_icmp)
pf_send_icmp(m, r->return_icmp >> 8,
r->return_icmp & 255, af, r);
else if ((af == AF_INET6) && r->return_icmp6)
pf_send_icmp(m, r->return_icmp6 >> 8,
r->return_icmp6 & 255, af, r);
}
if (r->action == PF_DROP)
return (PF_DROP);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pf_tag_packet(m, pd->pf_mtag, tag, rtableid)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
REASON_SET(&reason, PFRES_MEMORY);
return (PF_DROP);
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (r->keep_state || nr != NULL) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* create new state */
struct pf_state *s = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_src_node *sn = NULL;
/* check maximums */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (r->max_states && (r->states >= r->max_states)) {
pf_status.lcounters[LCNT_STATES]++;
REASON_SET(&reason, PFRES_MAXSTATES);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto cleanup;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
/* src node for filter rule */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((r->rule_flag & PFRULE_SRCTRACK ||
r->rpool.opts & PF_POOL_STICKYADDR) &&
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_insert_src_node(&sn, r, saddr, af) != 0) {
REASON_SET(&reason, PFRES_SRCLIMIT);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto cleanup;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
/* src node for translation rule */
if (nr != NULL && (nr->rpool.opts & PF_POOL_STICKYADDR) &&
((direction == PF_OUT &&
pf_insert_src_node(&nsn, nr, &pd->baddr, af) != 0) ||
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
(pf_insert_src_node(&nsn, nr, saddr, af) != 0))) {
REASON_SET(&reason, PFRES_SRCLIMIT);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto cleanup;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
s = pool_get(&pf_state_pl, PR_NOWAIT);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (s == NULL) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(&reason, PFRES_MEMORY);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
cleanup:
if (sn != NULL && sn->states == 0 && sn->expire == 0) {
RB_REMOVE(pf_src_tree, &tree_src_tracking, sn);
pf_status.scounters[SCNT_SRC_NODE_REMOVALS]++;
pf_status.src_nodes--;
pool_put(&pf_src_tree_pl, sn);
}
if (nsn != sn && nsn != NULL && nsn->states == 0 &&
nsn->expire == 0) {
RB_REMOVE(pf_src_tree, &tree_src_tracking, nsn);
pf_status.scounters[SCNT_SRC_NODE_REMOVALS]++;
pf_status.src_nodes--;
pool_put(&pf_src_tree_pl, nsn);
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_DROP);
}
bzero(s, sizeof(*s));
s->rule.ptr = r;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
s->nat_rule.ptr = nr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->anchor.ptr = a;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
STATE_INC_COUNTERS(s);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->allow_opts = r->allow_opts;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
s->log = r->log & PF_LOG_ALL;
if (nr != NULL)
s->log |= nr->log & PF_LOG_ALL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->proto = IPPROTO_UDP;
s->direction = direction;
s->af = af;
if (direction == PF_OUT) {
PF_ACPY(&s->gwy.addr, saddr, af);
s->gwy.port = uh->uh_sport;
PF_ACPY(&s->ext.addr, daddr, af);
s->ext.port = uh->uh_dport;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr != NULL) {
PF_ACPY(&s->lan.addr, &pd->baddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->lan.port = bport;
} else {
PF_ACPY(&s->lan.addr, &s->gwy.addr, af);
s->lan.port = s->gwy.port;
}
} else {
PF_ACPY(&s->lan.addr, daddr, af);
s->lan.port = uh->uh_dport;
PF_ACPY(&s->ext.addr, saddr, af);
s->ext.port = uh->uh_sport;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr != NULL) {
PF_ACPY(&s->gwy.addr, &pd->baddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->gwy.port = bport;
} else {
PF_ACPY(&s->gwy.addr, &s->lan.addr, af);
s->gwy.port = s->lan.port;
}
}
s->src.state = PFUDPS_SINGLE;
s->dst.state = PFUDPS_NO_TRAFFIC;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
s->creation = time_second;
s->expire = time_second;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->timeout = PFTM_UDP_FIRST_PACKET;
pf_set_rt_ifp(s, saddr);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (sn != NULL) {
s->src_node = sn;
s->src_node->states++;
}
if (nsn != NULL) {
PF_ACPY(&nsn->raddr, &pd->naddr, af);
s->nat_src_node = nsn;
s->nat_src_node->states++;
}
if (pf_insert_state(BOUND_IFACE(r, kif), s)) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(&reason, PFRES_STATEINS);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_src_tree_remove_state(s);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
STATE_DEC_COUNTERS(s);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pool_put(&pf_state_pl, s);
return (PF_DROP);
} else
*sm = s;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (tag > 0) {
pf_tag_ref(tag);
s->tag = tag;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
/* copy back packet headers if we performed NAT operations */
if (rewrite)
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
m_copyback(m, off, sizeof(*uh), (caddr_t)uh);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_PASS);
}
int
pf_test_icmp(struct pf_rule **rm, struct pf_state **sm, int direction,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pfi_kif *kif, struct mbuf *m, int off, void *h,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct pf_pdesc *pd, struct pf_rule **am, struct pf_ruleset **rsm,
struct ifqueue *ifq)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_rule *nr = NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_addr *saddr = pd->src, *daddr = pd->dst;
struct pf_rule *r, *a = NULL;
struct pf_ruleset *ruleset = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_src_node *nsn = NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
u_short reason;
Unbreak the build. Committed from the wrong directory.
2005-09-08 17:42:42 +00:00
u_int16_t icmpid = 0, bport, nport = 0;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
sa_family_t af = pd->af;
Shut up a couple of annoying gcc warnings. Do not enclose the fixes with #ifdefs in order to loop it back to OpenBSD after the next import. There are a some implicit asserts involved which might be better spelled out explicitly (af == AF_INET ...) Approved by: bms(mentor)
2004-03-30 18:28:30 +00:00
u_int8_t icmptype = 0; /* make the compiler happy */
u_int8_t icmpcode = 0; /* make the compiler happy */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
int state_icmp = 0;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int tag = -1, rtableid = -1;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef INET6
int rewrite = 0;
#endif /* INET6 */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
int asd = 0;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int match = 0;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (pf_check_congestion(ifq)) {
REASON_SET(&reason, PFRES_CONGEST);
return (PF_DROP);
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
switch (pd->proto) {
#ifdef INET
case IPPROTO_ICMP:
icmptype = pd->hdr.icmp->icmp_type;
icmpcode = pd->hdr.icmp->icmp_code;
icmpid = pd->hdr.icmp->icmp_id;
if (icmptype == ICMP_UNREACH ||
icmptype == ICMP_SOURCEQUENCH ||
icmptype == ICMP_REDIRECT ||
icmptype == ICMP_TIMXCEED ||
icmptype == ICMP_PARAMPROB)
state_icmp++;
break;
#endif /* INET */
#ifdef INET6
case IPPROTO_ICMPV6:
icmptype = pd->hdr.icmp6->icmp6_type;
icmpcode = pd->hdr.icmp6->icmp6_code;
icmpid = pd->hdr.icmp6->icmp6_id;
if (icmptype == ICMP6_DST_UNREACH ||
icmptype == ICMP6_PACKET_TOO_BIG ||
icmptype == ICMP6_TIME_EXCEEDED ||
icmptype == ICMP6_PARAM_PROB)
state_icmp++;
break;
#endif /* INET6 */
}
r = TAILQ_FIRST(pf_main_ruleset.rules[PF_RULESET_FILTER].active.ptr);
if (direction == PF_OUT) {
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
bport = nport = icmpid;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* check outgoing packet for BINAT/NAT */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((nr = pf_get_translation(pd, m, off, PF_OUT, kif, &nsn,
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
saddr, icmpid, daddr, icmpid, &pd->naddr, &nport)) !=
NULL) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
PF_ACPY(&pd->baddr, saddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
switch (af) {
#ifdef INET
case AF_INET:
pf_change_a(&saddr->v4.s_addr, pd->ip_sum,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pd->naddr.v4.s_addr, 0);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
pd->hdr.icmp->icmp_cksum = pf_cksum_fixup(
pd->hdr.icmp->icmp_cksum, icmpid, nport, 0);
pd->hdr.icmp->icmp_id = nport;
Unbreak the build. Committed from the wrong directory.
2005-09-08 17:42:42 +00:00
m_copyback(m, off, ICMP_MINLEN,
(caddr_t)pd->hdr.icmp);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
pf_change_a6(saddr, &pd->hdr.icmp6->icmp6_cksum,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
&pd->naddr, 0);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
rewrite++;
break;
#endif /* INET6 */
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr->natpass)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pd->nat_rule = nr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
} else {
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
bport = nport = icmpid;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* check incoming packet for BINAT/RDR */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((nr = pf_get_translation(pd, m, off, PF_IN, kif, &nsn,
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
saddr, icmpid, daddr, icmpid, &pd->naddr, &nport)) !=
NULL) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
PF_ACPY(&pd->baddr, daddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
switch (af) {
#ifdef INET
case AF_INET:
pf_change_a(&daddr->v4.s_addr,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pd->ip_sum, pd->naddr.v4.s_addr, 0);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
pf_change_a6(daddr, &pd->hdr.icmp6->icmp6_cksum,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
&pd->naddr, 0);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
rewrite++;
break;
#endif /* INET6 */
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr->natpass)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pd->nat_rule = nr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
}
while (r != NULL) {
r->evaluations++;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pfi_kif_match(r->kif, kif) == r->ifnot)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_IFP].ptr;
else if (r->direction && r->direction != direction)
r = r->skip[PF_SKIP_DIR].ptr;
else if (r->af && r->af != af)
r = r->skip[PF_SKIP_AF].ptr;
else if (r->proto && r->proto != pd->proto)
r = r->skip[PF_SKIP_PROTO].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (PF_MISMATCHAW(&r->src.addr, saddr, af,
r->src.neg, kif))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_SRC_ADDR].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (PF_MISMATCHAW(&r->dst.addr, daddr, af,
r->dst.neg, NULL))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_DST_ADDR].ptr;
else if (r->type && r->type != icmptype + 1)
r = TAILQ_NEXT(r, entries);
else if (r->code && r->code != icmpcode + 1)
r = TAILQ_NEXT(r, entries);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->tos && !(r->tos == pd->tos))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
else if (r->rule_flag & PFRULE_FRAGMENT)
r = TAILQ_NEXT(r, entries);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
else if (r->prob && r->prob <= arc4random())
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->match_tag && !pf_match_tag(m, r, pd->pf_mtag, &tag))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
else if (r->os_fingerprint != PF_OSFP_ANY)
r = TAILQ_NEXT(r, entries);
else {
if (r->tag)
tag = r->tag;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r->rtableid >= 0)
rtableid = r->rtableid;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (r->anchor == NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
match = 1;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
*rm = r;
*am = a;
*rsm = ruleset;
if ((*rm)->quick)
break;
r = TAILQ_NEXT(r, entries);
} else
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_step_into_anchor(&asd, &ruleset,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
PF_RULESET_FILTER, &r, &a, &match);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r == NULL && pf_step_out_of_anchor(&asd, &ruleset,
PF_RULESET_FILTER, &r, &a, &match))
break;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
r = *rm;
a = *am;
ruleset = *rsm;
REASON_SET(&reason, PFRES_MATCH);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r->log || (nr != NULL && nr->natpass && nr->log)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef INET6
if (rewrite)
m_copyback(m, off, sizeof(struct icmp6_hdr),
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)pd->hdr.icmp6);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#endif /* INET6 */
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
PFLOG_PACKET(kif, h, m, af, direction, reason, r->log ? r : nr,
a, ruleset, pd);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
if (r->action != PF_PASS)
return (PF_DROP);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pf_tag_packet(m, pd->pf_mtag, tag, rtableid)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
REASON_SET(&reason, PFRES_MEMORY);
return (PF_DROP);
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (!state_icmp && (r->keep_state || nr != NULL)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* create new state */
struct pf_state *s = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_src_node *sn = NULL;
/* check maximums */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (r->max_states && (r->states >= r->max_states)) {
pf_status.lcounters[LCNT_STATES]++;
REASON_SET(&reason, PFRES_MAXSTATES);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto cleanup;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
/* src node for filter rule */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((r->rule_flag & PFRULE_SRCTRACK ||
r->rpool.opts & PF_POOL_STICKYADDR) &&
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_insert_src_node(&sn, r, saddr, af) != 0) {
REASON_SET(&reason, PFRES_SRCLIMIT);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto cleanup;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
/* src node for translation rule */
if (nr != NULL && (nr->rpool.opts & PF_POOL_STICKYADDR) &&
((direction == PF_OUT &&
pf_insert_src_node(&nsn, nr, &pd->baddr, af) != 0) ||
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
(pf_insert_src_node(&nsn, nr, saddr, af) != 0))) {
REASON_SET(&reason, PFRES_SRCLIMIT);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto cleanup;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
s = pool_get(&pf_state_pl, PR_NOWAIT);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (s == NULL) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(&reason, PFRES_MEMORY);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
cleanup:
if (sn != NULL && sn->states == 0 && sn->expire == 0) {
RB_REMOVE(pf_src_tree, &tree_src_tracking, sn);
pf_status.scounters[SCNT_SRC_NODE_REMOVALS]++;
pf_status.src_nodes--;
pool_put(&pf_src_tree_pl, sn);
}
if (nsn != sn && nsn != NULL && nsn->states == 0 &&
nsn->expire == 0) {
RB_REMOVE(pf_src_tree, &tree_src_tracking, nsn);
pf_status.scounters[SCNT_SRC_NODE_REMOVALS]++;
pf_status.src_nodes--;
pool_put(&pf_src_tree_pl, nsn);
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_DROP);
}
bzero(s, sizeof(*s));
s->rule.ptr = r;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
s->nat_rule.ptr = nr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->anchor.ptr = a;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
STATE_INC_COUNTERS(s);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->allow_opts = r->allow_opts;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
s->log = r->log & PF_LOG_ALL;
if (nr != NULL)
s->log |= nr->log & PF_LOG_ALL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->proto = pd->proto;
s->direction = direction;
s->af = af;
if (direction == PF_OUT) {
PF_ACPY(&s->gwy.addr, saddr, af);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
s->gwy.port = nport;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
PF_ACPY(&s->ext.addr, daddr, af);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
s->ext.port = 0;
if (nr != NULL) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
PF_ACPY(&s->lan.addr, &pd->baddr, af);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
s->lan.port = bport;
} else {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
PF_ACPY(&s->lan.addr, &s->gwy.addr, af);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
s->lan.port = s->gwy.port;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
} else {
PF_ACPY(&s->lan.addr, daddr, af);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
s->lan.port = nport;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
PF_ACPY(&s->ext.addr, saddr, af);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
s->ext.port = 0;
if (nr != NULL) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
PF_ACPY(&s->gwy.addr, &pd->baddr, af);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
s->gwy.port = bport;
} else {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
PF_ACPY(&s->gwy.addr, &s->lan.addr, af);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
s->gwy.port = s->lan.port;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
s->creation = time_second;
s->expire = time_second;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->timeout = PFTM_ICMP_FIRST_PACKET;
pf_set_rt_ifp(s, saddr);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (sn != NULL) {
s->src_node = sn;
s->src_node->states++;
}
if (nsn != NULL) {
PF_ACPY(&nsn->raddr, &pd->naddr, af);
s->nat_src_node = nsn;
s->nat_src_node->states++;
}
if (pf_insert_state(BOUND_IFACE(r, kif), s)) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(&reason, PFRES_STATEINS);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_src_tree_remove_state(s);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
STATE_DEC_COUNTERS(s);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pool_put(&pf_state_pl, s);
return (PF_DROP);
} else
*sm = s;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (tag > 0) {
pf_tag_ref(tag);
s->tag = tag;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
#ifdef INET6
/* copy back packet headers if we performed IPv6 NAT operations */
if (rewrite)
m_copyback(m, off, sizeof(struct icmp6_hdr),
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)pd->hdr.icmp6);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#endif /* INET6 */
return (PF_PASS);
}
int
pf_test_other(struct pf_rule **rm, struct pf_state **sm, int direction,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pfi_kif *kif, struct mbuf *m, int off, void *h, struct pf_pdesc *pd,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct pf_rule **am, struct pf_ruleset **rsm, struct ifqueue *ifq)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_rule *nr = NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_rule *r, *a = NULL;
struct pf_ruleset *ruleset = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_src_node *nsn = NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_addr *saddr = pd->src, *daddr = pd->dst;
sa_family_t af = pd->af;
u_short reason;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int tag = -1, rtableid = -1;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
int asd = 0;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int match = 0;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (pf_check_congestion(ifq)) {
REASON_SET(&reason, PFRES_CONGEST);
return (PF_DROP);
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_FIRST(pf_main_ruleset.rules[PF_RULESET_FILTER].active.ptr);
if (direction == PF_OUT) {
/* check outgoing packet for BINAT/NAT */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((nr = pf_get_translation(pd, m, off, PF_OUT, kif, &nsn,
saddr, 0, daddr, 0, &pd->naddr, NULL)) != NULL) {
PF_ACPY(&pd->baddr, saddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
switch (af) {
#ifdef INET
case AF_INET:
pf_change_a(&saddr->v4.s_addr, pd->ip_sum,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pd->naddr.v4.s_addr, 0);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
PF_ACPY(saddr, &pd->naddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET6 */
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr->natpass)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pd->nat_rule = nr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
} else {
/* check incoming packet for BINAT/RDR */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((nr = pf_get_translation(pd, m, off, PF_IN, kif, &nsn,
saddr, 0, daddr, 0, &pd->naddr, NULL)) != NULL) {
PF_ACPY(&pd->baddr, daddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
switch (af) {
#ifdef INET
case AF_INET:
pf_change_a(&daddr->v4.s_addr,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pd->ip_sum, pd->naddr.v4.s_addr, 0);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
PF_ACPY(daddr, &pd->naddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET6 */
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr->natpass)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pd->nat_rule = nr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
}
while (r != NULL) {
r->evaluations++;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pfi_kif_match(r->kif, kif) == r->ifnot)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_IFP].ptr;
else if (r->direction && r->direction != direction)
r = r->skip[PF_SKIP_DIR].ptr;
else if (r->af && r->af != af)
r = r->skip[PF_SKIP_AF].ptr;
else if (r->proto && r->proto != pd->proto)
r = r->skip[PF_SKIP_PROTO].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (PF_MISMATCHAW(&r->src.addr, pd->src, af,
r->src.neg, kif))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_SRC_ADDR].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (PF_MISMATCHAW(&r->dst.addr, pd->dst, af,
r->dst.neg, NULL))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_DST_ADDR].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->tos && !(r->tos == pd->tos))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
else if (r->rule_flag & PFRULE_FRAGMENT)
r = TAILQ_NEXT(r, entries);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
else if (r->prob && r->prob <= arc4random())
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->match_tag && !pf_match_tag(m, r, pd->pf_mtag, &tag))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
else if (r->os_fingerprint != PF_OSFP_ANY)
r = TAILQ_NEXT(r, entries);
else {
if (r->tag)
tag = r->tag;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r->rtableid >= 0)
rtableid = r->rtableid;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (r->anchor == NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
match = 1;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
*rm = r;
*am = a;
*rsm = ruleset;
if ((*rm)->quick)
break;
r = TAILQ_NEXT(r, entries);
} else
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_step_into_anchor(&asd, &ruleset,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
PF_RULESET_FILTER, &r, &a, &match);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r == NULL && pf_step_out_of_anchor(&asd, &ruleset,
PF_RULESET_FILTER, &r, &a, &match))
break;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
r = *rm;
a = *am;
ruleset = *rsm;
REASON_SET(&reason, PFRES_MATCH);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r->log || (nr != NULL && nr->natpass && nr->log))
PFLOG_PACKET(kif, h, m, af, direction, reason, r->log ? r : nr,
a, ruleset, pd);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if ((r->action == PF_DROP) &&
((r->rule_flag & PFRULE_RETURNICMP) ||
(r->rule_flag & PFRULE_RETURN))) {
struct pf_addr *a = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr != NULL) {
if (direction == PF_OUT)
a = saddr;
else
a = daddr;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (a != NULL) {
switch (af) {
#ifdef INET
case AF_INET:
pf_change_a(&a->v4.s_addr, pd->ip_sum,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pd->baddr.v4.s_addr, 0);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
PF_ACPY(a, &pd->baddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET6 */
}
}
if ((af == AF_INET) && r->return_icmp)
pf_send_icmp(m, r->return_icmp >> 8,
r->return_icmp & 255, af, r);
else if ((af == AF_INET6) && r->return_icmp6)
pf_send_icmp(m, r->return_icmp6 >> 8,
r->return_icmp6 & 255, af, r);
}
if (r->action != PF_PASS)
return (PF_DROP);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pf_tag_packet(m, pd->pf_mtag, tag, rtableid)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
REASON_SET(&reason, PFRES_MEMORY);
return (PF_DROP);
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (r->keep_state || nr != NULL) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* create new state */
struct pf_state *s = NULL;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_src_node *sn = NULL;
/* check maximums */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (r->max_states && (r->states >= r->max_states)) {
pf_status.lcounters[LCNT_STATES]++;
REASON_SET(&reason, PFRES_MAXSTATES);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto cleanup;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
/* src node for filter rule */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if ((r->rule_flag & PFRULE_SRCTRACK ||
r->rpool.opts & PF_POOL_STICKYADDR) &&
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_insert_src_node(&sn, r, saddr, af) != 0) {
REASON_SET(&reason, PFRES_SRCLIMIT);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto cleanup;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
/* src node for translation rule */
if (nr != NULL && (nr->rpool.opts & PF_POOL_STICKYADDR) &&
((direction == PF_OUT &&
pf_insert_src_node(&nsn, nr, &pd->baddr, af) != 0) ||
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
(pf_insert_src_node(&nsn, nr, saddr, af) != 0))) {
REASON_SET(&reason, PFRES_SRCLIMIT);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto cleanup;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
s = pool_get(&pf_state_pl, PR_NOWAIT);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (s == NULL) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(&reason, PFRES_MEMORY);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
cleanup:
if (sn != NULL && sn->states == 0 && sn->expire == 0) {
RB_REMOVE(pf_src_tree, &tree_src_tracking, sn);
pf_status.scounters[SCNT_SRC_NODE_REMOVALS]++;
pf_status.src_nodes--;
pool_put(&pf_src_tree_pl, sn);
}
if (nsn != sn && nsn != NULL && nsn->states == 0 &&
nsn->expire == 0) {
RB_REMOVE(pf_src_tree, &tree_src_tracking, nsn);
pf_status.scounters[SCNT_SRC_NODE_REMOVALS]++;
pf_status.src_nodes--;
pool_put(&pf_src_tree_pl, nsn);
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_DROP);
}
bzero(s, sizeof(*s));
s->rule.ptr = r;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
s->nat_rule.ptr = nr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->anchor.ptr = a;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
STATE_INC_COUNTERS(s);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->allow_opts = r->allow_opts;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
s->log = r->log & PF_LOG_ALL;
if (nr != NULL)
s->log |= nr->log & PF_LOG_ALL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->proto = pd->proto;
s->direction = direction;
s->af = af;
if (direction == PF_OUT) {
PF_ACPY(&s->gwy.addr, saddr, af);
PF_ACPY(&s->ext.addr, daddr, af);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr != NULL)
PF_ACPY(&s->lan.addr, &pd->baddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
else
PF_ACPY(&s->lan.addr, &s->gwy.addr, af);
} else {
PF_ACPY(&s->lan.addr, daddr, af);
PF_ACPY(&s->ext.addr, saddr, af);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (nr != NULL)
PF_ACPY(&s->gwy.addr, &pd->baddr, af);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
else
PF_ACPY(&s->gwy.addr, &s->lan.addr, af);
}
s->src.state = PFOTHERS_SINGLE;
s->dst.state = PFOTHERS_NO_TRAFFIC;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
s->creation = time_second;
s->expire = time_second;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
s->timeout = PFTM_OTHER_FIRST_PACKET;
pf_set_rt_ifp(s, saddr);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (sn != NULL) {
s->src_node = sn;
s->src_node->states++;
}
if (nsn != NULL) {
PF_ACPY(&nsn->raddr, &pd->naddr, af);
s->nat_src_node = nsn;
s->nat_src_node->states++;
}
if (pf_insert_state(BOUND_IFACE(r, kif), s)) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(&reason, PFRES_STATEINS);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_src_tree_remove_state(s);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
STATE_DEC_COUNTERS(s);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pool_put(&pf_state_pl, s);
return (PF_DROP);
} else
*sm = s;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (tag > 0) {
pf_tag_ref(tag);
s->tag = tag;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
return (PF_PASS);
}
int
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_test_fragment(struct pf_rule **rm, int direction, struct pfi_kif *kif,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct mbuf *m, void *h, struct pf_pdesc *pd, struct pf_rule **am,
struct pf_ruleset **rsm)
{
struct pf_rule *r, *a = NULL;
struct pf_ruleset *ruleset = NULL;
sa_family_t af = pd->af;
u_short reason;
int tag = -1;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
int asd = 0;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int match = 0;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_FIRST(pf_main_ruleset.rules[PF_RULESET_FILTER].active.ptr);
while (r != NULL) {
r->evaluations++;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pfi_kif_match(r->kif, kif) == r->ifnot)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_IFP].ptr;
else if (r->direction && r->direction != direction)
r = r->skip[PF_SKIP_DIR].ptr;
else if (r->af && r->af != af)
r = r->skip[PF_SKIP_AF].ptr;
else if (r->proto && r->proto != pd->proto)
r = r->skip[PF_SKIP_PROTO].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (PF_MISMATCHAW(&r->src.addr, pd->src, af,
r->src.neg, kif))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_SRC_ADDR].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (PF_MISMATCHAW(&r->dst.addr, pd->dst, af,
r->dst.neg, NULL))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = r->skip[PF_SKIP_DST_ADDR].ptr;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->tos && !(r->tos == pd->tos))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
Reloop OpenBSD rev. 1.563: In pf_test_fragment(), ignore protocol-specific criteria for packets of different protocols. Reported by: des Obtained from: OpenBSD MFC after: 3 days
2007-11-21 10:12:52 +00:00
else if (r->os_fingerprint != PF_OSFP_ANY)
r = TAILQ_NEXT(r, entries);
else if (pd->proto == IPPROTO_UDP &&
(r->src.port_op || r->dst.port_op))
r = TAILQ_NEXT(r, entries);
else if (pd->proto == IPPROTO_TCP &&
(r->src.port_op || r->dst.port_op || r->flagset))
r = TAILQ_NEXT(r, entries);
else if ((pd->proto == IPPROTO_ICMP ||
pd->proto == IPPROTO_ICMPV6) &&
(r->type || r->code))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
else if (r->prob && r->prob <= arc4random())
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (r->match_tag && !pf_match_tag(m, r, pd->pf_mtag, &tag))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = TAILQ_NEXT(r, entries);
else {
if (r->anchor == NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
match = 1;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
*rm = r;
*am = a;
*rsm = ruleset;
if ((*rm)->quick)
break;
r = TAILQ_NEXT(r, entries);
} else
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_step_into_anchor(&asd, &ruleset,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
PF_RULESET_FILTER, &r, &a, &match);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (r == NULL && pf_step_out_of_anchor(&asd, &ruleset,
PF_RULESET_FILTER, &r, &a, &match))
break;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
r = *rm;
a = *am;
ruleset = *rsm;
REASON_SET(&reason, PFRES_MATCH);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (r->log)
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
PFLOG_PACKET(kif, h, m, af, direction, reason, r, a, ruleset,
pd);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (r->action != PF_PASS)
return (PF_DROP);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pf_tag_packet(m, pd->pf_mtag, tag, -1)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
REASON_SET(&reason, PFRES_MEMORY);
return (PF_DROP);
}
return (PF_PASS);
}
int
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
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,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
u_short *reason)
{
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_state_cmp key;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct tcphdr *th = pd->hdr.tcp;
u_int16_t win = ntohs(th->th_win);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
u_int32_t ack, end, seq, orig_seq;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
u_int8_t sws, dws;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
int ackskew;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
int copyback = 0;
struct pf_state_peer *src, *dst;
key.af = pd->af;
key.proto = IPPROTO_TCP;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (direction == PF_IN) {
PF_ACPY(&key.ext.addr, pd->src, key.af);
PF_ACPY(&key.gwy.addr, pd->dst, key.af);
key.ext.port = th->th_sport;
key.gwy.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.port = th->th_sport;
key.ext.port = th->th_dport;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
STATE_LOOKUP();
if (direction == (*state)->direction) {
src = &(*state)->src;
dst = &(*state)->dst;
} else {
src = &(*state)->dst;
dst = &(*state)->src;
}
if ((*state)->src.state == PF_TCPS_PROXY_SRC) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (direction != (*state)->direction) {
REASON_SET(reason, PFRES_SYNPROXY);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_SYNPROXY_DROP);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (th->th_flags & TH_SYN) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (ntohl(th->th_seq) != (*state)->src.seqlo) {
REASON_SET(reason, PFRES_SYNPROXY);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_DROP);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#ifdef __FreeBSD__
pf_send_tcp(NULL, (*state)->rule.ptr, pd->af, pd->dst,
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_send_tcp((*state)->rule.ptr, pd->af, pd->dst,
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pd->src, th->th_dport, th->th_sport,
(*state)->src.seqhi, ntohl(th->th_seq) + 1,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
TH_SYN|TH_ACK, 0, (*state)->src.mss, 0, 1,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
0, NULL, NULL);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(reason, PFRES_SYNPROXY);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_SYNPROXY_DROP);
} else if (!(th->th_flags & TH_ACK) ||
(ntohl(th->th_ack) != (*state)->src.seqhi + 1) ||
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
(ntohl(th->th_seq) != (*state)->src.seqlo + 1)) {
REASON_SET(reason, PFRES_SYNPROXY);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_DROP);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
} else if ((*state)->src_node != NULL &&
pf_src_connlimit(state)) {
REASON_SET(reason, PFRES_SRCLIMIT);
return (PF_DROP);
} else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
(*state)->src.state = PF_TCPS_PROXY_DST;
}
if ((*state)->src.state == PF_TCPS_PROXY_DST) {
struct pf_state_host *src, *dst;
if (direction == PF_OUT) {
src = &(*state)->gwy;
dst = &(*state)->ext;
} else {
src = &(*state)->ext;
dst = &(*state)->lan;
}
if (direction == (*state)->direction) {
if (((th->th_flags & (TH_SYN|TH_ACK)) != TH_ACK) ||
(ntohl(th->th_ack) != (*state)->src.seqhi + 1) ||
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
(ntohl(th->th_seq) != (*state)->src.seqlo + 1)) {
REASON_SET(reason, PFRES_SYNPROXY);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_DROP);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
(*state)->src.max_win = MAX(ntohs(th->th_win), 1);
if ((*state)->dst.seqhi == 1)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
(*state)->dst.seqhi = htonl(arc4random());
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#ifdef __FreeBSD__
pf_send_tcp(NULL, (*state)->rule.ptr, pd->af,
&src->addr,
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_send_tcp((*state)->rule.ptr, pd->af, &src->addr,
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
&dst->addr, src->port, dst->port,
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
(*state)->dst.seqhi, 0, TH_SYN, 0,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
(*state)->src.mss, 0, 0, (*state)->tag, NULL, NULL);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(reason, PFRES_SYNPROXY);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_SYNPROXY_DROP);
} else if (((th->th_flags & (TH_SYN|TH_ACK)) !=
(TH_SYN|TH_ACK)) ||
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
(ntohl(th->th_ack) != (*state)->dst.seqhi + 1)) {
REASON_SET(reason, PFRES_SYNPROXY);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_DROP);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
} else {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
(*state)->dst.max_win = MAX(ntohs(th->th_win), 1);
(*state)->dst.seqlo = ntohl(th->th_seq);
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#ifdef __FreeBSD__
pf_send_tcp(NULL, (*state)->rule.ptr, pd->af, pd->dst,
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_send_tcp((*state)->rule.ptr, pd->af, pd->dst,
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pd->src, th->th_dport, th->th_sport,
ntohl(th->th_ack), ntohl(th->th_seq) + 1,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
TH_ACK, (*state)->src.max_win, 0, 0, 0,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
(*state)->tag, NULL, NULL);
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#ifdef __FreeBSD__
pf_send_tcp(NULL, (*state)->rule.ptr, pd->af,
&src->addr,
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_send_tcp((*state)->rule.ptr, pd->af, &src->addr,
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
&dst->addr, src->port, dst->port,
(*state)->src.seqhi + 1, (*state)->src.seqlo + 1,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
TH_ACK, (*state)->dst.max_win, 0, 0, 1,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
0, NULL, NULL);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
(*state)->src.seqdiff = (*state)->dst.seqhi -
(*state)->src.seqlo;
(*state)->dst.seqdiff = (*state)->src.seqhi -
(*state)->dst.seqlo;
(*state)->src.seqhi = (*state)->src.seqlo +
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
(*state)->dst.max_win;
Bring back fix from rev. 1.28 which was lost during the import.
2005-05-04 16:03:01 +00:00
(*state)->dst.seqhi = (*state)->dst.seqlo +
(*state)->src.max_win;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
(*state)->src.wscale = (*state)->dst.wscale = 0;
(*state)->src.state = (*state)->dst.state =
TCPS_ESTABLISHED;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(reason, PFRES_SYNPROXY);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_SYNPROXY_DROP);
}
}
Import pf.c, rev 1.559 by markus: allow state reuse for tcp if both sides are in FIN_WAIT_2 and a new SYN arrives; ok dhartmei, henning, feedback aaron
2008-08-04 14:08:55 +00:00
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) {
printf("pf: state reuse ");
pf_print_state(*state);
pf_print_flags(th->th_flags);
printf("\n");
}
/* XXX make sure it's the same direction ?? */
(*state)->src.state = (*state)->dst.state = TCPS_CLOSED;
pf_unlink_state(*state);
*state = NULL;
return (PF_DROP);
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (src->wscale && dst->wscale && !(th->th_flags & TH_SYN)) {
sws = src->wscale & PF_WSCALE_MASK;
dws = dst->wscale & PF_WSCALE_MASK;
} else
sws = dws = 0;
/*
* Sequence tracking algorithm from Guido van Rooij's paper:
* http://www.madison-gurkha.com/publications/tcp_filtering/
* tcp_filtering.ps
*/
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
orig_seq = seq = ntohl(th->th_seq);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (src->seqlo == 0) {
/* First packet from this end. Set its state */
if ((pd->flags & PFDESC_TCP_NORM || dst->scrub) &&
src->scrub == NULL) {
if (pf_normalize_tcp_init(m, off, pd, th, src, dst)) {
REASON_SET(reason, PFRES_MEMORY);
return (PF_DROP);
}
}
/* Deferred generation of sequence number modulator */
if (dst->seqdiff && !src->seqdiff) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifdef __FreeBSD__
while ((src->seqdiff = pf_new_isn(*state) - seq) == 0)
;
#else
while ((src->seqdiff = tcp_rndiss_next() - seq) == 0)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
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 = 1;
} else {
ack = ntohl(th->th_ack);
}
end = seq + pd->p_len;
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);
if (src->wscale & PF_WSCALE_FLAG) {
/* Remove scale factor from initial
* window */
sws = src->wscale & PF_WSCALE_MASK;
win = ((u_int32_t)win + (1 << sws) - 1)
>> sws;
dws = dst->wscale & PF_WSCALE_MASK;
} else {
/* fixup other window */
dst->max_win <<= dst->wscale &
PF_WSCALE_MASK;
/* in case of a retrans SYN|ACK */
dst->wscale = 0;
}
}
}
if (th->th_flags & TH_FIN)
end++;
src->seqlo = seq;
if (src->state < TCPS_SYN_SENT)
src->state = TCPS_SYN_SENT;
/*
* May need to slide the window (seqhi may have been set by
* the crappy stack check or if we picked up the connection
* after establishment)
*/
if (src->seqhi == 1 ||
SEQ_GEQ(end + MAX(1, dst->max_win << dws), src->seqhi))
src->seqhi = end + MAX(1, dst->max_win << dws);
if (win > src->max_win)
src->max_win = win;
} else {
ack = ntohl(th->th_ack) - dst->seqdiff;
if (src->seqdiff) {
/* Modulate sequence numbers */
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 = 1;
}
end = seq + pd->p_len;
if (th->th_flags & TH_SYN)
end++;
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)) ||
/* broken tcp stacks do not set ack */
(dst->state < TCPS_SYN_SENT)) {
/*
* Many stacks (ours included) will set the ACK number in an
* FIN|ACK if the SYN times out -- no sequence to ACK.
*/
ack = dst->seqlo;
}
if (seq == end) {
/* Ease sequencing restrictions on no data packets */
seq = src->seqlo;
end = seq;
}
ackskew = dst->seqlo - ack;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
/*
* Need to demodulate the sequence numbers in any TCP SACK options
* (Selective ACK). We could optionally validate the SACK values
* against the current ACK window, either forwards or backwards, but
* I'm not confident that SACK has been implemented properly
* everywhere. It wouldn't surprise me if several stacks accidently
* SACK too far backwards of previously ACKed data. There really aren't
* any security implications of bad SACKing unless the target stack
* doesn't validate the option length correctly. Someone trying to
* spoof into a TCP connection won't bother blindly sending SACK
* options anyway.
*/
if (dst->seqdiff && (th->th_off << 2) > sizeof(struct tcphdr)) {
if (pf_modulate_sack(m, off, pd, th, dst))
copyback = 1;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#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 - (dst->max_win << dws)) &&
/* Retrans: not more than one window back */
(ackskew >= -MAXACKWINDOW) &&
/* Acking not more than one reassembled fragment backwards */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
(ackskew <= (MAXACKWINDOW << sws)) &&
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* Acking not more than one window forward */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
((th->th_flags & TH_RST) == 0 || orig_seq == src->seqlo ||
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
(orig_seq == src->seqlo + 1) || (pd->flags & PFDESC_IP_REAS) == 0)) {
/* Require an exact/+1 sequence match on resets when possible */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (dst->scrub || src->scrub) {
if (pf_normalize_tcp_stateful(m, off, pd, reason, th,
*state, src, dst, &copyback))
return (PF_DROP);
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* update max window */
if (src->max_win < win)
src->max_win = win;
/* synchronize sequencing */
if (SEQ_GT(end, src->seqlo))
src->seqlo = end;
/* slide the window of what the other end can send */
if (SEQ_GEQ(ack + (win << sws), dst->seqhi))
dst->seqhi = ack + MAX((win << sws), 1);
/* update states */
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)
src->state = TCPS_CLOSING;
if (th->th_flags & TH_ACK) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (dst->state == TCPS_SYN_SENT) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
dst->state = TCPS_ESTABLISHED;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (src->state == TCPS_ESTABLISHED &&
(*state)->src_node != NULL &&
pf_src_connlimit(state)) {
REASON_SET(reason, PFRES_SRCLIMIT);
return (PF_DROP);
}
} else if (dst->state == TCPS_CLOSING)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
dst->state = TCPS_FIN_WAIT_2;
}
if (th->th_flags & TH_RST)
src->state = dst->state = TCPS_TIME_WAIT;
/* update expire time */
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(*state)->expire = time_second;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (src->state >= TCPS_FIN_WAIT_2 &&
dst->state >= TCPS_FIN_WAIT_2)
(*state)->timeout = PFTM_TCP_CLOSED;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (src->state >= TCPS_CLOSING &&
dst->state >= TCPS_CLOSING)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
(*state)->timeout = PFTM_TCP_FIN_WAIT;
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)
(*state)->timeout = PFTM_TCP_CLOSING;
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 */
/*
* This currently handles three situations:
* 1) Stupid stacks will shotgun SYNs before their peer
* replies.
* 2) When PF catches an already established stream (the
* firewall rebooted, the state table was flushed, routes
* changed...)
* 3) Packets get funky immediately after the connection
* closes (this should catch Solaris spurious ACK|FINs
* that web servers like to spew after a close)
*
* This must be a little more careful than the above code
* since packet floods will also be caught here. We don't
* update the TTL here to mitigate the damage of a packet
* flood and so the same code can handle awkward establishment
* and a loosened connection close.
* In the establishment case, a correct peer response will
* validate the connection, go through the normal state code
* and keep updating the state TTL.
*/
if (pf_status.debug >= PF_DEBUG_MISC) {
printf("pf: loose state match: ");
pf_print_state(*state);
pf_print_flags(th->th_flags);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
printf(" seq=%u (%u) ack=%u len=%u ackskew=%d "
"pkts=%llu:%llu\n", seq, orig_seq, ack, pd->p_len,
#ifdef __FreeBSD__
ackskew, (unsigned long long)(*state)->packets[0],
(unsigned long long)(*state)->packets[1]);
#else
ackskew, (*state)->packets[0],
(*state)->packets[1]);
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (dst->scrub || src->scrub) {
if (pf_normalize_tcp_stateful(m, off, pd, reason, th,
*state, src, dst, &copyback))
return (PF_DROP);
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* update max window */
if (src->max_win < win)
src->max_win = win;
/* synchronize sequencing */
if (SEQ_GT(end, src->seqlo))
src->seqlo = end;
/* slide the window of what the other end can send */
if (SEQ_GEQ(ack + (win << sws), dst->seqhi))
dst->seqhi = ack + MAX((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)
src->state = TCPS_CLOSING;
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) {
/* Send RST for state mismatches during handshake */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (!(th->th_flags & TH_RST))
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#ifdef __FreeBSD__
pf_send_tcp(m, (*state)->rule.ptr, pd->af,
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pf_send_tcp((*state)->rule.ptr, pd->af,
Introduce a new entry point, mac_create_mbuf_from_firewall. This entry point exists to allow the mandatory access control policy to properly initialize mbufs generated by the firewall. An example where this might happen is keep alive packets, or ICMP error packets in response to other packets. This takes care of kernel panics associated with un-initialize mbuf labels when the firewall generates packets. [1] I modified this patch from it's original version, the initial patch introduced a number of entry points which were programmatically equivalent. So I introduced only one. Instead, we should leverage mac_create_mbuf_netlayer() which is used for similar situations, an example being icmp_error() This will minimize the impact associated with the MFC Submitted by: mlaier [1] MFC after: 1 week This is a RELENG_6 candidate
2006-09-12 04:25:13 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pd->dst, pd->src, th->th_dport,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
th->th_sport, ntohl(th->th_ack), 0,
TH_RST, 0, 0,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
(*state)->rule.ptr->return_ttl, 1, 0,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pd->eh, kif->pfik_ifp);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
src->seqlo = 0;
src->seqhi = 1;
src->max_win = 1;
} else if (pf_status.debug >= PF_DEBUG_MISC) {
printf("pf: BAD state: ");
pf_print_state(*state);
pf_print_flags(th->th_flags);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
printf(" seq=%u (%u) ack=%u len=%u ackskew=%d "
"pkts=%llu:%llu dir=%s,%s\n",
seq, orig_seq, ack, pd->p_len, ackskew,
#ifdef __FreeBSD__
(unsigned long long)(*state)->packets[0],
(unsigned long long)(*state)->packets[1],
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
(*state)->packets[0], (*state)->packets[1],
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
direction == PF_IN ? "in" : "out",
direction == (*state)->direction ? "fwd" : "rev");
printf("pf: State failure on: %c %c %c %c | %c %c\n",
SEQ_GEQ(src->seqhi, end) ? ' ' : '1',
SEQ_GEQ(seq, src->seqlo - (dst->max_win << dws)) ?
' ': '2',
(ackskew >= -MAXACKWINDOW) ? ' ' : '3',
(ackskew <= (MAXACKWINDOW << sws)) ? ' ' : '4',
SEQ_GEQ(src->seqhi + MAXACKWINDOW, end) ?' ' :'5',
SEQ_GEQ(seq, src->seqlo - MAXACKWINDOW) ?' ' :'6');
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(reason, PFRES_BADSTATE);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_DROP);
}
/* Any packets which have gotten here are to be passed */
/* translate source/destination address, if necessary */
if (STATE_TRANSLATE(*state)) {
if (direction == PF_OUT)
pf_change_ap(pd->src, &th->th_sport, pd->ip_sum,
&th->th_sum, &(*state)->gwy.addr,
(*state)->gwy.port, 0, pd->af);
else
pf_change_ap(pd->dst, &th->th_dport, pd->ip_sum,
&th->th_sum, &(*state)->lan.addr,
(*state)->lan.port, 0, pd->af);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
m_copyback(m, off, sizeof(*th), (caddr_t)th);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
} else if (copyback) {
/* Copyback sequence modulation or stateful scrub changes */
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
m_copyback(m, off, sizeof(*th), (caddr_t)th);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
return (PF_PASS);
}
int
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
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)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
struct pf_state_peer *src, *dst;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_state_cmp key;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct udphdr *uh = pd->hdr.udp;
key.af = pd->af;
key.proto = IPPROTO_UDP;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (direction == PF_IN) {
PF_ACPY(&key.ext.addr, pd->src, key.af);
PF_ACPY(&key.gwy.addr, pd->dst, key.af);
key.ext.port = uh->uh_sport;
key.gwy.port = uh->uh_dport;
} else {
PF_ACPY(&key.lan.addr, pd->src, key.af);
PF_ACPY(&key.ext.addr, pd->dst, key.af);
key.lan.port = uh->uh_sport;
key.ext.port = uh->uh_dport;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
STATE_LOOKUP();
if (direction == (*state)->direction) {
src = &(*state)->src;
dst = &(*state)->dst;
} else {
src = &(*state)->dst;
dst = &(*state)->src;
}
/* update states */
if (src->state < PFUDPS_SINGLE)
src->state = PFUDPS_SINGLE;
if (dst->state == PFUDPS_SINGLE)
dst->state = PFUDPS_MULTIPLE;
/* update expire time */
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(*state)->expire = time_second;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (src->state == PFUDPS_MULTIPLE && dst->state == PFUDPS_MULTIPLE)
(*state)->timeout = PFTM_UDP_MULTIPLE;
else
(*state)->timeout = PFTM_UDP_SINGLE;
/* translate source/destination address, if necessary */
if (STATE_TRANSLATE(*state)) {
if (direction == PF_OUT)
pf_change_ap(pd->src, &uh->uh_sport, pd->ip_sum,
&uh->uh_sum, &(*state)->gwy.addr,
(*state)->gwy.port, 1, pd->af);
else
pf_change_ap(pd->dst, &uh->uh_dport, pd->ip_sum,
&uh->uh_sum, &(*state)->lan.addr,
(*state)->lan.port, 1, pd->af);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
m_copyback(m, off, sizeof(*uh), (caddr_t)uh);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
return (PF_PASS);
}
int
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_test_state_icmp(struct pf_state **state, int direction, struct pfi_kif *kif,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct mbuf *m, int off, void *h, struct pf_pdesc *pd, u_short *reason)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
struct pf_addr *saddr = pd->src, *daddr = pd->dst;
Shut up a couple of annoying gcc warnings. Do not enclose the fixes with #ifdefs in order to loop it back to OpenBSD after the next import. There are a some implicit asserts involved which might be better spelled out explicitly (af == AF_INET ...) Approved by: bms(mentor)
2004-03-30 18:28:30 +00:00
u_int16_t icmpid = 0; /* make the compiler happy */
u_int16_t *icmpsum = NULL; /* make the compiler happy */
u_int8_t icmptype = 0; /* make the compiler happy */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
int state_icmp = 0;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_state_cmp key;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
switch (pd->proto) {
#ifdef INET
case IPPROTO_ICMP:
icmptype = pd->hdr.icmp->icmp_type;
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)
state_icmp++;
break;
#endif /* INET */
#ifdef INET6
case IPPROTO_ICMPV6:
icmptype = pd->hdr.icmp6->icmp6_type;
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)
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;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (direction == PF_IN) {
PF_ACPY(&key.ext.addr, pd->src, key.af);
PF_ACPY(&key.gwy.addr, pd->dst, key.af);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
key.ext.port = 0;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
key.gwy.port = icmpid;
} else {
PF_ACPY(&key.lan.addr, pd->src, key.af);
PF_ACPY(&key.ext.addr, pd->dst, key.af);
key.lan.port = icmpid;
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
key.ext.port = 0;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
STATE_LOOKUP();
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(*state)->expire = time_second;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
(*state)->timeout = PFTM_ICMP_ERROR_REPLY;
/* translate source/destination address, if necessary */
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
if (STATE_TRANSLATE(*state)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (direction == PF_OUT) {
switch (pd->af) {
#ifdef INET
case AF_INET:
pf_change_a(&saddr->v4.s_addr,
pd->ip_sum,
(*state)->gwy.addr.v4.s_addr, 0);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
pd->hdr.icmp->icmp_cksum =
pf_cksum_fixup(
pd->hdr.icmp->icmp_cksum, icmpid,
(*state)->gwy.port, 0);
pd->hdr.icmp->icmp_id =
(*state)->gwy.port;
m_copyback(m, off, ICMP_MINLEN,
Unbreak the build. Committed from the wrong directory.
2005-09-08 17:42:42 +00:00
(caddr_t)pd->hdr.icmp);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
pf_change_a6(saddr,
&pd->hdr.icmp6->icmp6_cksum,
&(*state)->gwy.addr, 0);
m_copyback(m, off,
sizeof(struct icmp6_hdr),
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)pd->hdr.icmp6);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET6 */
}
} else {
switch (pd->af) {
#ifdef INET
case AF_INET:
pf_change_a(&daddr->v4.s_addr,
pd->ip_sum,
(*state)->lan.addr.v4.s_addr, 0);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
pd->hdr.icmp->icmp_cksum =
pf_cksum_fixup(
pd->hdr.icmp->icmp_cksum, icmpid,
(*state)->lan.port, 0);
pd->hdr.icmp->icmp_id =
(*state)->lan.port;
m_copyback(m, off, ICMP_MINLEN,
Unbreak the build. Committed from the wrong directory.
2005-09-08 17:42:42 +00:00
(caddr_t)pd->hdr.icmp);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
pf_change_a6(daddr,
&pd->hdr.icmp6->icmp6_cksum,
&(*state)->lan.addr, 0);
m_copyback(m, off,
sizeof(struct icmp6_hdr),
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)pd->hdr.icmp6);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET6 */
}
}
}
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;
#ifdef INET
struct ip h2;
#endif /* INET */
#ifdef INET6
struct ip6_hdr h2_6;
int terminal = 0;
#endif /* INET6 */
Shut up a couple of annoying gcc warnings. Do not enclose the fixes with #ifdefs in order to loop it back to OpenBSD after the next import. There are a some implicit asserts involved which might be better spelled out explicitly (af == AF_INET ...) Approved by: bms(mentor)
2004-03-30 18:28:30 +00:00
int ipoff2 = 0; /* make the compiler happy */
int off2 = 0; /* make the compiler happy */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pd2.af = pd->af;
switch (pd->af) {
#ifdef INET
case AF_INET:
/* offset of h2 in mbuf chain */
ipoff2 = off + ICMP_MINLEN;
if (!pf_pull_hdr(m, ipoff2, &h2, sizeof(h2),
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
NULL, reason, pd2.af)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
DPFPRINTF(PF_DEBUG_MISC,
("pf: ICMP error message too short "
"(ip)\n"));
return (PF_DROP);
}
/*
* ICMP error messages don't refer to non-first
* fragments
*/
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (h2.ip_off & htons(IP_OFFMASK)) {
REASON_SET(reason, PFRES_FRAG);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_DROP);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* offset of protocol header that follows h2 */
off2 = ipoff2 + (h2.ip_hl << 2);
pd2.proto = h2.ip_p;
pd2.src = (struct pf_addr *)&h2.ip_src;
pd2.dst = (struct pf_addr *)&h2.ip_dst;
pd2.ip_sum = &h2.ip_sum;
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
ipoff2 = off + sizeof(struct icmp6_hdr);
if (!pf_pull_hdr(m, ipoff2, &h2_6, sizeof(h2_6),
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
NULL, reason, pd2.af)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
DPFPRINTF(PF_DEBUG_MISC,
("pf: ICMP error message too short "
"(ip6)\n"));
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.ip_sum = NULL;
off2 = ipoff2 + sizeof(h2_6);
do {
switch (pd2.proto) {
case IPPROTO_FRAGMENT:
/*
* ICMPv6 error messages for
* non-first fragments
*/
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(reason, PFRES_FRAG);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_DROP);
case IPPROTO_AH:
case IPPROTO_HOPOPTS:
case IPPROTO_ROUTING:
case IPPROTO_DSTOPTS: {
/* get next header and header length */
struct ip6_ext opt6;
if (!pf_pull_hdr(m, off2, &opt6,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
sizeof(opt6), NULL, reason,
pd2.af)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
DPFPRINTF(PF_DEBUG_MISC,
("pf: ICMPv6 short opt\n"));
return (PF_DROP);
}
if (pd2.proto == IPPROTO_AH)
off2 += (opt6.ip6e_len + 2) * 4;
else
off2 += (opt6.ip6e_len + 1) * 8;
pd2.proto = opt6.ip6e_nxt;
/* goto the next header */
break;
}
default:
terminal++;
break;
}
} while (!terminal);
break;
#endif /* INET6 */
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifdef __FreeBSD__
default:
panic("AF not supported: %d", pd->af);
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
switch (pd2.proto) {
case IPPROTO_TCP: {
struct tcphdr th;
u_int32_t seq;
struct pf_state_peer *src, *dst;
u_int8_t dws;
Commit import of OpenBSD-stable fix: Fix by dhartmei@ and mcbride@ 1.433 Properly m_copyback() modified TCP sequence number after demodulation 1.432 Fix icmp checksum when sequence number modlation is being used. Also fix a daddr vs saddr cut-n-paste error in ICMP error handling. Fixes PR 3724 Obtained from: OpenBSD Reviewed by: dhartmei Approved by: rwatson
2004-04-11 17:35:40 +00:00
int copyback = 0;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/*
* 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.
*/
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (!pf_pull_hdr(m, off2, &th, 8, NULL, reason,
pd2.af)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
DPFPRINTF(PF_DEBUG_MISC,
("pf: ICMP error message too short "
"(tcp)\n"));
return (PF_DROP);
}
key.af = pd2.af;
key.proto = IPPROTO_TCP;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (direction == PF_IN) {
PF_ACPY(&key.ext.addr, pd2.dst, key.af);
PF_ACPY(&key.gwy.addr, pd2.src, key.af);
key.ext.port = th.th_dport;
key.gwy.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.port = th.th_dport;
key.ext.port = th.th_sport;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
STATE_LOOKUP();
if (direction == (*state)->direction) {
src = &(*state)->dst;
dst = &(*state)->src;
} else {
src = &(*state)->src;
dst = &(*state)->dst;
}
When checking the sequence number of a TCP header embedded in an ICMP error message, do not access th_flags. The field is beyond the first eight bytes of the header that are required to be present and were pulled up in the mbuf. A random value of th_flags can have TH_SYN set, which made the sequence number comparison not apply the window scaling factor, which led to legitimate ICMP(v6) packets getting blocked with "BAD ICMP" debug log messages (if enabled with pfctl -xm), thus breaking PMTU discovery. Triggering the bug requires TCP window scaling to be enabled (sysctl net.inet.tcp.rfc1323, enabled by default) on both end- points of the TCP connection. Large scaling factors increase the probability of triggering the bug. PR: kern/115413: [ipv6] ipv6 pmtu not working Tested by: Jacek Zapala Reviewed by: mlaier Approved by: re (kensmith)
2007-08-23 09:30:58 +00:00
if (src->wscale && dst->wscale)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
dws = dst->wscale & PF_WSCALE_MASK;
else
dws = 0;
/* Demodulate sequence number */
seq = ntohl(th.th_seq) - src->seqdiff;
Commit import of OpenBSD-stable fix: Fix by dhartmei@ and mcbride@ 1.433 Properly m_copyback() modified TCP sequence number after demodulation 1.432 Fix icmp checksum when sequence number modlation is being used. Also fix a daddr vs saddr cut-n-paste error in ICMP error handling. Fixes PR 3724 Obtained from: OpenBSD Reviewed by: dhartmei Approved by: rwatson
2004-04-11 17:35:40 +00:00
if (src->seqdiff) {
pf_change_a(&th.th_seq, icmpsum,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
htonl(seq), 0);
Commit import of OpenBSD-stable fix: Fix by dhartmei@ and mcbride@ 1.433 Properly m_copyback() modified TCP sequence number after demodulation 1.432 Fix icmp checksum when sequence number modlation is being used. Also fix a daddr vs saddr cut-n-paste error in ICMP error handling. Fixes PR 3724 Obtained from: OpenBSD Reviewed by: dhartmei Approved by: rwatson
2004-04-11 17:35:40 +00:00
copyback = 1;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (!SEQ_GEQ(src->seqhi, seq) ||
!SEQ_GEQ(seq, src->seqlo - (dst->max_win << dws))) {
if (pf_status.debug >= PF_DEBUG_MISC) {
printf("pf: BAD ICMP %d:%d ",
icmptype, pd->hdr.icmp->icmp_code);
pf_print_host(pd->src, 0, pd->af);
printf(" -> ");
pf_print_host(pd->dst, 0, pd->af);
printf(" state: ");
pf_print_state(*state);
printf(" seq=%u\n", seq);
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(reason, PFRES_BADSTATE);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_DROP);
}
if (STATE_TRANSLATE(*state)) {
if (direction == PF_IN) {
pf_change_icmp(pd2.src, &th.th_sport,
Commit import of OpenBSD-stable fix: Fix by dhartmei@ and mcbride@ 1.433 Properly m_copyback() modified TCP sequence number after demodulation 1.432 Fix icmp checksum when sequence number modlation is being used. Also fix a daddr vs saddr cut-n-paste error in ICMP error handling. Fixes PR 3724 Obtained from: OpenBSD Reviewed by: dhartmei Approved by: rwatson
2004-04-11 17:35:40 +00:00
daddr, &(*state)->lan.addr,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
(*state)->lan.port, NULL,
pd2.ip_sum, icmpsum,
pd->ip_sum, 0, pd2.af);
} else {
pf_change_icmp(pd2.dst, &th.th_dport,
saddr, &(*state)->gwy.addr,
(*state)->gwy.port, NULL,
pd2.ip_sum, icmpsum,
pd->ip_sum, 0, pd2.af);
}
Commit import of OpenBSD-stable fix: Fix by dhartmei@ and mcbride@ 1.433 Properly m_copyback() modified TCP sequence number after demodulation 1.432 Fix icmp checksum when sequence number modlation is being used. Also fix a daddr vs saddr cut-n-paste error in ICMP error handling. Fixes PR 3724 Obtained from: OpenBSD Reviewed by: dhartmei Approved by: rwatson
2004-04-11 17:35:40 +00:00
copyback = 1;
}
if (copyback) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
switch (pd2.af) {
#ifdef INET
case AF_INET:
m_copyback(m, off, ICMP_MINLEN,
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)pd->hdr.icmp);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
m_copyback(m, ipoff2, sizeof(h2),
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)&h2);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
m_copyback(m, off,
sizeof(struct icmp6_hdr),
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)pd->hdr.icmp6);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
m_copyback(m, ipoff2, sizeof(h2_6),
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)&h2_6);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET6 */
}
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
m_copyback(m, off2, 8, (caddr_t)&th);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
return (PF_PASS);
break;
}
case IPPROTO_UDP: {
struct udphdr uh;
if (!pf_pull_hdr(m, off2, &uh, sizeof(uh),
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
NULL, reason, pd2.af)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
DPFPRINTF(PF_DEBUG_MISC,
("pf: ICMP error message too short "
"(udp)\n"));
return (PF_DROP);
}
key.af = pd2.af;
key.proto = IPPROTO_UDP;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (direction == PF_IN) {
PF_ACPY(&key.ext.addr, pd2.dst, key.af);
PF_ACPY(&key.gwy.addr, pd2.src, key.af);
key.ext.port = uh.uh_dport;
key.gwy.port = uh.uh_sport;
} else {
PF_ACPY(&key.lan.addr, pd2.dst, key.af);
PF_ACPY(&key.ext.addr, pd2.src, key.af);
key.lan.port = uh.uh_dport;
key.ext.port = uh.uh_sport;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
STATE_LOOKUP();
if (STATE_TRANSLATE(*state)) {
if (direction == PF_IN) {
pf_change_icmp(pd2.src, &uh.uh_sport,
daddr, &(*state)->lan.addr,
(*state)->lan.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)->gwy.addr,
(*state)->gwy.port, &uh.uh_sum,
pd2.ip_sum, icmpsum,
pd->ip_sum, 1, pd2.af);
}
switch (pd2.af) {
#ifdef INET
case AF_INET:
m_copyback(m, off, ICMP_MINLEN,
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)pd->hdr.icmp);
m_copyback(m, ipoff2, sizeof(h2),
(caddr_t)&h2);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
m_copyback(m, off,
sizeof(struct icmp6_hdr),
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)pd->hdr.icmp6);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
m_copyback(m, ipoff2, sizeof(h2_6),
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)&h2_6);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET6 */
}
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
m_copyback(m, off2, sizeof(uh),
(caddr_t)&uh);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
return (PF_PASS);
break;
}
#ifdef INET
case IPPROTO_ICMP: {
struct icmp iih;
if (!pf_pull_hdr(m, off2, &iih, ICMP_MINLEN,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
NULL, reason, pd2.af)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
DPFPRINTF(PF_DEBUG_MISC,
("pf: ICMP error message too short i"
"(icmp)\n"));
return (PF_DROP);
}
key.af = pd2.af;
key.proto = IPPROTO_ICMP;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (direction == PF_IN) {
PF_ACPY(&key.ext.addr, pd2.dst, key.af);
PF_ACPY(&key.gwy.addr, pd2.src, key.af);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
key.ext.port = 0;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
key.gwy.port = iih.icmp_id;
} else {
PF_ACPY(&key.lan.addr, pd2.dst, key.af);
PF_ACPY(&key.ext.addr, pd2.src, key.af);
key.lan.port = iih.icmp_id;
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
key.ext.port = 0;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
STATE_LOOKUP();
if (STATE_TRANSLATE(*state)) {
if (direction == PF_IN) {
pf_change_icmp(pd2.src, &iih.icmp_id,
daddr, &(*state)->lan.addr,
(*state)->lan.port, NULL,
pd2.ip_sum, icmpsum,
pd->ip_sum, 0, AF_INET);
} else {
pf_change_icmp(pd2.dst, &iih.icmp_id,
saddr, &(*state)->gwy.addr,
(*state)->gwy.port, NULL,
pd2.ip_sum, icmpsum,
pd->ip_sum, 0, AF_INET);
}
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
m_copyback(m, off, ICMP_MINLEN,
(caddr_t)pd->hdr.icmp);
m_copyback(m, ipoff2, sizeof(h2),
(caddr_t)&h2);
m_copyback(m, off2, ICMP_MINLEN,
(caddr_t)&iih);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
return (PF_PASS);
break;
}
#endif /* INET */
#ifdef INET6
case IPPROTO_ICMPV6: {
struct icmp6_hdr iih;
if (!pf_pull_hdr(m, off2, &iih,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
sizeof(struct icmp6_hdr), NULL, reason, pd2.af)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
DPFPRINTF(PF_DEBUG_MISC,
("pf: ICMP error message too short "
"(icmp6)\n"));
return (PF_DROP);
}
key.af = pd2.af;
key.proto = IPPROTO_ICMPV6;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (direction == PF_IN) {
PF_ACPY(&key.ext.addr, pd2.dst, key.af);
PF_ACPY(&key.gwy.addr, pd2.src, key.af);
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
key.ext.port = 0;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
key.gwy.port = iih.icmp6_id;
} else {
PF_ACPY(&key.lan.addr, pd2.dst, key.af);
PF_ACPY(&key.ext.addr, pd2.src, key.af);
key.lan.port = iih.icmp6_id;
Commit imported changes to HEAD: pf_ioctl.c Revision 1.153 Sun Aug 7 11:37:33 2005 UTC by dhartmei | verify ticket in DIOCADDADDR, from Boris Polevoy, ok deraadt@ pf_ioctl.c Revision 1.158 Mon Sep 5 14:51:08 2005 UTC by dhartmei | in DIOCCHANGERULE, properly initialize table, if used in NAT rule. | from Boris Polevoy <vapcom at mail dot ru>, ok mcbride@ pf.c Revision 1.502 Mon Aug 22 11:54:25 2005 UTC by dhartmei | when nat'ing icmp 'connections', replace icmp id with proxy values | (similar to proxy ports for tcp/udp). not all clients use | per-invokation random ids, this allows multiple concurrent | connections from such clients. | thanks for testing to Rod Whitworth, "looks ok" markus@ pf.c Revision 1.501 Mon Aug 22 09:48:05 2005 UTC by dhartmei | fix rdr to bitmask replacement address pool. patch from Max Laier, | reported by Boris Polevoy, tested by Jean Debogue, ok henning@ Obtained from: OpenBSD MFC after: 3 days
2005-09-08 15:06:52 +00:00
key.ext.port = 0;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
STATE_LOOKUP();
if (STATE_TRANSLATE(*state)) {
if (direction == PF_IN) {
pf_change_icmp(pd2.src, &iih.icmp6_id,
daddr, &(*state)->lan.addr,
(*state)->lan.port, NULL,
pd2.ip_sum, icmpsum,
pd->ip_sum, 0, AF_INET6);
} else {
pf_change_icmp(pd2.dst, &iih.icmp6_id,
saddr, &(*state)->gwy.addr,
(*state)->gwy.port, NULL,
pd2.ip_sum, icmpsum,
pd->ip_sum, 0, AF_INET6);
}
m_copyback(m, off, sizeof(struct icmp6_hdr),
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)pd->hdr.icmp6);
m_copyback(m, ipoff2, sizeof(h2_6),
(caddr_t)&h2_6);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
m_copyback(m, off2, sizeof(struct icmp6_hdr),
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)&iih);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
return (PF_PASS);
break;
}
#endif /* INET6 */
default: {
key.af = pd2.af;
key.proto = pd2.proto;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (direction == PF_IN) {
PF_ACPY(&key.ext.addr, pd2.dst, key.af);
PF_ACPY(&key.gwy.addr, pd2.src, key.af);
key.ext.port = 0;
key.gwy.port = 0;
} else {
PF_ACPY(&key.lan.addr, pd2.dst, key.af);
PF_ACPY(&key.ext.addr, pd2.src, key.af);
key.lan.port = 0;
key.ext.port = 0;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
STATE_LOOKUP();
if (STATE_TRANSLATE(*state)) {
if (direction == PF_IN) {
pf_change_icmp(pd2.src, NULL,
daddr, &(*state)->lan.addr,
0, NULL,
pd2.ip_sum, icmpsum,
pd->ip_sum, 0, pd2.af);
} else {
pf_change_icmp(pd2.dst, NULL,
saddr, &(*state)->gwy.addr,
0, NULL,
pd2.ip_sum, icmpsum,
pd->ip_sum, 0, pd2.af);
}
switch (pd2.af) {
#ifdef INET
case AF_INET:
m_copyback(m, off, ICMP_MINLEN,
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)pd->hdr.icmp);
m_copyback(m, ipoff2, sizeof(h2),
(caddr_t)&h2);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
m_copyback(m, off,
sizeof(struct icmp6_hdr),
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)pd->hdr.icmp6);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
m_copyback(m, ipoff2, sizeof(h2_6),
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(caddr_t)&h2_6);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET6 */
}
}
return (PF_PASS);
break;
}
}
}
}
int
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_test_state_other(struct pf_state **state, int direction, struct pfi_kif *kif,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct pf_pdesc *pd)
{
struct pf_state_peer *src, *dst;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_state_cmp key;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
key.af = pd->af;
key.proto = pd->proto;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (direction == PF_IN) {
PF_ACPY(&key.ext.addr, pd->src, key.af);
PF_ACPY(&key.gwy.addr, pd->dst, key.af);
key.ext.port = 0;
key.gwy.port = 0;
} else {
PF_ACPY(&key.lan.addr, pd->src, key.af);
PF_ACPY(&key.ext.addr, pd->dst, key.af);
key.lan.port = 0;
key.ext.port = 0;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
STATE_LOOKUP();
if (direction == (*state)->direction) {
src = &(*state)->src;
dst = &(*state)->dst;
} else {
src = &(*state)->dst;
dst = &(*state)->src;
}
/* update states */
if (src->state < PFOTHERS_SINGLE)
src->state = PFOTHERS_SINGLE;
if (dst->state == PFOTHERS_SINGLE)
dst->state = PFOTHERS_MULTIPLE;
/* update expire time */
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
(*state)->expire = time_second;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (src->state == PFOTHERS_MULTIPLE && dst->state == PFOTHERS_MULTIPLE)
(*state)->timeout = PFTM_OTHER_MULTIPLE;
else
(*state)->timeout = PFTM_OTHER_SINGLE;
/* translate source/destination address, if necessary */
if (STATE_TRANSLATE(*state)) {
if (direction == PF_OUT)
switch (pd->af) {
#ifdef INET
case AF_INET:
pf_change_a(&pd->src->v4.s_addr,
pd->ip_sum, (*state)->gwy.addr.v4.s_addr,
0);
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
PF_ACPY(pd->src, &(*state)->gwy.addr, pd->af);
break;
#endif /* INET6 */
}
else
switch (pd->af) {
#ifdef INET
case AF_INET:
pf_change_a(&pd->dst->v4.s_addr,
pd->ip_sum, (*state)->lan.addr.v4.s_addr,
0);
break;
#endif /* INET */
#ifdef INET6
case AF_INET6:
PF_ACPY(pd->dst, &(*state)->lan.addr, pd->af);
break;
#endif /* INET6 */
}
}
return (PF_PASS);
}
/*
* ipoff and off are measured from the start of the mbuf chain.
* h must be at "ipoff" on the mbuf chain.
*/
void *
pf_pull_hdr(struct mbuf *m, int off, void *p, int len,
u_short *actionp, u_short *reasonp, sa_family_t af)
{
switch (af) {
#ifdef INET
case AF_INET: {
struct ip *h = mtod(m, struct ip *);
u_int16_t fragoff = (ntohs(h->ip_off) & IP_OFFMASK) << 3;
if (fragoff) {
if (fragoff >= len)
ACTION_SET(actionp, PF_PASS);
else {
ACTION_SET(actionp, PF_DROP);
REASON_SET(reasonp, PFRES_FRAG);
}
return (NULL);
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (m->m_pkthdr.len < off + len ||
ntohs(h->ip_len) < off + len) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
ACTION_SET(actionp, PF_DROP);
REASON_SET(reasonp, PFRES_SHORT);
return (NULL);
}
break;
}
#endif /* INET */
#ifdef INET6
case AF_INET6: {
struct ip6_hdr *h = mtod(m, struct ip6_hdr *);
if (m->m_pkthdr.len < 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);
}
break;
}
#endif /* INET6 */
}
m_copydata(m, off, len, p);
return (p);
}
int
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_routable(struct pf_addr *addr, sa_family_t af, struct pfi_kif *kif)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
struct sockaddr_in *dst;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
int ret = 1;
int check_mpath;
#ifndef __FreeBSD__
extern int ipmultipath;
#endif
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#ifdef INET6
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifndef __FreeBSD__
extern int ip6_multipath;
#endif
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
struct sockaddr_in6 *dst6;
struct route_in6 ro;
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct route ro;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct radix_node *rn;
struct rtentry *rt;
struct ifnet *ifp;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
check_mpath = 0;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
bzero(&ro, sizeof(ro));
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
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;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifndef __FreeBSD__ /* MULTIPATH_ROUTING */
if (ipmultipath)
check_mpath = 1;
#endif
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
break;
#ifdef 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;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifndef __FreeBSD__ /* MULTIPATH_ROUTING */
if (ip6_multipath)
check_mpath = 1;
#endif
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
break;
#endif /* INET6 */
default:
return (0);
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
/* Skip checks for ipsec interfaces */
if (kif != NULL && kif->pfik_ifp->if_type == IFT_ENC)
goto out;
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Add code to allow the system to handle multiple routing tables. This particular implementation is designed to be fully backwards compatible and to be MFC-able to 7.x (and 6.x) Currently the only protocol that can make use of the multiple tables is IPv4 Similar functionality exists in OpenBSD and Linux. From my notes: ----- One thing where FreeBSD has been falling behind, and which by chance I have some time to work on is "policy based routing", which allows different packet streams to be routed by more than just the destination address. Constraints: ------------ I want to make some form of this available in the 6.x tree (and by extension 7.x) , but FreeBSD in general needs it so I might as well do it in -current and back port the portions I need. One of the ways that this can be done is to have the ability to instantiate multiple kernel routing tables (which I will now refer to as "Forwarding Information Bases" or "FIBs" for political correctness reasons). Which FIB a particular packet uses to make the next hop decision can be decided by a number of mechanisms. The policies these mechanisms implement are the "Policies" referred to in "Policy based routing". One of the constraints I have if I try to back port this work to 6.x is that it must be implemented as a EXTENSION to the existing ABIs in 6.x so that third party applications do not need to be recompiled in timespan of the branch. This first version will not have some of the bells and whistles that will come with later versions. It will, for example, be limited to 16 tables in the first commit. Implementation method, Compatible version. (part 1) ------------------------------- For this reason I have implemented a "sufficient subset" of a multiple routing table solution in Perforce, and back-ported it to 6.x. (also in Perforce though not always caught up with what I have done in -current/P4). The subset allows a number of FIBs to be defined at compile time (8 is sufficient for my purposes in 6.x) and implements the changes needed to allow IPV4 to use them. I have not done the changes for ipv6 simply because I do not need it, and I do not have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it. Other protocol families are left untouched and should there be users with proprietary protocol families, they should continue to work and be oblivious to the existence of the extra FIBs. To understand how this is done, one must know that the current FIB code starts everything off with a single dimensional array of pointers to FIB head structures (One per protocol family), each of which in turn points to the trie of routes available to that family. The basic change in the ABI compatible version of the change is to extent that array to be a 2 dimensional array, so that instead of protocol family X looking at rt_tables[X] for the table it needs, it looks at rt_tables[Y][X] when for all protocol families except ipv4 Y is always 0. Code that is unaware of the change always just sees the first row of the table, which of course looks just like the one dimensional array that existed before. The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign() are all maintained, but refer only to the first row of the array, so that existing callers in proprietary protocols can continue to do the "right thing". Some new entry points are added, for the exclusive use of ipv4 code called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(), which have an extra argument which refers the code to the correct row. In addition, there are some new entry points (currently called rtalloc_fib() and friends) that check the Address family being looked up and call either rtalloc() (and friends) if the protocol is not IPv4 forcing the action to row 0 or to the appropriate row if it IS IPv4 (and that info is available). These are for calling from code that is not specific to any particular protocol. The way these are implemented would change in the non ABI preserving code to be added later. One feature of the first version of the code is that for ipv4, the interface routes show up automatically on all the FIBs, so that no matter what FIB you select you always have the basic direct attached hosts available to you. (rtinit() does this automatically). You CAN delete an interface route from one FIB should you want to but by default it's there. ARP information is also available in each FIB. It's assumed that the same machine would have the same MAC address, regardless of which FIB you are using to get to it. This brings us as to how the correct FIB is selected for an outgoing IPV4 packet. Firstly, all packets have a FIB associated with them. if nothing has been done to change it, it will be FIB 0. The FIB is changed in the following ways. Packets fall into one of a number of classes. 1/ locally generated packets, coming from a socket/PCB. Such packets select a FIB from a number associated with the socket/PCB. This in turn is inherited from the process, but can be changed by a socket option. The process in turn inherits it on fork. I have written a utility call setfib that acts a bit like nice.. setfib -3 ping target.example.com # will use fib 3 for ping. It is an obvious extension to make it a property of a jail but I have not done so. It can be achieved by combining the setfib and jail commands. 2/ packets received on an interface for forwarding. By default these packets would use table 0, (or possibly a number settable in a sysctl(not yet)). but prior to routing the firewall can inspect them (see below). (possibly in the future you may be able to associate a FIB with packets received on an interface.. An ifconfig arg, but not yet.) 3/ packets inspected by a packet classifier, which can arbitrarily associate a fib with it on a packet by packet basis. A fib assigned to a packet by a packet classifier (such as ipfw) would over-ride a fib associated by a more default source. (such as cases 1 or 2). 4/ a tcp listen socket associated with a fib will generate accept sockets that are associated with that same fib. 5/ Packets generated in response to some other packet (e.g. reset or icmp packets). These should use the FIB associated with the packet being reponded to. 6/ Packets generated during encapsulation. gif, tun and other tunnel interfaces will encapsulate using the FIB that was in effect withthe proces that set up the tunnel. thus setfib 1 ifconfig gif0 [tunnel instructions] will set the fib for the tunnel to use to be fib 1. Routing messages would be associated with their process, and thus select one FIB or another. messages from the kernel would be associated with the fib they refer to and would only be received by a routing socket associated with that fib. (not yet implemented) In addition Netstat has been edited to be able to cope with the fact that the array is now 2 dimensional. (It looks in system memory using libkvm (!)). Old versions of netstat see only the first FIB. In addition two sysctls are added to give: a) the number of FIBs compiled in (active) b) the default FIB of the calling process. Early testing experience: ------------------------- Basically our (IronPort's) appliance does this functionality already using ipfw fwd but that method has some drawbacks. For example, It can't fully simulate a routing table because it can't influence the socket's choice of local address when a connect() is done. Testing during the generating of these changes has been remarkably smooth so far. Multiple tables have co-existed with no notable side effects, and packets have been routes accordingly. ipfw has grown 2 new keywords: setfib N ip from anay to any count ip from any to any fib N In pf there seems to be a requirement to be able to give symbolic names to the fibs but I do not have that capacity. I am not sure if it is required. SCTP has interestingly enough built in support for this, called VRFs in Cisco parlance. it will be interesting to see how that handles it when it suddenly actually does something. Where to next: -------------------- After committing the ABI compatible version and MFCing it, I'd like to proceed in a forward direction in -current. this will result in some roto-tilling in the routing code. Firstly: the current code's idea of having a separate tree per protocol family, all of the same format, and pointed to by the 1 dimensional array is a bit silly. Especially when one considers that there is code that makes assumptions about every protocol having the same internal structures there. Some protocols don't WANT that sort of structure. (for example the whole idea of a netmask is foreign to appletalk). This needs to be made opaque to the external code. My suggested first change is to add routing method pointers to the 'domain' structure, along with information pointing the data. instead of having an array of pointers to uniform structures, there would be an array pointing to the 'domain' structures for each protocol address domain (protocol family), and the methods this reached would be called. The methods would have an argument that gives FIB number, but the protocol would be free to ignore it. When the ABI can be changed it raises the possibilty of the addition of a fib entry into the "struct route". Currently, the structure contains the sockaddr of the desination, and the resulting fib entry. To make this work fully, one could add a fib number so that given an address and a fib, one can find the third element, the fib entry. Interaction with the ARP layer/ LL layer would need to be revisited as well. Qing Li has been working on this already. This work was sponsored by Ironport Systems/Cisco Reviewed by: several including rwatson, bz and mlair (parts each) Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
/* XXX MRT not always INET */ /* stick with table 0 though */
if (af == AF_INET)
This main goals of this project are: 1. separating L2 tables (ARP, NDP) from the L3 routing tables 2. removing as much locking dependencies among these layers as possible to allow for some parallelism in the search operations 3. simplify the logic in the routing code, The most notable end result is the obsolescent of the route cloning (RTF_CLONING) concept, which translated into code reduction in both IPv4 ARP and IPv6 NDP related modules, and size reduction in struct rtentry{}. The change in design obsoletes the semantics of RTF_CLONING, RTF_WASCLONE and RTF_LLINFO routing flags. The userland applications such as "arp" and "ndp" have been modified to reflect those changes. The output from "netstat -r" shows only the routing entries. Quite a few developers have contributed to this project in the past: Glebius Smirnoff, Luigi Rizzo, Alessandro Cerri, and Andre Oppermann. And most recently: - Kip Macy revised the locking code completely, thus completing the last piece of the puzzle, Kip has also been conducting active functional testing - Sam Leffler has helped me improving/refactoring the code, and provided valuable reviews - Julian Elischer setup the perforce tree for me and has helped me maintaining that branch before the svn conversion
2008-12-15 06:10:57 +00:00
in_rtalloc_ign((struct route *)&ro, 0, 0);
Add code to allow the system to handle multiple routing tables. This particular implementation is designed to be fully backwards compatible and to be MFC-able to 7.x (and 6.x) Currently the only protocol that can make use of the multiple tables is IPv4 Similar functionality exists in OpenBSD and Linux. From my notes: ----- One thing where FreeBSD has been falling behind, and which by chance I have some time to work on is "policy based routing", which allows different packet streams to be routed by more than just the destination address. Constraints: ------------ I want to make some form of this available in the 6.x tree (and by extension 7.x) , but FreeBSD in general needs it so I might as well do it in -current and back port the portions I need. One of the ways that this can be done is to have the ability to instantiate multiple kernel routing tables (which I will now refer to as "Forwarding Information Bases" or "FIBs" for political correctness reasons). Which FIB a particular packet uses to make the next hop decision can be decided by a number of mechanisms. The policies these mechanisms implement are the "Policies" referred to in "Policy based routing". One of the constraints I have if I try to back port this work to 6.x is that it must be implemented as a EXTENSION to the existing ABIs in 6.x so that third party applications do not need to be recompiled in timespan of the branch. This first version will not have some of the bells and whistles that will come with later versions. It will, for example, be limited to 16 tables in the first commit. Implementation method, Compatible version. (part 1) ------------------------------- For this reason I have implemented a "sufficient subset" of a multiple routing table solution in Perforce, and back-ported it to 6.x. (also in Perforce though not always caught up with what I have done in -current/P4). The subset allows a number of FIBs to be defined at compile time (8 is sufficient for my purposes in 6.x) and implements the changes needed to allow IPV4 to use them. I have not done the changes for ipv6 simply because I do not need it, and I do not have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it. Other protocol families are left untouched and should there be users with proprietary protocol families, they should continue to work and be oblivious to the existence of the extra FIBs. To understand how this is done, one must know that the current FIB code starts everything off with a single dimensional array of pointers to FIB head structures (One per protocol family), each of which in turn points to the trie of routes available to that family. The basic change in the ABI compatible version of the change is to extent that array to be a 2 dimensional array, so that instead of protocol family X looking at rt_tables[X] for the table it needs, it looks at rt_tables[Y][X] when for all protocol families except ipv4 Y is always 0. Code that is unaware of the change always just sees the first row of the table, which of course looks just like the one dimensional array that existed before. The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign() are all maintained, but refer only to the first row of the array, so that existing callers in proprietary protocols can continue to do the "right thing". Some new entry points are added, for the exclusive use of ipv4 code called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(), which have an extra argument which refers the code to the correct row. In addition, there are some new entry points (currently called rtalloc_fib() and friends) that check the Address family being looked up and call either rtalloc() (and friends) if the protocol is not IPv4 forcing the action to row 0 or to the appropriate row if it IS IPv4 (and that info is available). These are for calling from code that is not specific to any particular protocol. The way these are implemented would change in the non ABI preserving code to be added later. One feature of the first version of the code is that for ipv4, the interface routes show up automatically on all the FIBs, so that no matter what FIB you select you always have the basic direct attached hosts available to you. (rtinit() does this automatically). You CAN delete an interface route from one FIB should you want to but by default it's there. ARP information is also available in each FIB. It's assumed that the same machine would have the same MAC address, regardless of which FIB you are using to get to it. This brings us as to how the correct FIB is selected for an outgoing IPV4 packet. Firstly, all packets have a FIB associated with them. if nothing has been done to change it, it will be FIB 0. The FIB is changed in the following ways. Packets fall into one of a number of classes. 1/ locally generated packets, coming from a socket/PCB. Such packets select a FIB from a number associated with the socket/PCB. This in turn is inherited from the process, but can be changed by a socket option. The process in turn inherits it on fork. I have written a utility call setfib that acts a bit like nice.. setfib -3 ping target.example.com # will use fib 3 for ping. It is an obvious extension to make it a property of a jail but I have not done so. It can be achieved by combining the setfib and jail commands. 2/ packets received on an interface for forwarding. By default these packets would use table 0, (or possibly a number settable in a sysctl(not yet)). but prior to routing the firewall can inspect them (see below). (possibly in the future you may be able to associate a FIB with packets received on an interface.. An ifconfig arg, but not yet.) 3/ packets inspected by a packet classifier, which can arbitrarily associate a fib with it on a packet by packet basis. A fib assigned to a packet by a packet classifier (such as ipfw) would over-ride a fib associated by a more default source. (such as cases 1 or 2). 4/ a tcp listen socket associated with a fib will generate accept sockets that are associated with that same fib. 5/ Packets generated in response to some other packet (e.g. reset or icmp packets). These should use the FIB associated with the packet being reponded to. 6/ Packets generated during encapsulation. gif, tun and other tunnel interfaces will encapsulate using the FIB that was in effect withthe proces that set up the tunnel. thus setfib 1 ifconfig gif0 [tunnel instructions] will set the fib for the tunnel to use to be fib 1. Routing messages would be associated with their process, and thus select one FIB or another. messages from the kernel would be associated with the fib they refer to and would only be received by a routing socket associated with that fib. (not yet implemented) In addition Netstat has been edited to be able to cope with the fact that the array is now 2 dimensional. (It looks in system memory using libkvm (!)). Old versions of netstat see only the first FIB. In addition two sysctls are added to give: a) the number of FIBs compiled in (active) b) the default FIB of the calling process. Early testing experience: ------------------------- Basically our (IronPort's) appliance does this functionality already using ipfw fwd but that method has some drawbacks. For example, It can't fully simulate a routing table because it can't influence the socket's choice of local address when a connect() is done. Testing during the generating of these changes has been remarkably smooth so far. Multiple tables have co-existed with no notable side effects, and packets have been routes accordingly. ipfw has grown 2 new keywords: setfib N ip from anay to any count ip from any to any fib N In pf there seems to be a requirement to be able to give symbolic names to the fibs but I do not have that capacity. I am not sure if it is required. SCTP has interestingly enough built in support for this, called VRFs in Cisco parlance. it will be interesting to see how that handles it when it suddenly actually does something. Where to next: -------------------- After committing the ABI compatible version and MFCing it, I'd like to proceed in a forward direction in -current. this will result in some roto-tilling in the routing code. Firstly: the current code's idea of having a separate tree per protocol family, all of the same format, and pointed to by the 1 dimensional array is a bit silly. Especially when one considers that there is code that makes assumptions about every protocol having the same internal structures there. Some protocols don't WANT that sort of structure. (for example the whole idea of a netmask is foreign to appletalk). This needs to be made opaque to the external code. My suggested first change is to add routing method pointers to the 'domain' structure, along with information pointing the data. instead of having an array of pointers to uniform structures, there would be an array pointing to the 'domain' structures for each protocol address domain (protocol family), and the methods this reached would be called. The methods would have an argument that gives FIB number, but the protocol would be free to ignore it. When the ABI can be changed it raises the possibilty of the addition of a fib entry into the "struct route". Currently, the structure contains the sockaddr of the desination, and the resulting fib entry. To make this work fully, one could add a fib number so that given an address and a fib, one can find the third element, the fib entry. Interaction with the ARP layer/ LL layer would need to be revisited as well. Qing Li has been working on this already. This work was sponsored by Ironport Systems/Cisco Reviewed by: several including rwatson, bz and mlair (parts each) Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
else
This main goals of this project are: 1. separating L2 tables (ARP, NDP) from the L3 routing tables 2. removing as much locking dependencies among these layers as possible to allow for some parallelism in the search operations 3. simplify the logic in the routing code, The most notable end result is the obsolescent of the route cloning (RTF_CLONING) concept, which translated into code reduction in both IPv4 ARP and IPv6 NDP related modules, and size reduction in struct rtentry{}. The change in design obsoletes the semantics of RTF_CLONING, RTF_WASCLONE and RTF_LLINFO routing flags. The userland applications such as "arp" and "ndp" have been modified to reflect those changes. The output from "netstat -r" shows only the routing entries. Quite a few developers have contributed to this project in the past: Glebius Smirnoff, Luigi Rizzo, Alessandro Cerri, and Andre Oppermann. And most recently: - Kip Macy revised the locking code completely, thus completing the last piece of the puzzle, Kip has also been conducting active functional testing - Sam Leffler has helped me improving/refactoring the code, and provided valuable reviews - Julian Elischer setup the perforce tree for me and has helped me maintaining that branch before the svn conversion
2008-12-15 06:10:57 +00:00
rtalloc_ign((struct route *)&ro, 0);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#else /* ! __FreeBSD__ */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
rtalloc_noclone((struct route *)&ro, NO_CLONING);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (ro.ro_rt != NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
/* No interface given, this is a no-route check */
if (kif == NULL)
goto out;
if (kif->pfik_ifp == NULL) {
ret = 0;
goto out;
}
/* Perform uRPF check if passed input interface */
ret = 0;
rn = (struct radix_node *)ro.ro_rt;
do {
rt = (struct rtentry *)rn;
#ifndef __FreeBSD__ /* CARPDEV */
if (rt->rt_ifp->if_type == IFT_CARP)
ifp = rt->rt_ifp->if_carpdev;
else
#endif
ifp = rt->rt_ifp;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (kif->pfik_ifp == ifp)
ret = 1;
#ifdef __FreeBSD__ /* MULTIPATH_ROUTING */
rn = NULL;
#else
rn = rn_mpath_next(rn);
#endif
} while (check_mpath == 1 && rn != NULL && ret == 0);
} else
ret = 0;
out:
if (ro.ro_rt != NULL)
RTFREE(ro.ro_rt);
return (ret);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
}
int
pf_rtlabel_match(struct pf_addr *addr, sa_family_t af, struct pf_addr_wrap *aw)
{
struct sockaddr_in *dst;
#ifdef INET6
struct sockaddr_in6 *dst6;
struct route_in6 ro;
#else
struct route ro;
#endif
int ret = 0;
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;
break;
#ifdef 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;
break;
#endif /* INET6 */
default:
return (0);
}
#ifdef __FreeBSD__
# ifdef RTF_PRCLONING
rtalloc_ign((struct route *)&ro, (RTF_CLONING|RTF_PRCLONING));
# else /* !RTF_PRCLONING */
Add code to allow the system to handle multiple routing tables. This particular implementation is designed to be fully backwards compatible and to be MFC-able to 7.x (and 6.x) Currently the only protocol that can make use of the multiple tables is IPv4 Similar functionality exists in OpenBSD and Linux. From my notes: ----- One thing where FreeBSD has been falling behind, and which by chance I have some time to work on is "policy based routing", which allows different packet streams to be routed by more than just the destination address. Constraints: ------------ I want to make some form of this available in the 6.x tree (and by extension 7.x) , but FreeBSD in general needs it so I might as well do it in -current and back port the portions I need. One of the ways that this can be done is to have the ability to instantiate multiple kernel routing tables (which I will now refer to as "Forwarding Information Bases" or "FIBs" for political correctness reasons). Which FIB a particular packet uses to make the next hop decision can be decided by a number of mechanisms. The policies these mechanisms implement are the "Policies" referred to in "Policy based routing". One of the constraints I have if I try to back port this work to 6.x is that it must be implemented as a EXTENSION to the existing ABIs in 6.x so that third party applications do not need to be recompiled in timespan of the branch. This first version will not have some of the bells and whistles that will come with later versions. It will, for example, be limited to 16 tables in the first commit. Implementation method, Compatible version. (part 1) ------------------------------- For this reason I have implemented a "sufficient subset" of a multiple routing table solution in Perforce, and back-ported it to 6.x. (also in Perforce though not always caught up with what I have done in -current/P4). The subset allows a number of FIBs to be defined at compile time (8 is sufficient for my purposes in 6.x) and implements the changes needed to allow IPV4 to use them. I have not done the changes for ipv6 simply because I do not need it, and I do not have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it. Other protocol families are left untouched and should there be users with proprietary protocol families, they should continue to work and be oblivious to the existence of the extra FIBs. To understand how this is done, one must know that the current FIB code starts everything off with a single dimensional array of pointers to FIB head structures (One per protocol family), each of which in turn points to the trie of routes available to that family. The basic change in the ABI compatible version of the change is to extent that array to be a 2 dimensional array, so that instead of protocol family X looking at rt_tables[X] for the table it needs, it looks at rt_tables[Y][X] when for all protocol families except ipv4 Y is always 0. Code that is unaware of the change always just sees the first row of the table, which of course looks just like the one dimensional array that existed before. The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign() are all maintained, but refer only to the first row of the array, so that existing callers in proprietary protocols can continue to do the "right thing". Some new entry points are added, for the exclusive use of ipv4 code called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(), which have an extra argument which refers the code to the correct row. In addition, there are some new entry points (currently called rtalloc_fib() and friends) that check the Address family being looked up and call either rtalloc() (and friends) if the protocol is not IPv4 forcing the action to row 0 or to the appropriate row if it IS IPv4 (and that info is available). These are for calling from code that is not specific to any particular protocol. The way these are implemented would change in the non ABI preserving code to be added later. One feature of the first version of the code is that for ipv4, the interface routes show up automatically on all the FIBs, so that no matter what FIB you select you always have the basic direct attached hosts available to you. (rtinit() does this automatically). You CAN delete an interface route from one FIB should you want to but by default it's there. ARP information is also available in each FIB. It's assumed that the same machine would have the same MAC address, regardless of which FIB you are using to get to it. This brings us as to how the correct FIB is selected for an outgoing IPV4 packet. Firstly, all packets have a FIB associated with them. if nothing has been done to change it, it will be FIB 0. The FIB is changed in the following ways. Packets fall into one of a number of classes. 1/ locally generated packets, coming from a socket/PCB. Such packets select a FIB from a number associated with the socket/PCB. This in turn is inherited from the process, but can be changed by a socket option. The process in turn inherits it on fork. I have written a utility call setfib that acts a bit like nice.. setfib -3 ping target.example.com # will use fib 3 for ping. It is an obvious extension to make it a property of a jail but I have not done so. It can be achieved by combining the setfib and jail commands. 2/ packets received on an interface for forwarding. By default these packets would use table 0, (or possibly a number settable in a sysctl(not yet)). but prior to routing the firewall can inspect them (see below). (possibly in the future you may be able to associate a FIB with packets received on an interface.. An ifconfig arg, but not yet.) 3/ packets inspected by a packet classifier, which can arbitrarily associate a fib with it on a packet by packet basis. A fib assigned to a packet by a packet classifier (such as ipfw) would over-ride a fib associated by a more default source. (such as cases 1 or 2). 4/ a tcp listen socket associated with a fib will generate accept sockets that are associated with that same fib. 5/ Packets generated in response to some other packet (e.g. reset or icmp packets). These should use the FIB associated with the packet being reponded to. 6/ Packets generated during encapsulation. gif, tun and other tunnel interfaces will encapsulate using the FIB that was in effect withthe proces that set up the tunnel. thus setfib 1 ifconfig gif0 [tunnel instructions] will set the fib for the tunnel to use to be fib 1. Routing messages would be associated with their process, and thus select one FIB or another. messages from the kernel would be associated with the fib they refer to and would only be received by a routing socket associated with that fib. (not yet implemented) In addition Netstat has been edited to be able to cope with the fact that the array is now 2 dimensional. (It looks in system memory using libkvm (!)). Old versions of netstat see only the first FIB. In addition two sysctls are added to give: a) the number of FIBs compiled in (active) b) the default FIB of the calling process. Early testing experience: ------------------------- Basically our (IronPort's) appliance does this functionality already using ipfw fwd but that method has some drawbacks. For example, It can't fully simulate a routing table because it can't influence the socket's choice of local address when a connect() is done. Testing during the generating of these changes has been remarkably smooth so far. Multiple tables have co-existed with no notable side effects, and packets have been routes accordingly. ipfw has grown 2 new keywords: setfib N ip from anay to any count ip from any to any fib N In pf there seems to be a requirement to be able to give symbolic names to the fibs but I do not have that capacity. I am not sure if it is required. SCTP has interestingly enough built in support for this, called VRFs in Cisco parlance. it will be interesting to see how that handles it when it suddenly actually does something. Where to next: -------------------- After committing the ABI compatible version and MFCing it, I'd like to proceed in a forward direction in -current. this will result in some roto-tilling in the routing code. Firstly: the current code's idea of having a separate tree per protocol family, all of the same format, and pointed to by the 1 dimensional array is a bit silly. Especially when one considers that there is code that makes assumptions about every protocol having the same internal structures there. Some protocols don't WANT that sort of structure. (for example the whole idea of a netmask is foreign to appletalk). This needs to be made opaque to the external code. My suggested first change is to add routing method pointers to the 'domain' structure, along with information pointing the data. instead of having an array of pointers to uniform structures, there would be an array pointing to the 'domain' structures for each protocol address domain (protocol family), and the methods this reached would be called. The methods would have an argument that gives FIB number, but the protocol would be free to ignore it. When the ABI can be changed it raises the possibilty of the addition of a fib entry into the "struct route". Currently, the structure contains the sockaddr of the desination, and the resulting fib entry. To make this work fully, one could add a fib number so that given an address and a fib, one can find the third element, the fib entry. Interaction with the ARP layer/ LL layer would need to be revisited as well. Qing Li has been working on this already. This work was sponsored by Ironport Systems/Cisco Reviewed by: several including rwatson, bz and mlair (parts each) Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
if (af == AF_INET)
This main goals of this project are: 1. separating L2 tables (ARP, NDP) from the L3 routing tables 2. removing as much locking dependencies among these layers as possible to allow for some parallelism in the search operations 3. simplify the logic in the routing code, The most notable end result is the obsolescent of the route cloning (RTF_CLONING) concept, which translated into code reduction in both IPv4 ARP and IPv6 NDP related modules, and size reduction in struct rtentry{}. The change in design obsoletes the semantics of RTF_CLONING, RTF_WASCLONE and RTF_LLINFO routing flags. The userland applications such as "arp" and "ndp" have been modified to reflect those changes. The output from "netstat -r" shows only the routing entries. Quite a few developers have contributed to this project in the past: Glebius Smirnoff, Luigi Rizzo, Alessandro Cerri, and Andre Oppermann. And most recently: - Kip Macy revised the locking code completely, thus completing the last piece of the puzzle, Kip has also been conducting active functional testing - Sam Leffler has helped me improving/refactoring the code, and provided valuable reviews - Julian Elischer setup the perforce tree for me and has helped me maintaining that branch before the svn conversion
2008-12-15 06:10:57 +00:00
in_rtalloc_ign((struct route *)&ro, 0, 0);
Add code to allow the system to handle multiple routing tables. This particular implementation is designed to be fully backwards compatible and to be MFC-able to 7.x (and 6.x) Currently the only protocol that can make use of the multiple tables is IPv4 Similar functionality exists in OpenBSD and Linux. From my notes: ----- One thing where FreeBSD has been falling behind, and which by chance I have some time to work on is "policy based routing", which allows different packet streams to be routed by more than just the destination address. Constraints: ------------ I want to make some form of this available in the 6.x tree (and by extension 7.x) , but FreeBSD in general needs it so I might as well do it in -current and back port the portions I need. One of the ways that this can be done is to have the ability to instantiate multiple kernel routing tables (which I will now refer to as "Forwarding Information Bases" or "FIBs" for political correctness reasons). Which FIB a particular packet uses to make the next hop decision can be decided by a number of mechanisms. The policies these mechanisms implement are the "Policies" referred to in "Policy based routing". One of the constraints I have if I try to back port this work to 6.x is that it must be implemented as a EXTENSION to the existing ABIs in 6.x so that third party applications do not need to be recompiled in timespan of the branch. This first version will not have some of the bells and whistles that will come with later versions. It will, for example, be limited to 16 tables in the first commit. Implementation method, Compatible version. (part 1) ------------------------------- For this reason I have implemented a "sufficient subset" of a multiple routing table solution in Perforce, and back-ported it to 6.x. (also in Perforce though not always caught up with what I have done in -current/P4). The subset allows a number of FIBs to be defined at compile time (8 is sufficient for my purposes in 6.x) and implements the changes needed to allow IPV4 to use them. I have not done the changes for ipv6 simply because I do not need it, and I do not have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it. Other protocol families are left untouched and should there be users with proprietary protocol families, they should continue to work and be oblivious to the existence of the extra FIBs. To understand how this is done, one must know that the current FIB code starts everything off with a single dimensional array of pointers to FIB head structures (One per protocol family), each of which in turn points to the trie of routes available to that family. The basic change in the ABI compatible version of the change is to extent that array to be a 2 dimensional array, so that instead of protocol family X looking at rt_tables[X] for the table it needs, it looks at rt_tables[Y][X] when for all protocol families except ipv4 Y is always 0. Code that is unaware of the change always just sees the first row of the table, which of course looks just like the one dimensional array that existed before. The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign() are all maintained, but refer only to the first row of the array, so that existing callers in proprietary protocols can continue to do the "right thing". Some new entry points are added, for the exclusive use of ipv4 code called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(), which have an extra argument which refers the code to the correct row. In addition, there are some new entry points (currently called rtalloc_fib() and friends) that check the Address family being looked up and call either rtalloc() (and friends) if the protocol is not IPv4 forcing the action to row 0 or to the appropriate row if it IS IPv4 (and that info is available). These are for calling from code that is not specific to any particular protocol. The way these are implemented would change in the non ABI preserving code to be added later. One feature of the first version of the code is that for ipv4, the interface routes show up automatically on all the FIBs, so that no matter what FIB you select you always have the basic direct attached hosts available to you. (rtinit() does this automatically). You CAN delete an interface route from one FIB should you want to but by default it's there. ARP information is also available in each FIB. It's assumed that the same machine would have the same MAC address, regardless of which FIB you are using to get to it. This brings us as to how the correct FIB is selected for an outgoing IPV4 packet. Firstly, all packets have a FIB associated with them. if nothing has been done to change it, it will be FIB 0. The FIB is changed in the following ways. Packets fall into one of a number of classes. 1/ locally generated packets, coming from a socket/PCB. Such packets select a FIB from a number associated with the socket/PCB. This in turn is inherited from the process, but can be changed by a socket option. The process in turn inherits it on fork. I have written a utility call setfib that acts a bit like nice.. setfib -3 ping target.example.com # will use fib 3 for ping. It is an obvious extension to make it a property of a jail but I have not done so. It can be achieved by combining the setfib and jail commands. 2/ packets received on an interface for forwarding. By default these packets would use table 0, (or possibly a number settable in a sysctl(not yet)). but prior to routing the firewall can inspect them (see below). (possibly in the future you may be able to associate a FIB with packets received on an interface.. An ifconfig arg, but not yet.) 3/ packets inspected by a packet classifier, which can arbitrarily associate a fib with it on a packet by packet basis. A fib assigned to a packet by a packet classifier (such as ipfw) would over-ride a fib associated by a more default source. (such as cases 1 or 2). 4/ a tcp listen socket associated with a fib will generate accept sockets that are associated with that same fib. 5/ Packets generated in response to some other packet (e.g. reset or icmp packets). These should use the FIB associated with the packet being reponded to. 6/ Packets generated during encapsulation. gif, tun and other tunnel interfaces will encapsulate using the FIB that was in effect withthe proces that set up the tunnel. thus setfib 1 ifconfig gif0 [tunnel instructions] will set the fib for the tunnel to use to be fib 1. Routing messages would be associated with their process, and thus select one FIB or another. messages from the kernel would be associated with the fib they refer to and would only be received by a routing socket associated with that fib. (not yet implemented) In addition Netstat has been edited to be able to cope with the fact that the array is now 2 dimensional. (It looks in system memory using libkvm (!)). Old versions of netstat see only the first FIB. In addition two sysctls are added to give: a) the number of FIBs compiled in (active) b) the default FIB of the calling process. Early testing experience: ------------------------- Basically our (IronPort's) appliance does this functionality already using ipfw fwd but that method has some drawbacks. For example, It can't fully simulate a routing table because it can't influence the socket's choice of local address when a connect() is done. Testing during the generating of these changes has been remarkably smooth so far. Multiple tables have co-existed with no notable side effects, and packets have been routes accordingly. ipfw has grown 2 new keywords: setfib N ip from anay to any count ip from any to any fib N In pf there seems to be a requirement to be able to give symbolic names to the fibs but I do not have that capacity. I am not sure if it is required. SCTP has interestingly enough built in support for this, called VRFs in Cisco parlance. it will be interesting to see how that handles it when it suddenly actually does something. Where to next: -------------------- After committing the ABI compatible version and MFCing it, I'd like to proceed in a forward direction in -current. this will result in some roto-tilling in the routing code. Firstly: the current code's idea of having a separate tree per protocol family, all of the same format, and pointed to by the 1 dimensional array is a bit silly. Especially when one considers that there is code that makes assumptions about every protocol having the same internal structures there. Some protocols don't WANT that sort of structure. (for example the whole idea of a netmask is foreign to appletalk). This needs to be made opaque to the external code. My suggested first change is to add routing method pointers to the 'domain' structure, along with information pointing the data. instead of having an array of pointers to uniform structures, there would be an array pointing to the 'domain' structures for each protocol address domain (protocol family), and the methods this reached would be called. The methods would have an argument that gives FIB number, but the protocol would be free to ignore it. When the ABI can be changed it raises the possibilty of the addition of a fib entry into the "struct route". Currently, the structure contains the sockaddr of the desination, and the resulting fib entry. To make this work fully, one could add a fib number so that given an address and a fib, one can find the third element, the fib entry. Interaction with the ARP layer/ LL layer would need to be revisited as well. Qing Li has been working on this already. This work was sponsored by Ironport Systems/Cisco Reviewed by: several including rwatson, bz and mlair (parts each) Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
else
This main goals of this project are: 1. separating L2 tables (ARP, NDP) from the L3 routing tables 2. removing as much locking dependencies among these layers as possible to allow for some parallelism in the search operations 3. simplify the logic in the routing code, The most notable end result is the obsolescent of the route cloning (RTF_CLONING) concept, which translated into code reduction in both IPv4 ARP and IPv6 NDP related modules, and size reduction in struct rtentry{}. The change in design obsoletes the semantics of RTF_CLONING, RTF_WASCLONE and RTF_LLINFO routing flags. The userland applications such as "arp" and "ndp" have been modified to reflect those changes. The output from "netstat -r" shows only the routing entries. Quite a few developers have contributed to this project in the past: Glebius Smirnoff, Luigi Rizzo, Alessandro Cerri, and Andre Oppermann. And most recently: - Kip Macy revised the locking code completely, thus completing the last piece of the puzzle, Kip has also been conducting active functional testing - Sam Leffler has helped me improving/refactoring the code, and provided valuable reviews - Julian Elischer setup the perforce tree for me and has helped me maintaining that branch before the svn conversion
2008-12-15 06:10:57 +00:00
rtalloc_ign((struct route *)&ro, 0);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
# endif
#else /* ! __FreeBSD__ */
rtalloc_noclone((struct route *)&ro, NO_CLONING);
#endif
if (ro.ro_rt != NULL) {
#ifdef __FreeBSD__
/* XXX_IMPORT: later */
#else
if (ro.ro_rt->rt_labelid == aw->v.rtlabel)
ret = 1;
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
RTFREE(ro.ro_rt);
}
return (ret);
}
#ifdef INET
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
void
pf_route(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_state *s, struct pf_pdesc *pd)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
struct mbuf *m0, *m1;
struct route iproute;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct route *ro = NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
struct sockaddr_in *dst;
struct ip *ip;
struct ifnet *ifp = NULL;
struct pf_addr naddr;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_src_node *sn = NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
int error = 0;
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
int sw_csum;
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifdef IPSEC
struct m_tag *mtag;
#endif /* IPSEC */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (m == NULL || *m == NULL || r == NULL ||
(dir != PF_IN && dir != PF_OUT) || oifp == NULL)
panic("pf_route: invalid parameters");
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pd->pf_mtag->routed++ > 3) {
m0 = *m;
*m = NULL;
goto bad;
Merge in a stable fix from OpenBSD: MFC: Fix by dhartmei@ change pf_route() loop detection: introduce a counter (number of times a packet is routed already) in the mbuf tag, allow at most four times. Fixes some legitimate cases broken by the previous change. Reviewed by: dhartmei
2004-07-17 17:15:15 +00:00
}
if (r->rt == PF_DUPTO) {
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Merge in a stable fix from OpenBSD: MFC: Fix by dhartmei@ change pf_route() loop detection: introduce a counter (number of times a packet is routed already) in the mbuf tag, allow at most four times. Fixes some legitimate cases broken by the previous change. Reviewed by: dhartmei
2004-07-17 17:15:15 +00:00
if ((m0 = m_dup(*m, M_DONTWAIT)) == NULL)
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#else
Merge in a stable fix from OpenBSD: MFC: Fix by dhartmei@ change pf_route() loop detection: introduce a counter (number of times a packet is routed already) in the mbuf tag, allow at most four times. Fixes some legitimate cases broken by the previous change. Reviewed by: dhartmei
2004-07-17 17:15:15 +00:00
if ((m0 = m_copym2(*m, 0, M_COPYALL, M_NOWAIT)) == NULL)
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return;
} else {
if ((r->rt == PF_REPLYTO) == (r->direction == dir))
return;
m0 = *m;
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (m0->m_len < sizeof(struct ip)) {
DPFPRINTF(PF_DEBUG_URGENT,
("pf_route: m0->m_len < sizeof(struct ip)\n"));
goto bad;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
ip = mtod(m0, struct ip *);
ro = &iproute;
bzero((caddr_t)ro, sizeof(*ro));
dst = satosin(&ro->ro_dst);
dst->sin_family = AF_INET;
dst->sin_len = sizeof(*dst);
dst->sin_addr = ip->ip_dst;
if (r->rt == PF_FASTROUTE) {
Add code to allow the system to handle multiple routing tables. This particular implementation is designed to be fully backwards compatible and to be MFC-able to 7.x (and 6.x) Currently the only protocol that can make use of the multiple tables is IPv4 Similar functionality exists in OpenBSD and Linux. From my notes: ----- One thing where FreeBSD has been falling behind, and which by chance I have some time to work on is "policy based routing", which allows different packet streams to be routed by more than just the destination address. Constraints: ------------ I want to make some form of this available in the 6.x tree (and by extension 7.x) , but FreeBSD in general needs it so I might as well do it in -current and back port the portions I need. One of the ways that this can be done is to have the ability to instantiate multiple kernel routing tables (which I will now refer to as "Forwarding Information Bases" or "FIBs" for political correctness reasons). Which FIB a particular packet uses to make the next hop decision can be decided by a number of mechanisms. The policies these mechanisms implement are the "Policies" referred to in "Policy based routing". One of the constraints I have if I try to back port this work to 6.x is that it must be implemented as a EXTENSION to the existing ABIs in 6.x so that third party applications do not need to be recompiled in timespan of the branch. This first version will not have some of the bells and whistles that will come with later versions. It will, for example, be limited to 16 tables in the first commit. Implementation method, Compatible version. (part 1) ------------------------------- For this reason I have implemented a "sufficient subset" of a multiple routing table solution in Perforce, and back-ported it to 6.x. (also in Perforce though not always caught up with what I have done in -current/P4). The subset allows a number of FIBs to be defined at compile time (8 is sufficient for my purposes in 6.x) and implements the changes needed to allow IPV4 to use them. I have not done the changes for ipv6 simply because I do not need it, and I do not have enough knowledge of ipv6 (e.g. neighbor discovery) needed to do it. Other protocol families are left untouched and should there be users with proprietary protocol families, they should continue to work and be oblivious to the existence of the extra FIBs. To understand how this is done, one must know that the current FIB code starts everything off with a single dimensional array of pointers to FIB head structures (One per protocol family), each of which in turn points to the trie of routes available to that family. The basic change in the ABI compatible version of the change is to extent that array to be a 2 dimensional array, so that instead of protocol family X looking at rt_tables[X] for the table it needs, it looks at rt_tables[Y][X] when for all protocol families except ipv4 Y is always 0. Code that is unaware of the change always just sees the first row of the table, which of course looks just like the one dimensional array that existed before. The entry points rtrequest(), rtalloc(), rtalloc1(), rtalloc_ign() are all maintained, but refer only to the first row of the array, so that existing callers in proprietary protocols can continue to do the "right thing". Some new entry points are added, for the exclusive use of ipv4 code called in_rtrequest(), in_rtalloc(), in_rtalloc1() and in_rtalloc_ign(), which have an extra argument which refers the code to the correct row. In addition, there are some new entry points (currently called rtalloc_fib() and friends) that check the Address family being looked up and call either rtalloc() (and friends) if the protocol is not IPv4 forcing the action to row 0 or to the appropriate row if it IS IPv4 (and that info is available). These are for calling from code that is not specific to any particular protocol. The way these are implemented would change in the non ABI preserving code to be added later. One feature of the first version of the code is that for ipv4, the interface routes show up automatically on all the FIBs, so that no matter what FIB you select you always have the basic direct attached hosts available to you. (rtinit() does this automatically). You CAN delete an interface route from one FIB should you want to but by default it's there. ARP information is also available in each FIB. It's assumed that the same machine would have the same MAC address, regardless of which FIB you are using to get to it. This brings us as to how the correct FIB is selected for an outgoing IPV4 packet. Firstly, all packets have a FIB associated with them. if nothing has been done to change it, it will be FIB 0. The FIB is changed in the following ways. Packets fall into one of a number of classes. 1/ locally generated packets, coming from a socket/PCB. Such packets select a FIB from a number associated with the socket/PCB. This in turn is inherited from the process, but can be changed by a socket option. The process in turn inherits it on fork. I have written a utility call setfib that acts a bit like nice.. setfib -3 ping target.example.com # will use fib 3 for ping. It is an obvious extension to make it a property of a jail but I have not done so. It can be achieved by combining the setfib and jail commands. 2/ packets received on an interface for forwarding. By default these packets would use table 0, (or possibly a number settable in a sysctl(not yet)). but prior to routing the firewall can inspect them (see below). (possibly in the future you may be able to associate a FIB with packets received on an interface.. An ifconfig arg, but not yet.) 3/ packets inspected by a packet classifier, which can arbitrarily associate a fib with it on a packet by packet basis. A fib assigned to a packet by a packet classifier (such as ipfw) would over-ride a fib associated by a more default source. (such as cases 1 or 2). 4/ a tcp listen socket associated with a fib will generate accept sockets that are associated with that same fib. 5/ Packets generated in response to some other packet (e.g. reset or icmp packets). These should use the FIB associated with the packet being reponded to. 6/ Packets generated during encapsulation. gif, tun and other tunnel interfaces will encapsulate using the FIB that was in effect withthe proces that set up the tunnel. thus setfib 1 ifconfig gif0 [tunnel instructions] will set the fib for the tunnel to use to be fib 1. Routing messages would be associated with their process, and thus select one FIB or another. messages from the kernel would be associated with the fib they refer to and would only be received by a routing socket associated with that fib. (not yet implemented) In addition Netstat has been edited to be able to cope with the fact that the array is now 2 dimensional. (It looks in system memory using libkvm (!)). Old versions of netstat see only the first FIB. In addition two sysctls are added to give: a) the number of FIBs compiled in (active) b) the default FIB of the calling process. Early testing experience: ------------------------- Basically our (IronPort's) appliance does this functionality already using ipfw fwd but that method has some drawbacks. For example, It can't fully simulate a routing table because it can't influence the socket's choice of local address when a connect() is done. Testing during the generating of these changes has been remarkably smooth so far. Multiple tables have co-existed with no notable side effects, and packets have been routes accordingly. ipfw has grown 2 new keywords: setfib N ip from anay to any count ip from any to any fib N In pf there seems to be a requirement to be able to give symbolic names to the fibs but I do not have that capacity. I am not sure if it is required. SCTP has interestingly enough built in support for this, called VRFs in Cisco parlance. it will be interesting to see how that handles it when it suddenly actually does something. Where to next: -------------------- After committing the ABI compatible version and MFCing it, I'd like to proceed in a forward direction in -current. this will result in some roto-tilling in the routing code. Firstly: the current code's idea of having a separate tree per protocol family, all of the same format, and pointed to by the 1 dimensional array is a bit silly. Especially when one considers that there is code that makes assumptions about every protocol having the same internal structures there. Some protocols don't WANT that sort of structure. (for example the whole idea of a netmask is foreign to appletalk). This needs to be made opaque to the external code. My suggested first change is to add routing method pointers to the 'domain' structure, along with information pointing the data. instead of having an array of pointers to uniform structures, there would be an array pointing to the 'domain' structures for each protocol address domain (protocol family), and the methods this reached would be called. The methods would have an argument that gives FIB number, but the protocol would be free to ignore it. When the ABI can be changed it raises the possibilty of the addition of a fib entry into the "struct route". Currently, the structure contains the sockaddr of the desination, and the resulting fib entry. To make this work fully, one could add a fib number so that given an address and a fib, one can find the third element, the fib entry. Interaction with the ARP layer/ LL layer would need to be revisited as well. Qing Li has been working on this already. This work was sponsored by Ironport Systems/Cisco Reviewed by: several including rwatson, bz and mlair (parts each) Obtained from: Ironport systems/Cisco
2008-05-09 23:03:00 +00:00
in_rtalloc(ro, 0);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (ro->ro_rt == 0) {
Update stats in struct ipstat using four new macros, IPSTAT_ADD(), IPSTAT_INC(), IPSTAT_SUB(), and IPSTAT_DEC(), rather than directly manipulating the fields across the kernel. This will make it easier to change the implementation of these statistics, such as using per-CPU versions of the data structures. MFC after: 3 days
2009-04-11 23:35:20 +00:00
IPSTAT_INC(ips_noroute);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
goto bad;
}
ifp = ro->ro_rt->rt_ifp;
ro->ro_rt->rt_use++;
if (ro->ro_rt->rt_flags & RTF_GATEWAY)
dst = satosin(ro->ro_rt->rt_gateway);
} else {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (TAILQ_EMPTY(&r->rpool.list)) {
DPFPRINTF(PF_DEBUG_URGENT,
("pf_route: TAILQ_EMPTY(&r->rpool.list)\n"));
goto bad;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (s == NULL) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_map_addr(AF_INET, r, (struct pf_addr *)&ip->ip_src,
&naddr, NULL, &sn);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (!PF_AZERO(&naddr, AF_INET))
dst->sin_addr.s_addr = naddr.v4.s_addr;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
ifp = r->rpool.cur->kif ?
r->rpool.cur->kif->pfik_ifp : NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
} else {
if (!PF_AZERO(&s->rt_addr, AF_INET))
dst->sin_addr.s_addr =
s->rt_addr.v4.s_addr;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
ifp = s->rt_kif ? s->rt_kif->pfik_ifp : NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
}
if (ifp == NULL)
goto bad;
Import two fixes from the OpenBSD stable branch: - prevent an endless loop with route-to lo0, fixes PR 3736 (dhartmei@) - The rule_number parameter for pf_get_pool() needs to be 32 bits, not 8 - this fixes corruption of the address pools with large rulesets. (mcbride@, pb@) Reviewed-by: dhartmei
2004-06-17 16:59:47 +00:00
if (oifp != ifp) {
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_UNLOCK();
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (pf_test(PF_OUT, ifp, &m0, NULL, NULL) != PF_PASS) {
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_LOCK();
goto bad;
} else if (m0 == NULL) {
PF_LOCK();
goto done;
}
PF_LOCK();
#else
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (pf_test(PF_OUT, ifp, &m0, NULL) != PF_PASS)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
goto bad;
else if (m0 == NULL)
goto done;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (m0->m_len < sizeof(struct ip)) {
DPFPRINTF(PF_DEBUG_URGENT,
("pf_route: m0->m_len < sizeof(struct ip)\n"));
goto bad;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
ip = mtod(m0, struct ip *);
}
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
/* Copied from FreeBSD 5.1-CURRENT ip_output. */
m0->m_pkthdr.csum_flags |= CSUM_IP;
sw_csum = m0->m_pkthdr.csum_flags & ~ifp->if_hwassist;
if (sw_csum & CSUM_DELAY_DATA) {
/*
* XXX: in_delayed_cksum assumes HBO for ip->ip_len (at least)
*/
NTOHS(ip->ip_len);
NTOHS(ip->ip_off); /* XXX: needed? */
in_delayed_cksum(m0);
HTONS(ip->ip_len);
HTONS(ip->ip_off);
sw_csum &= ~CSUM_DELAY_DATA;
}
m0->m_pkthdr.csum_flags &= ifp->if_hwassist;
if (ntohs(ip->ip_len) <= ifp->if_mtu ||
(ifp->if_hwassist & CSUM_FRAGMENT &&
((ip->ip_off & htons(IP_DF)) == 0))) {
/*
* ip->ip_len = htons(ip->ip_len);
* ip->ip_off = htons(ip->ip_off);
*/
ip->ip_sum = 0;
if (sw_csum & CSUM_DELAY_IP) {
/* From KAME */
if (ip->ip_v == IPVERSION &&
(ip->ip_hl << 2) == sizeof(*ip)) {
ip->ip_sum = in_cksum_hdr(ip);
} else {
ip->ip_sum = in_cksum(m0, ip->ip_hl << 2);
}
}
PF_UNLOCK();
Change if_output to take a struct route as its fourth argument in order to allow passing a cached struct llentry * down to L2 Reviewed by: rwatson
2009-04-16 20:30:28 +00:00
error = (*ifp->if_output)(ifp, m0, sintosa(dst), ro);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_LOCK();
goto done;
}
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* Copied from ip_output. */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#ifdef IPSEC
/*
* If deferred crypto processing is needed, check that the
* interface supports it.
*/
if ((mtag = m_tag_find(m0, PACKET_TAG_IPSEC_OUT_CRYPTO_NEEDED, NULL))
!= NULL && (ifp->if_capabilities & IFCAP_IPSEC) == 0) {
/* Notify IPsec to do its own crypto. */
ipsp_skipcrypto_unmark((struct tdb_ident *)(mtag + 1));
goto bad;
}
#endif /* IPSEC */
/* Catch routing changes wrt. hardware checksumming for TCP or UDP. */
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (m0->m_pkthdr.csum_flags & M_TCPV4_CSUM_OUT) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (!(ifp->if_capabilities & IFCAP_CSUM_TCPv4) ||
ifp->if_bridge != NULL) {
in_delayed_cksum(m0);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
m0->m_pkthdr.csum_flags &= ~M_TCPV4_CSUM_OUT; /* Clear */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
} else if (m0->m_pkthdr.csum_flags & M_UDPV4_CSUM_OUT) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (!(ifp->if_capabilities & IFCAP_CSUM_UDPv4) ||
ifp->if_bridge != NULL) {
in_delayed_cksum(m0);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
m0->m_pkthdr.csum_flags &= ~M_UDPV4_CSUM_OUT; /* Clear */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (ntohs(ip->ip_len) <= ifp->if_mtu) {
if ((ifp->if_capabilities & IFCAP_CSUM_IPv4) &&
ifp->if_bridge == NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
m0->m_pkthdr.csum_flags |= M_IPV4_CSUM_OUT;
Update stats in struct ipstat using four new macros, IPSTAT_ADD(), IPSTAT_INC(), IPSTAT_SUB(), and IPSTAT_DEC(), rather than directly manipulating the fields across the kernel. This will make it easier to change the implementation of these statistics, such as using per-CPU versions of the data structures. MFC after: 3 days
2009-04-11 23:35:20 +00:00
IPSTAT_INC(ips_outhwcsum);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
} else {
ip->ip_sum = 0;
ip->ip_sum = in_cksum(m0, ip->ip_hl << 2);
}
/* Update relevant hardware checksum stats for TCP/UDP */
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (m0->m_pkthdr.csum_flags & M_TCPV4_CSUM_OUT)
Update stats in struct tcpstat using two new macros, TCPSTAT_ADD() and TCPSTAT_INC(), rather than directly manipulating the fields across the kernel. This will make it easier to change the implementation of these statistics, such as using per-CPU versions of the data structures. MFC after: 3 days
2009-04-11 22:07:19 +00:00
TCPSTAT_INC(tcpstat.tcps_outhwcsum);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
else if (m0->m_pkthdr.csum_flags & M_UDPV4_CSUM_OUT)
Update stats in struct udpstat using two new macros, UDPSTAT_ADD() and UDPSTAT_INC(), rather than directly manipulating the fields across the kernel. This will make it easier to change the implementation of these statistics, such as using per-CPU versions of the data structures. MFC after: 3 days
2009-04-12 11:42:40 +00:00
UDPSTAT_INC(udps_outhwcsum);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
error = (*ifp->if_output)(ifp, m0, sintosa(dst), NULL);
goto done;
}
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/*
* Too large for interface; fragment if possible.
* Must be able to put at least 8 bytes per fragment.
*/
if (ip->ip_off & htons(IP_DF)) {
Update stats in struct ipstat using four new macros, IPSTAT_ADD(), IPSTAT_INC(), IPSTAT_SUB(), and IPSTAT_DEC(), rather than directly manipulating the fields across the kernel. This will make it easier to change the implementation of these statistics, such as using per-CPU versions of the data structures. MFC after: 3 days
2009-04-11 23:35:20 +00:00
IPSTAT_INC(ips_cantfrag);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (r->rt != PF_DUPTO) {
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
/* icmp_error() expects host byte ordering */
NTOHS(ip->ip_len);
NTOHS(ip->ip_off);
PF_UNLOCK();
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
icmp_error(m0, ICMP_UNREACH, ICMP_UNREACH_NEEDFRAG, 0,
if_mtu not ifp_mtu.
2005-05-04 19:51:06 +00:00
ifp->if_mtu);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_LOCK();
Combine rev. 1.29 and 1.30 to something that will make sense for future imports.
2005-05-04 15:55:29 +00:00
#else
icmp_error(m0, ICMP_UNREACH, ICMP_UNREACH_NEEDFRAG, 0,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
ifp->if_mtu);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
goto done;
} else
goto bad;
}
m1 = m0;
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
/*
* XXX: is cheaper + less error prone than own function
*/
NTOHS(ip->ip_len);
NTOHS(ip->ip_off);
error = ip_fragment(ip, &m0, ifp->if_mtu, ifp->if_hwassist, sw_csum);
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
error = ip_fragment(m0, ifp, ifp->if_mtu);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Commit two fixes from OpenBSD's stable branch: - Fix binat for incoming connections when a netblock (not just a single address) is used for source in the binat rule. closes PR 3535, reported by Karl O.Pinc. ok henning@, cedric@ - Fix a problem related to empty anchor rulesets, which could cause a kernel panic. Approved by: bms(mentor)
2004-03-28 20:04:12 +00:00
if (error) {
#ifndef __FreeBSD__ /* ip_fragment does not do m_freem() on FreeBSD */
m0 = NULL;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
goto bad;
Commit two fixes from OpenBSD's stable branch: - Fix binat for incoming connections when a netblock (not just a single address) is used for source in the binat rule. closes PR 3535, reported by Karl O.Pinc. ok henning@, cedric@ - Fix a problem related to empty anchor rulesets, which could cause a kernel panic. Approved by: bms(mentor)
2004-03-28 20:04:12 +00:00
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
for (m0 = m1; m0; m0 = m1) {
m1 = m0->m_nextpkt;
m0->m_nextpkt = 0;
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
if (error == 0) {
PF_UNLOCK();
error = (*ifp->if_output)(ifp, m0, sintosa(dst),
NULL);
PF_LOCK();
} else
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (error == 0)
error = (*ifp->if_output)(ifp, m0, sintosa(dst),
NULL);
else
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
m_freem(m0);
}
if (error == 0)
Update stats in struct ipstat using four new macros, IPSTAT_ADD(), IPSTAT_INC(), IPSTAT_SUB(), and IPSTAT_DEC(), rather than directly manipulating the fields across the kernel. This will make it easier to change the implementation of these statistics, such as using per-CPU versions of the data structures. MFC after: 3 days
2009-04-11 23:35:20 +00:00
IPSTAT_INC(ips_fragmented);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
done:
if (r->rt != PF_DUPTO)
*m = NULL;
if (ro == &iproute && ro->ro_rt)
RTFREE(ro->ro_rt);
return;
bad:
m_freem(m0);
goto done;
}
#endif /* INET */
#ifdef INET6
void
pf_route6(struct mbuf **m, struct pf_rule *r, int dir, struct ifnet *oifp,
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct pf_state *s, struct pf_pdesc *pd)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
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;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_src_node *sn = NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
int error = 0;
if (m == NULL || *m == NULL || r == NULL ||
(dir != PF_IN && dir != PF_OUT) || oifp == NULL)
panic("pf_route6: invalid parameters");
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pd->pf_mtag->routed++ > 3) {
m0 = *m;
*m = NULL;
goto bad;
Merge in a stable fix from OpenBSD: MFC: Fix by dhartmei@ change pf_route() loop detection: introduce a counter (number of times a packet is routed already) in the mbuf tag, allow at most four times. Fixes some legitimate cases broken by the previous change. Reviewed by: dhartmei
2004-07-17 17:15:15 +00:00
}
if (r->rt == PF_DUPTO) {
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Merge in a stable fix from OpenBSD: MFC: Fix by dhartmei@ change pf_route() loop detection: introduce a counter (number of times a packet is routed already) in the mbuf tag, allow at most four times. Fixes some legitimate cases broken by the previous change. Reviewed by: dhartmei
2004-07-17 17:15:15 +00:00
if ((m0 = m_dup(*m, M_DONTWAIT)) == NULL)
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#else
Merge in a stable fix from OpenBSD: MFC: Fix by dhartmei@ change pf_route() loop detection: introduce a counter (number of times a packet is routed already) in the mbuf tag, allow at most four times. Fixes some legitimate cases broken by the previous change. Reviewed by: dhartmei
2004-07-17 17:15:15 +00:00
if ((m0 = m_copym2(*m, 0, M_COPYALL, M_NOWAIT)) == NULL)
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return;
} else {
if ((r->rt == PF_REPLYTO) == (r->direction == dir))
return;
m0 = *m;
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (m0->m_len < sizeof(struct ip6_hdr)) {
DPFPRINTF(PF_DEBUG_URGENT,
("pf_route6: m0->m_len < sizeof(struct ip6_hdr)\n"));
goto bad;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
ip6 = mtod(m0, struct ip6_hdr *);
ro = &ip6route;
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_addr = ip6->ip6_dst;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
/* Cheat. XXX why only in the v6 case??? */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (r->rt == PF_FASTROUTE) {
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Define semantic of M_SKIP_FIREWALL more precisely, i.e. also pass associated icmp_error() packets. While here retire PACKET_TAG_PF_GENERATED (which served the same purpose) and use M_SKIP_FIREWALL in pf as well. This should speed up things a bit as we get rid of the tag allocations. Discussed with: juli
2004-07-17 05:10:06 +00:00
m0->m_flags |= M_SKIP_FIREWALL;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_UNLOCK();
ip6_output(m0, NULL, NULL, 0, NULL, NULL, NULL);
PF_LOCK();
#else
Define semantic of M_SKIP_FIREWALL more precisely, i.e. also pass associated icmp_error() packets. While here retire PACKET_TAG_PF_GENERATED (which served the same purpose) and use M_SKIP_FIREWALL in pf as well. This should speed up things a bit as we get rid of the tag allocations. Discussed with: juli
2004-07-17 05:10:06 +00:00
mtag = m_tag_get(PACKET_TAG_PF_GENERATED, 0, M_NOWAIT);
if (mtag == NULL)
goto bad;
m_tag_prepend(m0, mtag);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pd->pf_mtag->flags |= PF_TAG_GENERATED;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
ip6_output(m0, NULL, NULL, 0, NULL, NULL);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return;
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (TAILQ_EMPTY(&r->rpool.list)) {
DPFPRINTF(PF_DEBUG_URGENT,
("pf_route6: TAILQ_EMPTY(&r->rpool.list)\n"));
goto bad;
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (s == NULL) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_map_addr(AF_INET6, r, (struct pf_addr *)&ip6->ip6_src,
&naddr, NULL, &sn);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (!PF_AZERO(&naddr, AF_INET6))
PF_ACPY((struct pf_addr *)&dst->sin6_addr,
&naddr, AF_INET6);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
ifp = r->rpool.cur->kif ? r->rpool.cur->kif->pfik_ifp : NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
} else {
if (!PF_AZERO(&s->rt_addr, AF_INET6))
PF_ACPY((struct pf_addr *)&dst->sin6_addr,
&s->rt_addr, AF_INET6);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
ifp = s->rt_kif ? s->rt_kif->pfik_ifp : NULL;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
if (ifp == NULL)
goto bad;
if (oifp != ifp) {
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Merge in a stable fix from OpenBSD: MFC: Fix by dhartmei@ change pf_route() loop detection: introduce a counter (number of times a packet is routed already) in the mbuf tag, allow at most four times. Fixes some legitimate cases broken by the previous change. Reviewed by: dhartmei
2004-07-17 17:15:15 +00:00
PF_UNLOCK();
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (pf_test6(PF_OUT, ifp, &m0, NULL, NULL) != PF_PASS) {
Merge in a stable fix from OpenBSD: MFC: Fix by dhartmei@ change pf_route() loop detection: introduce a counter (number of times a packet is routed already) in the mbuf tag, allow at most four times. Fixes some legitimate cases broken by the previous change. Reviewed by: dhartmei
2004-07-17 17:15:15 +00:00
PF_LOCK();
goto bad;
} else if (m0 == NULL) {
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_LOCK();
Merge in a stable fix from OpenBSD: MFC: Fix by dhartmei@ change pf_route() loop detection: introduce a counter (number of times a packet is routed already) in the mbuf tag, allow at most four times. Fixes some legitimate cases broken by the previous change. Reviewed by: dhartmei
2004-07-17 17:15:15 +00:00
goto done;
}
PF_LOCK();
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#else
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (pf_test6(PF_OUT, ifp, &m0, NULL) != PF_PASS)
Merge in a stable fix from OpenBSD: MFC: Fix by dhartmei@ change pf_route() loop detection: introduce a counter (number of times a packet is routed already) in the mbuf tag, allow at most four times. Fixes some legitimate cases broken by the previous change. Reviewed by: dhartmei
2004-07-17 17:15:15 +00:00
goto bad;
else if (m0 == NULL)
goto done;
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (m0->m_len < sizeof(struct ip6_hdr)) {
DPFPRINTF(PF_DEBUG_URGENT,
("pf_route6: m0->m_len < sizeof(struct ip6_hdr)\n"));
goto bad;
}
Merge in a stable fix from OpenBSD: MFC: Fix by dhartmei@ change pf_route() loop detection: introduce a counter (number of times a packet is routed already) in the mbuf tag, allow at most four times. Fixes some legitimate cases broken by the previous change. Reviewed by: dhartmei
2004-07-17 17:15:15 +00:00
ip6 = mtod(m0, struct ip6_hdr *);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
/*
* If the packet is too large for the outgoing interface,
* send back an icmp6 error.
*/
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (IN6_IS_SCOPE_EMBED(&dst->sin6_addr))
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
dst->sin6_addr.s6_addr16[1] = htons(ifp->if_index);
if ((u_long)m0->m_pkthdr.len <= ifp->if_mtu) {
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_UNLOCK();
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
error = nd6_output(ifp, ifp, m0, dst, NULL);
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_LOCK();
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
} else {
in6_ifstat_inc(ifp, ifs6_in_toobig);
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
if (r->rt != PF_DUPTO) {
PF_UNLOCK();
icmp6_error(m0, ICMP6_PACKET_TOO_BIG, 0, ifp->if_mtu);
PF_LOCK();
} else
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (r->rt != PF_DUPTO)
icmp6_error(m0, ICMP6_PACKET_TOO_BIG, 0, ifp->if_mtu);
else
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
goto bad;
}
done:
if (r->rt != PF_DUPTO)
*m = NULL;
return;
bad:
m_freem(m0);
goto done;
}
#endif /* INET6 */
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
/*
Refine pf_check_proto_cksum() a bit in order to avoid additional in_pseudo() calls further down the stack. If we find the cksum to be okay we pretend that the hardware did all the work and hence keep the upper layers from checking again. Submitted by: Pyun YongHyeon
2004-07-23 03:31:42 +00:00
* FreeBSD supports cksum offloads for the following drivers.
Removed the remnants of gx(4).
2004-11-08 20:24:52 +00:00
* em(4), fxp(4), ixgb(4), lge(4), ndis(4), nge(4), re(4),
Refine pf_check_proto_cksum() a bit in order to avoid additional in_pseudo() calls further down the stack. If we find the cksum to be okay we pretend that the hardware did all the work and hence keep the upper layers from checking again. Submitted by: Pyun YongHyeon
2004-07-23 03:31:42 +00:00
* ti(4), txp(4), xl(4)
*
* CSUM_DATA_VALID | CSUM_PSEUDO_HDR :
* network driver performed cksum including pseudo header, need to verify
* csum_data
* CSUM_DATA_VALID :
* network driver performed cksum, needs to additional pseudo header
* cksum computation with partial csum_data(i.e. lack of H/W support for
* pseudo header, for instance hme(4), sk(4) and possibly gem(4))
*
* After validating the cksum of packet, set both flag CSUM_DATA_VALID and
* CSUM_PSEUDO_HDR in order to avoid recomputation of the cksum in upper
* TCP/UDP layer.
* Also, set csum_data to 0xffff to force cksum validation.
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
*/
int
pf_check_proto_cksum(struct mbuf *m, int off, int len, u_int8_t p, sa_family_t af)
{
u_int16_t sum = 0;
int hw_assist = 0;
struct ip *ip;
if (off < sizeof(struct ip) || len < sizeof(struct udphdr))
return (1);
if (m->m_pkthdr.len < off + len)
return (1);
switch (p) {
case IPPROTO_TCP:
if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) {
if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR) {
sum = m->m_pkthdr.csum_data;
} else {
ip = mtod(m, struct ip *);
sum = in_pseudo(ip->ip_src.s_addr,
One more round on the H/W supported checksum code: Fix NICs that can't do the pseudo header. We really need the TCP packet length here. This happens to end up in ip->ip_len in tcp_input.c, but here we should get it from the len function variable instead. Submitted by: yongari Tested by: Nicolas Linard, yongari (sparc64 + hme) MFC after: 5 days
2004-09-11 11:18:25 +00:00
ip->ip_dst.s_addr, htonl((u_short)len +
m->m_pkthdr.csum_data + IPPROTO_TCP));
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
}
sum ^= 0xffff;
++hw_assist;
}
break;
case IPPROTO_UDP:
if (m->m_pkthdr.csum_flags & CSUM_DATA_VALID) {
if (m->m_pkthdr.csum_flags & CSUM_PSEUDO_HDR) {
sum = m->m_pkthdr.csum_data;
} else {
ip = mtod(m, struct ip *);
sum = in_pseudo(ip->ip_src.s_addr,
ip->ip_dst.s_addr, htonl((u_short)len +
m->m_pkthdr.csum_data + IPPROTO_UDP));
}
sum ^= 0xffff;
++hw_assist;
}
break;
case IPPROTO_ICMP:
#ifdef INET6
case IPPROTO_ICMPV6:
#endif /* INET6 */
break;
default:
return (1);
}
if (!hw_assist) {
switch (af) {
case AF_INET:
if (p == IPPROTO_ICMP) {
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 (m->m_len < sizeof(struct ip))
return (1);
sum = in4_cksum(m, p, off, len);
}
break;
#ifdef INET6
case AF_INET6:
if (m->m_len < sizeof(struct ip6_hdr))
return (1);
sum = in6_cksum(m, p, off, len);
break;
#endif /* INET6 */
default:
return (1);
}
}
if (sum) {
switch (p) {
case IPPROTO_TCP:
Step 1.5 of importing the network stack virtualization infrastructure from the vimage project, as per plan established at devsummit 08/08: http://wiki.freebsd.org/Image/Notes200808DevSummit Introduce INIT_VNET_*() initializer macros, VNET_FOREACH() iterator macros, and CURVNET_SET() context setting macros, all currently resolving to NOPs. Prepare for virtualization of selected SYSCTL objects by introducing a family of SYSCTL_V_*() macros, currently resolving to their global counterparts, i.e. SYSCTL_V_INT() == SYSCTL_INT(). Move selected #defines from sys/sys/vimage.h to newly introduced header files specific to virtualized subsystems (sys/net/vnet.h, sys/netinet/vinet.h etc.). All the changes are verified to have zero functional impact at this point in time by doing MD5 comparision between pre- and post-change object files(*). (*) netipsec/keysock.c did not validate depending on compile time options. Implemented by: julian, bz, brooks, zec Reviewed by: julian, bz, brooks, kris, rwatson, ... Approved by: julian (mentor) Obtained from: //depot/projects/vimage-commit2/... X-MFC after: never Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-10-02 15:37:58 +00:00
{
Update stats in struct tcpstat using two new macros, TCPSTAT_ADD() and TCPSTAT_INC(), rather than directly manipulating the fields across the kernel. This will make it easier to change the implementation of these statistics, such as using per-CPU versions of the data structures. MFC after: 3 days
2009-04-11 22:07:19 +00:00
TCPSTAT_INC(tcps_rcvbadsum);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
break;
Step 1.5 of importing the network stack virtualization infrastructure from the vimage project, as per plan established at devsummit 08/08: http://wiki.freebsd.org/Image/Notes200808DevSummit Introduce INIT_VNET_*() initializer macros, VNET_FOREACH() iterator macros, and CURVNET_SET() context setting macros, all currently resolving to NOPs. Prepare for virtualization of selected SYSCTL objects by introducing a family of SYSCTL_V_*() macros, currently resolving to their global counterparts, i.e. SYSCTL_V_INT() == SYSCTL_INT(). Move selected #defines from sys/sys/vimage.h to newly introduced header files specific to virtualized subsystems (sys/net/vnet.h, sys/netinet/vinet.h etc.). All the changes are verified to have zero functional impact at this point in time by doing MD5 comparision between pre- and post-change object files(*). (*) netipsec/keysock.c did not validate depending on compile time options. Implemented by: julian, bz, brooks, zec Reviewed by: julian, bz, brooks, kris, rwatson, ... Approved by: julian (mentor) Obtained from: //depot/projects/vimage-commit2/... X-MFC after: never Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-10-02 15:37:58 +00:00
}
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
case IPPROTO_UDP:
Step 1.5 of importing the network stack virtualization infrastructure from the vimage project, as per plan established at devsummit 08/08: http://wiki.freebsd.org/Image/Notes200808DevSummit Introduce INIT_VNET_*() initializer macros, VNET_FOREACH() iterator macros, and CURVNET_SET() context setting macros, all currently resolving to NOPs. Prepare for virtualization of selected SYSCTL objects by introducing a family of SYSCTL_V_*() macros, currently resolving to their global counterparts, i.e. SYSCTL_V_INT() == SYSCTL_INT(). Move selected #defines from sys/sys/vimage.h to newly introduced header files specific to virtualized subsystems (sys/net/vnet.h, sys/netinet/vinet.h etc.). All the changes are verified to have zero functional impact at this point in time by doing MD5 comparision between pre- and post-change object files(*). (*) netipsec/keysock.c did not validate depending on compile time options. Implemented by: julian, bz, brooks, zec Reviewed by: julian, bz, brooks, kris, rwatson, ... Approved by: julian (mentor) Obtained from: //depot/projects/vimage-commit2/... X-MFC after: never Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-10-02 15:37:58 +00:00
{
Update stats in struct udpstat using two new macros, UDPSTAT_ADD() and UDPSTAT_INC(), rather than directly manipulating the fields across the kernel. This will make it easier to change the implementation of these statistics, such as using per-CPU versions of the data structures. MFC after: 3 days
2009-04-12 11:42:40 +00:00
UDPSTAT_INC(udps_badsum);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
break;
Step 1.5 of importing the network stack virtualization infrastructure from the vimage project, as per plan established at devsummit 08/08: http://wiki.freebsd.org/Image/Notes200808DevSummit Introduce INIT_VNET_*() initializer macros, VNET_FOREACH() iterator macros, and CURVNET_SET() context setting macros, all currently resolving to NOPs. Prepare for virtualization of selected SYSCTL objects by introducing a family of SYSCTL_V_*() macros, currently resolving to their global counterparts, i.e. SYSCTL_V_INT() == SYSCTL_INT(). Move selected #defines from sys/sys/vimage.h to newly introduced header files specific to virtualized subsystems (sys/net/vnet.h, sys/netinet/vinet.h etc.). All the changes are verified to have zero functional impact at this point in time by doing MD5 comparision between pre- and post-change object files(*). (*) netipsec/keysock.c did not validate depending on compile time options. Implemented by: julian, bz, brooks, zec Reviewed by: julian, bz, brooks, kris, rwatson, ... Approved by: julian (mentor) Obtained from: //depot/projects/vimage-commit2/... X-MFC after: never Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-10-02 15:37:58 +00:00
}
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
case IPPROTO_ICMP:
Step 1.5 of importing the network stack virtualization infrastructure from the vimage project, as per plan established at devsummit 08/08: http://wiki.freebsd.org/Image/Notes200808DevSummit Introduce INIT_VNET_*() initializer macros, VNET_FOREACH() iterator macros, and CURVNET_SET() context setting macros, all currently resolving to NOPs. Prepare for virtualization of selected SYSCTL objects by introducing a family of SYSCTL_V_*() macros, currently resolving to their global counterparts, i.e. SYSCTL_V_INT() == SYSCTL_INT(). Move selected #defines from sys/sys/vimage.h to newly introduced header files specific to virtualized subsystems (sys/net/vnet.h, sys/netinet/vinet.h etc.). All the changes are verified to have zero functional impact at this point in time by doing MD5 comparision between pre- and post-change object files(*). (*) netipsec/keysock.c did not validate depending on compile time options. Implemented by: julian, bz, brooks, zec Reviewed by: julian, bz, brooks, kris, rwatson, ... Approved by: julian (mentor) Obtained from: //depot/projects/vimage-commit2/... X-MFC after: never Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-10-02 15:37:58 +00:00
{
Update stats in struct icmpstat and icmp6stat using four new macros: ICMPSTAT_ADD(), ICMPSTAT_INC(), ICMP6STAT_ADD(), and ICMP6STAT_INC(), rather than directly manipulating the fields of these structures across the kernel. This will make it easier to change the implementation of these statistics, such as using per-CPU versions of the data structures. In on case, icmp6stat members are manipulated indirectly, by icmp6_errcount(), and this will require further work to fix for per-CPU stats. MFC after: 3 days
2009-04-12 13:22:33 +00:00
ICMPSTAT_INC(icps_checksum);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
break;
Step 1.5 of importing the network stack virtualization infrastructure from the vimage project, as per plan established at devsummit 08/08: http://wiki.freebsd.org/Image/Notes200808DevSummit Introduce INIT_VNET_*() initializer macros, VNET_FOREACH() iterator macros, and CURVNET_SET() context setting macros, all currently resolving to NOPs. Prepare for virtualization of selected SYSCTL objects by introducing a family of SYSCTL_V_*() macros, currently resolving to their global counterparts, i.e. SYSCTL_V_INT() == SYSCTL_INT(). Move selected #defines from sys/sys/vimage.h to newly introduced header files specific to virtualized subsystems (sys/net/vnet.h, sys/netinet/vinet.h etc.). All the changes are verified to have zero functional impact at this point in time by doing MD5 comparision between pre- and post-change object files(*). (*) netipsec/keysock.c did not validate depending on compile time options. Implemented by: julian, bz, brooks, zec Reviewed by: julian, bz, brooks, kris, rwatson, ... Approved by: julian (mentor) Obtained from: //depot/projects/vimage-commit2/... X-MFC after: never Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-10-02 15:37:58 +00:00
}
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#ifdef INET6
case IPPROTO_ICMPV6:
Step 1.5 of importing the network stack virtualization infrastructure from the vimage project, as per plan established at devsummit 08/08: http://wiki.freebsd.org/Image/Notes200808DevSummit Introduce INIT_VNET_*() initializer macros, VNET_FOREACH() iterator macros, and CURVNET_SET() context setting macros, all currently resolving to NOPs. Prepare for virtualization of selected SYSCTL objects by introducing a family of SYSCTL_V_*() macros, currently resolving to their global counterparts, i.e. SYSCTL_V_INT() == SYSCTL_INT(). Move selected #defines from sys/sys/vimage.h to newly introduced header files specific to virtualized subsystems (sys/net/vnet.h, sys/netinet/vinet.h etc.). All the changes are verified to have zero functional impact at this point in time by doing MD5 comparision between pre- and post-change object files(*). (*) netipsec/keysock.c did not validate depending on compile time options. Implemented by: julian, bz, brooks, zec Reviewed by: julian, bz, brooks, kris, rwatson, ... Approved by: julian (mentor) Obtained from: //depot/projects/vimage-commit2/... X-MFC after: never Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-10-02 15:37:58 +00:00
{
Update stats in struct icmpstat and icmp6stat using four new macros: ICMPSTAT_ADD(), ICMPSTAT_INC(), ICMP6STAT_ADD(), and ICMP6STAT_INC(), rather than directly manipulating the fields of these structures across the kernel. This will make it easier to change the implementation of these statistics, such as using per-CPU versions of the data structures. In on case, icmp6stat members are manipulated indirectly, by icmp6_errcount(), and this will require further work to fix for per-CPU stats. MFC after: 3 days
2009-04-12 13:22:33 +00:00
ICMP6STAT_INC(icp6s_checksum);
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
break;
Step 1.5 of importing the network stack virtualization infrastructure from the vimage project, as per plan established at devsummit 08/08: http://wiki.freebsd.org/Image/Notes200808DevSummit Introduce INIT_VNET_*() initializer macros, VNET_FOREACH() iterator macros, and CURVNET_SET() context setting macros, all currently resolving to NOPs. Prepare for virtualization of selected SYSCTL objects by introducing a family of SYSCTL_V_*() macros, currently resolving to their global counterparts, i.e. SYSCTL_V_INT() == SYSCTL_INT(). Move selected #defines from sys/sys/vimage.h to newly introduced header files specific to virtualized subsystems (sys/net/vnet.h, sys/netinet/vinet.h etc.). All the changes are verified to have zero functional impact at this point in time by doing MD5 comparision between pre- and post-change object files(*). (*) netipsec/keysock.c did not validate depending on compile time options. Implemented by: julian, bz, brooks, zec Reviewed by: julian, bz, brooks, kris, rwatson, ... Approved by: julian (mentor) Obtained from: //depot/projects/vimage-commit2/... X-MFC after: never Sponsored by: NLnet Foundation, The FreeBSD Foundation
2008-10-02 15:37:58 +00:00
}
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
#endif /* INET6 */
}
return (1);
Refine pf_check_proto_cksum() a bit in order to avoid additional in_pseudo() calls further down the stack. If we find the cksum to be okay we pretend that the hardware did all the work and hence keep the upper layers from checking again. Submitted by: Pyun YongHyeon
2004-07-23 03:31:42 +00:00
} else {
if (p == IPPROTO_TCP || p == IPPROTO_UDP) {
m->m_pkthdr.csum_flags |=
(CSUM_DATA_VALID | CSUM_PSEUDO_HDR);
m->m_pkthdr.csum_data = 0xffff;
}
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
}
return (0);
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#else /* !__FreeBSD__ */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/*
* check protocol (tcp/udp/icmp/icmp6) checksum and set mbuf flag
* off is the offset where the protocol header starts
* len is the total length of protocol header plus payload
* returns 0 when the checksum is valid, otherwise returns 1.
*/
int
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
pf_check_proto_cksum(struct mbuf *m, int off, int len, u_int8_t p,
sa_family_t af)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
u_int16_t flag_ok, flag_bad;
u_int16_t sum;
switch (p) {
case IPPROTO_TCP:
flag_ok = M_TCP_CSUM_IN_OK;
flag_bad = M_TCP_CSUM_IN_BAD;
break;
case IPPROTO_UDP:
flag_ok = M_UDP_CSUM_IN_OK;
flag_bad = M_UDP_CSUM_IN_BAD;
break;
case IPPROTO_ICMP:
#ifdef INET6
case IPPROTO_ICMPV6:
#endif /* INET6 */
flag_ok = flag_bad = 0;
break;
default:
return (1);
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (m->m_pkthdr.csum_flags & flag_ok)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (0);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (m->m_pkthdr.csum_flags & flag_bad)
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (1);
if (off < sizeof(struct ip) || len < sizeof(struct udphdr))
return (1);
if (m->m_pkthdr.len < off + len)
return (1);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
switch (af) {
#ifdef INET
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
case AF_INET:
if (p == IPPROTO_ICMP) {
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 (m->m_len < sizeof(struct ip))
return (1);
sum = in4_cksum(m, p, off, len);
}
break;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* INET */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef INET6
case AF_INET6:
if (m->m_len < sizeof(struct ip6_hdr))
return (1);
sum = in6_cksum(m, p, off, len);
break;
#endif /* INET6 */
default:
return (1);
}
if (sum) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
m->m_pkthdr.csum_flags |= flag_bad;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
switch (p) {
case IPPROTO_TCP:
Update stats in struct tcpstat using two new macros, TCPSTAT_ADD() and TCPSTAT_INC(), rather than directly manipulating the fields across the kernel. This will make it easier to change the implementation of these statistics, such as using per-CPU versions of the data structures. MFC after: 3 days
2009-04-11 22:07:19 +00:00
TCPSTAT_INC(tcps_rcvbadsum);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
case IPPROTO_UDP:
Update stats in struct udpstat using two new macros, UDPSTAT_ADD() and UDPSTAT_INC(), rather than directly manipulating the fields across the kernel. This will make it easier to change the implementation of these statistics, such as using per-CPU versions of the data structures. MFC after: 3 days
2009-04-12 11:42:40 +00:00
UDPSTAT_INC(udps_badsum);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
case IPPROTO_ICMP:
Update stats in struct icmpstat and icmp6stat using four new macros: ICMPSTAT_ADD(), ICMPSTAT_INC(), ICMP6STAT_ADD(), and ICMP6STAT_INC(), rather than directly manipulating the fields of these structures across the kernel. This will make it easier to change the implementation of these statistics, such as using per-CPU versions of the data structures. In on case, icmp6stat members are manipulated indirectly, by icmp6_errcount(), and this will require further work to fix for per-CPU stats. MFC after: 3 days
2009-04-12 13:22:33 +00:00
ICMPSTAT_INC(icps_checksum);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#ifdef INET6
case IPPROTO_ICMPV6:
Update stats in struct icmpstat and icmp6stat using four new macros: ICMPSTAT_ADD(), ICMPSTAT_INC(), ICMP6STAT_ADD(), and ICMP6STAT_INC(), rather than directly manipulating the fields of these structures across the kernel. This will make it easier to change the implementation of these statistics, such as using per-CPU versions of the data structures. In on case, icmp6stat members are manipulated indirectly, by icmp6_errcount(), and this will require further work to fix for per-CPU stats. MFC after: 3 days
2009-04-12 13:22:33 +00:00
ICMP6STAT_INC(icp6s_checksum);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
#endif /* INET6 */
}
return (1);
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
m->m_pkthdr.csum_flags |= flag_ok;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
return (0);
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#endif /* __FreeBSD__ */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef INET
int
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_test(int dir, struct ifnet *ifp, struct mbuf **m0,
struct ether_header *eh, struct inpcb *inp)
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_test(int dir, struct ifnet *ifp, struct mbuf **m0,
struct ether_header *eh)
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pfi_kif *kif;
u_short action, reason = 0, log = 0;
struct mbuf *m = *m0;
struct ip *h = NULL; /* make the compiler happy */
struct pf_rule *a = NULL, *r = &pf_default_rule, *tr, *nr;
struct pf_state *s = NULL;
struct pf_ruleset *ruleset = NULL;
struct pf_pdesc pd;
int off, dirndx, pqid = 0;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_LOCK();
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (!pf_status.running)
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
{
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_UNLOCK();
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
return (PF_PASS);
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#endif
memset(&pd, 0, sizeof(pd));
if ((pd.pf_mtag = pf_get_mtag(m)) == NULL) {
#ifdef __FreeBSD__
PF_UNLOCK();
#endif
DPFPRINTF(PF_DEBUG_URGENT,
("pf_test: pf_get_mtag returned NULL\n"));
return (PF_DROP);
}
#ifdef __FreeBSD__
if (m->m_flags & M_SKIP_FIREWALL) {
PF_UNLOCK();
return (PF_PASS);
}
#else
if (pd.pf_mtag->flags & PF_TAG_GENERATED)
return (PF_PASS);
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#ifdef __FreeBSD__
/* XXX_IMPORT: later */
#else
if (ifp->if_type == IFT_CARP && ifp->if_carpdev)
ifp = ifp->if_carpdev;
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
kif = (struct pfi_kif *)ifp->if_pf_kif;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (kif == NULL) {
#ifdef __FreeBSD__
PF_UNLOCK();
#endif
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
DPFPRINTF(PF_DEBUG_URGENT,
("pf_test: kif == NULL, if_xname %s\n", ifp->if_xname));
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
return (PF_DROP);
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (kif->pfik_flags & PFI_IFLAG_SKIP) {
#ifdef __FreeBSD__
PF_UNLOCK();
#endif
return (PF_PASS);
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
M_ASSERTPKTHDR(m);
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef DIAGNOSTIC
if ((m->m_flags & M_PKTHDR) == 0)
panic("non-M_PKTHDR is passed to pf_test");
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* DIAGNOSTIC */
#endif /* __FreeBSD__ */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (m->m_pkthdr.len < (int)sizeof(*h)) {
action = PF_DROP;
REASON_SET(&reason, PFRES_SHORT);
log = 1;
goto done;
}
/* We do IP header normalization and packet reassembly here */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (pf_normalize_ip(m0, dir, kif, &reason, &pd) != PF_PASS) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
action = PF_DROP;
goto done;
}
m = *m0;
h = mtod(m, struct ip *);
off = h->ip_hl << 2;
if (off < (int)sizeof(*h)) {
action = PF_DROP;
REASON_SET(&reason, PFRES_SHORT);
log = 1;
goto done;
}
pd.src = (struct pf_addr *)&h->ip_src;
pd.dst = (struct pf_addr *)&h->ip_dst;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
PF_ACPY(&pd.baddr, dir == PF_OUT ? pd.src : pd.dst, AF_INET);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pd.ip_sum = &h->ip_sum;
pd.proto = h->ip_p;
pd.af = AF_INET;
pd.tos = h->ip_tos;
pd.tot_len = ntohs(h->ip_len);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pd.eh = eh;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
/* handle fragments that didn't get reassembled by normalization */
if (h->ip_off & htons(IP_MF | IP_OFFMASK)) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_test_fragment(&r, dir, kif, m, h,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
&pd, &a, &ruleset);
goto done;
}
switch (h->ip_p) {
case IPPROTO_TCP: {
struct tcphdr th;
pd.hdr.tcp = &th;
if (!pf_pull_hdr(m, off, &th, sizeof(th),
&action, &reason, AF_INET)) {
log = action != PF_PASS;
goto done;
}
if (dir == PF_IN && pf_check_proto_cksum(m, off,
ntohs(h->ip_len) - off, IPPROTO_TCP, AF_INET)) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
REASON_SET(&reason, PFRES_PROTCKSUM);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
action = PF_DROP;
goto done;
}
pd.p_len = pd.tot_len - off - (th.th_off << 2);
if ((th.th_flags & TH_ACK) && pd.p_len == 0)
pqid = 1;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_normalize_tcp(dir, kif, m, 0, off, h, &pd);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (action == PF_DROP)
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto done;
action = pf_test_state_tcp(&s, dir, kif, m, off, h, &pd,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
&reason);
if (action == PF_PASS) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#if NPFSYNC
pfsync_update_state(s);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* NPFSYNC */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = s->rule.ptr;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
a = s->anchor.ptr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
log = s->log;
} else if (s == NULL)
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
action = pf_test_tcp(&r, &s, dir, kif,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
m, off, h, &pd, &a, &ruleset, NULL, inp);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_test_tcp(&r, &s, dir, kif,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
m, off, h, &pd, &a, &ruleset, &ipintrq);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
}
case IPPROTO_UDP: {
struct udphdr uh;
pd.hdr.udp = &uh;
if (!pf_pull_hdr(m, off, &uh, sizeof(uh),
&action, &reason, AF_INET)) {
log = action != PF_PASS;
goto done;
}
if (dir == PF_IN && uh.uh_sum && pf_check_proto_cksum(m,
off, ntohs(h->ip_len) - off, IPPROTO_UDP, AF_INET)) {
action = PF_DROP;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
REASON_SET(&reason, PFRES_PROTCKSUM);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
goto done;
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (uh.uh_dport == 0 ||
ntohs(uh.uh_ulen) > m->m_pkthdr.len - off ||
ntohs(uh.uh_ulen) < sizeof(struct udphdr)) {
action = PF_DROP;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
REASON_SET(&reason, PFRES_SHORT);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto done;
}
action = pf_test_state_udp(&s, dir, kif, m, off, h, &pd);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (action == PF_PASS) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#if NPFSYNC
pfsync_update_state(s);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* NPFSYNC */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = s->rule.ptr;
a = s->anchor.ptr;
log = s->log;
} else if (s == NULL)
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
action = pf_test_udp(&r, &s, dir, kif,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
m, off, h, &pd, &a, &ruleset, NULL, inp);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_test_udp(&r, &s, dir, kif,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
m, off, h, &pd, &a, &ruleset, &ipintrq);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
}
case IPPROTO_ICMP: {
struct icmp ih;
pd.hdr.icmp = &ih;
if (!pf_pull_hdr(m, off, &ih, ICMP_MINLEN,
&action, &reason, AF_INET)) {
log = action != PF_PASS;
goto done;
}
if (dir == PF_IN && pf_check_proto_cksum(m, off,
ntohs(h->ip_len) - off, IPPROTO_ICMP, AF_INET)) {
action = PF_DROP;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
REASON_SET(&reason, PFRES_PROTCKSUM);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
goto done;
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
action = pf_test_state_icmp(&s, dir, kif, m, off, h, &pd,
&reason);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (action == PF_PASS) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#if NPFSYNC
pfsync_update_state(s);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* NPFSYNC */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = s->rule.ptr;
a = s->anchor.ptr;
log = s->log;
} else if (s == NULL)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#ifdef __FreeBSD__
action = pf_test_icmp(&r, &s, dir, kif,
m, off, h, &pd, &a, &ruleset, NULL);
#else
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_test_icmp(&r, &s, dir, kif,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
m, off, h, &pd, &a, &ruleset, &ipintrq);
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
}
default:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_test_state_other(&s, dir, kif, &pd);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (action == PF_PASS) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#if NPFSYNC
pfsync_update_state(s);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* NPFSYNC */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = s->rule.ptr;
a = s->anchor.ptr;
log = s->log;
} else if (s == NULL)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#ifdef __FreeBSD__
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_test_other(&r, &s, dir, kif, m, off, h,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
&pd, &a, &ruleset, NULL);
#else
action = pf_test_other(&r, &s, dir, kif, m, off, h,
&pd, &a, &ruleset, &ipintrq);
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
}
done:
if (action == PF_PASS && h->ip_hl > 5 &&
!((s && s->allow_opts) || r->allow_opts)) {
action = PF_DROP;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(&reason, PFRES_IPOPTIONS);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
log = 1;
DPFPRINTF(PF_DEBUG_MISC,
("pf: dropping packet with ip options\n"));
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if ((s && s->tag) || r->rtableid)
pf_tag_packet(m, pd.pf_mtag, s ? s->tag : 0, r->rtableid);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef ALTQ
if (action == PF_PASS && r->qid) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pqid || (pd.tos & IPTOS_LOWDELAY))
pd.pf_mtag->qid = r->pqid;
else
pd.pf_mtag->qid = r->qid;
/* add hints for ecn */
pd.pf_mtag->af = AF_INET;
pd.pf_mtag->hdr = h;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* ALTQ */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
/*
* 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) &&
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
(ntohl(pd.dst->v4.s_addr) >> IN_CLASSA_NSHIFT) == IN_LOOPBACKNET)
pd.pf_mtag->flags |= PF_TAG_TRANSLATE_LOCALHOST;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (log) {
struct pf_rule *lr;
if (s != NULL && s->nat_rule.ptr != NULL &&
s->nat_rule.ptr->log & PF_LOG_ALL)
lr = s->nat_rule.ptr;
else
lr = r;
PFLOG_PACKET(kif, h, m, AF_INET, dir, reason, lr, a, ruleset,
&pd);
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
kif->pfik_bytes[0][dir == PF_OUT][action != PF_PASS] += pd.tot_len;
kif->pfik_packets[0][dir == PF_OUT][action != PF_PASS]++;
if (action == PF_PASS || r->action == PF_DROP) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
dirndx = (dir == PF_OUT);
r->packets[dirndx]++;
r->bytes[dirndx] += pd.tot_len;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a != NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
a->packets[dirndx]++;
a->bytes[dirndx] += pd.tot_len;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
if (s != NULL) {
if (s->nat_rule.ptr != NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
s->nat_rule.ptr->packets[dirndx]++;
s->nat_rule.ptr->bytes[dirndx] += pd.tot_len;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
if (s->src_node != NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
s->src_node->packets[dirndx]++;
s->src_node->bytes[dirndx] += pd.tot_len;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
if (s->nat_src_node != NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
s->nat_src_node->packets[dirndx]++;
s->nat_src_node->bytes[dirndx] += pd.tot_len;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
dirndx = (dir == s->direction) ? 0 : 1;
s->packets[dirndx]++;
s->bytes[dirndx] += pd.tot_len;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
tr = r;
nr = (s != NULL) ? s->nat_rule.ptr : pd.nat_rule;
if (nr != NULL) {
struct pf_addr *x;
/*
* XXX: we need to make sure that the addresses
* passed to pfr_update_stats() are the same than
* the addresses used during matching (pfr_match)
*/
if (r == &pf_default_rule) {
tr = nr;
x = (s == NULL || s->direction == dir) ?
&pd.baddr : &pd.naddr;
} else
x = (s == NULL || s->direction == dir) ?
&pd.naddr : &pd.baddr;
if (x == &pd.baddr || s == NULL) {
/* we need to change the address */
if (dir == PF_OUT)
pd.src = x;
else
pd.dst = x;
}
}
if (tr->src.addr.type == PF_ADDR_TABLE)
pfr_update_stats(tr->src.addr.p.tbl, (s == NULL ||
s->direction == dir) ? pd.src : pd.dst, pd.af,
pd.tot_len, dir == PF_OUT, r->action == PF_PASS,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
tr->src.neg);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (tr->dst.addr.type == PF_ADDR_TABLE)
pfr_update_stats(tr->dst.addr.p.tbl, (s == NULL ||
s->direction == dir) ? pd.dst : pd.src, pd.af,
pd.tot_len, dir == PF_OUT, r->action == PF_PASS,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
tr->dst.neg);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (action == PF_SYNPROXY_DROP) {
m_freem(*m0);
*m0 = NULL;
action = PF_PASS;
} else if (r->rt)
/* pf_route can free the mbuf causing *m0 to become NULL */
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_route(m0, r, dir, ifp, s, &pd);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_UNLOCK();
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (action);
}
#endif /* INET */
#ifdef INET6
int
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_test6(int dir, struct ifnet *ifp, struct mbuf **m0,
struct ether_header *eh, struct inpcb *inp)
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pf_test6(int dir, struct ifnet *ifp, struct mbuf **m0,
struct ether_header *eh)
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
{
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pfi_kif *kif;
u_short action, reason = 0, log = 0;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
struct mbuf *m = *m0, *n = NULL;
struct ip6_hdr *h;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
struct pf_rule *a = NULL, *r = &pf_default_rule, *tr, *nr;
struct pf_state *s = NULL;
struct pf_ruleset *ruleset = NULL;
struct pf_pdesc pd;
From OpenBSD, rev. 1.524, 1.528, 1.529 Deal with IPv6 routing headers (see FreeBSD-SA-07:03.ipv6 for background) Block IPv6 packets with routing headers by default, unless 'allow-opts' is specified. Block RH0 unconditionally. Deal with ip6_plen 0. MFC after: 1 week Discussed with: mlaier
2007-05-21 20:08:59 +00:00
int off, terminal = 0, dirndx, rh_cnt = 0;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_LOCK();
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (!pf_status.running)
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
{
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_UNLOCK();
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (PF_PASS);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#endif
memset(&pd, 0, sizeof(pd));
if ((pd.pf_mtag = pf_get_mtag(m)) == NULL) {
#ifdef __FreeBSD__
PF_UNLOCK();
#endif
DPFPRINTF(PF_DEBUG_URGENT,
("pf_test6: pf_get_mtag returned NULL\n"));
return (PF_DROP);
}
if (pd.pf_mtag->flags & PF_TAG_GENERATED)
return (PF_PASS);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#ifdef __FreeBSD__
/* XXX_IMPORT: later */
#else
if (ifp->if_type == IFT_CARP && ifp->if_carpdev)
ifp = ifp->if_carpdev;
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
kif = (struct pfi_kif *)ifp->if_pf_kif;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (kif == NULL) {
#ifdef __FreeBSD__
PF_UNLOCK();
#endif
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
DPFPRINTF(PF_DEBUG_URGENT,
("pf_test6: kif == NULL, if_xname %s\n", ifp->if_xname));
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
return (PF_DROP);
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (kif->pfik_flags & PFI_IFLAG_SKIP) {
#ifdef __FreeBSD__
PF_UNLOCK();
#endif
return (PF_PASS);
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
M_ASSERTPKTHDR(m);
#else
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef DIAGNOSTIC
if ((m->m_flags & M_PKTHDR) == 0)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
panic("non-M_PKTHDR is passed to pf_test6");
#endif /* DIAGNOSTIC */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#endif
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
#ifdef __FreeBSD__
h = NULL; /* make the compiler happy */
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (m->m_pkthdr.len < (int)sizeof(*h)) {
action = PF_DROP;
REASON_SET(&reason, PFRES_SHORT);
log = 1;
goto done;
}
/* We do IP header normalization and packet reassembly here */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
if (pf_normalize_ip6(m0, dir, kif, &reason, &pd) != PF_PASS) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
action = PF_DROP;
goto done;
}
m = *m0;
h = mtod(m, struct ip6_hdr *);
From OpenBSD, rev. 1.524, 1.528, 1.529 Deal with IPv6 routing headers (see FreeBSD-SA-07:03.ipv6 for background) Block IPv6 packets with routing headers by default, unless 'allow-opts' is specified. Block RH0 unconditionally. Deal with ip6_plen 0. MFC after: 1 week Discussed with: mlaier
2007-05-21 20:08:59 +00:00
#if 1
/*
* we do not support jumbogram yet. if we keep going, zero ip6_plen
* will do something bad, so drop the packet for now.
*/
if (htons(h->ip6_plen) == 0) {
action = PF_DROP;
REASON_SET(&reason, PFRES_NORM); /*XXX*/
goto done;
}
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pd.src = (struct pf_addr *)&h->ip6_src;
pd.dst = (struct pf_addr *)&h->ip6_dst;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
PF_ACPY(&pd.baddr, dir == PF_OUT ? pd.src : pd.dst, AF_INET6);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
pd.ip_sum = NULL;
pd.af = AF_INET6;
pd.tos = 0;
pd.tot_len = ntohs(h->ip6_plen) + sizeof(struct ip6_hdr);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
pd.eh = eh;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
off = ((caddr_t)h - m->m_data) + sizeof(struct ip6_hdr);
pd.proto = h->ip6_nxt;
do {
switch (pd.proto) {
case IPPROTO_FRAGMENT:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_test_fragment(&r, dir, kif, m, h,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
&pd, &a, &ruleset);
if (action == PF_DROP)
REASON_SET(&reason, PFRES_FRAG);
goto done;
From OpenBSD, rev. 1.524, 1.528, 1.529 Deal with IPv6 routing headers (see FreeBSD-SA-07:03.ipv6 for background) Block IPv6 packets with routing headers by default, unless 'allow-opts' is specified. Block RH0 unconditionally. Deal with ip6_plen 0. MFC after: 1 week Discussed with: mlaier
2007-05-21 20:08:59 +00:00
case IPPROTO_ROUTING: {
struct ip6_rthdr rthdr;
if (rh_cnt++) {
DPFPRINTF(PF_DEBUG_MISC,
("pf: IPv6 more than one rthdr\n"));
action = PF_DROP;
REASON_SET(&reason, PFRES_IPOPTIONS);
log = 1;
goto done;
}
if (!pf_pull_hdr(m, off, &rthdr, sizeof(rthdr), NULL,
&reason, pd.af)) {
DPFPRINTF(PF_DEBUG_MISC,
("pf: IPv6 short rthdr\n"));
action = PF_DROP;
REASON_SET(&reason, PFRES_SHORT);
log = 1;
goto done;
}
if (rthdr.ip6r_type == IPV6_RTHDR_TYPE_0) {
DPFPRINTF(PF_DEBUG_MISC,
("pf: IPv6 rthdr0\n"));
action = PF_DROP;
REASON_SET(&reason, PFRES_IPOPTIONS);
log = 1;
goto done;
}
/* fallthrough */
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
case IPPROTO_AH:
case IPPROTO_HOPOPTS:
case IPPROTO_DSTOPTS: {
/* get next header and header length */
struct ip6_ext opt6;
if (!pf_pull_hdr(m, off, &opt6, sizeof(opt6),
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
NULL, &reason, pd.af)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
DPFPRINTF(PF_DEBUG_MISC,
("pf: IPv6 short opt\n"));
action = PF_DROP;
log = 1;
goto done;
}
if (pd.proto == IPPROTO_AH)
off += (opt6.ip6e_len + 2) * 4;
else
off += (opt6.ip6e_len + 1) * 8;
pd.proto = opt6.ip6e_nxt;
/* goto the next header */
break;
}
default:
terminal++;
break;
}
} while (!terminal);
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
/* if there's no routing header, use unmodified mbuf for checksumming */
if (!n)
n = m;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
switch (pd.proto) {
case IPPROTO_TCP: {
struct tcphdr th;
pd.hdr.tcp = &th;
if (!pf_pull_hdr(m, off, &th, sizeof(th),
&action, &reason, AF_INET6)) {
log = action != PF_PASS;
goto done;
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (dir == PF_IN && pf_check_proto_cksum(n, off,
IPv6 packets can contain headers (like options) before the TCP/UDP/ICMP6 header. pf finds the first TCP/UDP/ICMP6 header to filter by traversing the header chain. In the case where headers are skipped, the protocol checksum verification used the wrong length (included the skipped headers), leading to incorrectly mismatching checksums. Such IPv6 packets with headers were silently dropped. Discovered by: Bernhard Schmidt MFC after: 1 week
2004-12-05 12:15:43 +00:00
ntohs(h->ip6_plen) - (off - sizeof(struct ip6_hdr)),
IPPROTO_TCP, AF_INET6)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
action = PF_DROP;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(&reason, PFRES_PROTCKSUM);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
goto done;
}
pd.p_len = pd.tot_len - off - (th.th_off << 2);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_normalize_tcp(dir, kif, m, 0, off, h, &pd);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (action == PF_DROP)
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto done;
action = pf_test_state_tcp(&s, dir, kif, m, off, h, &pd,
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
&reason);
if (action == PF_PASS) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#if NPFSYNC
pfsync_update_state(s);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* NPFSYNC */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = s->rule.ptr;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
a = s->anchor.ptr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
log = s->log;
} else if (s == NULL)
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
action = pf_test_tcp(&r, &s, dir, kif,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
m, off, h, &pd, &a, &ruleset, NULL, inp);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_test_tcp(&r, &s, dir, kif,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
m, off, h, &pd, &a, &ruleset, &ip6intrq);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
}
case IPPROTO_UDP: {
struct udphdr uh;
pd.hdr.udp = &uh;
if (!pf_pull_hdr(m, off, &uh, sizeof(uh),
&action, &reason, AF_INET6)) {
log = action != PF_PASS;
goto done;
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (dir == PF_IN && uh.uh_sum && pf_check_proto_cksum(n,
IPv6 packets can contain headers (like options) before the TCP/UDP/ICMP6 header. pf finds the first TCP/UDP/ICMP6 header to filter by traversing the header chain. In the case where headers are skipped, the protocol checksum verification used the wrong length (included the skipped headers), leading to incorrectly mismatching checksums. Such IPv6 packets with headers were silently dropped. Discovered by: Bernhard Schmidt MFC after: 1 week
2004-12-05 12:15:43 +00:00
off, ntohs(h->ip6_plen) - (off - sizeof(struct ip6_hdr)),
IPPROTO_UDP, AF_INET6)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
action = PF_DROP;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(&reason, PFRES_PROTCKSUM);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
goto done;
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (uh.uh_dport == 0 ||
ntohs(uh.uh_ulen) > m->m_pkthdr.len - off ||
ntohs(uh.uh_ulen) < sizeof(struct udphdr)) {
action = PF_DROP;
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
REASON_SET(&reason, PFRES_SHORT);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
goto done;
}
action = pf_test_state_udp(&s, dir, kif, m, off, h, &pd);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (action == PF_PASS) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#if NPFSYNC
pfsync_update_state(s);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* NPFSYNC */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = s->rule.ptr;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
a = s->anchor.ptr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
log = s->log;
} else if (s == NULL)
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#ifdef __FreeBSD__
action = pf_test_udp(&r, &s, dir, kif,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
m, off, h, &pd, &a, &ruleset, NULL, inp);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#else
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_test_udp(&r, &s, dir, kif,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
m, off, h, &pd, &a, &ruleset, &ip6intrq);
Add an additional struct inpcb * argument to pfil(9) in order to enable passing along socket information. This is required to work around a LOR with the socket code which results in an easy reproducible hard lockup with debug.mpsafenet=1. This commit does *not* fix the LOR, but enables us to do so later. The missing piece is to turn the filter locking into a leaf lock and will follow in a seperate (later) commit. This will hopefully be MT5'ed in order to fix the problem for RELENG_5 in forseeable future. Suggested by: rwatson A lot of work by: csjp (he'd be even more helpful w/o mentor-reviews ;) Reviewed by: rwatson, csjp Tested by: -pf, -ipfw, LINT, csjp and myself MFC after: 3 days LOR IDs: 14 - 17 (not fixed yet)
2004-09-29 04:54:33 +00:00
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
}
case IPPROTO_ICMPV6: {
struct icmp6_hdr ih;
pd.hdr.icmp6 = &ih;
if (!pf_pull_hdr(m, off, &ih, sizeof(ih),
&action, &reason, AF_INET6)) {
log = action != PF_PASS;
goto done;
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (dir == PF_IN && pf_check_proto_cksum(n, off,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
ntohs(h->ip6_plen) - (off - sizeof(struct ip6_hdr)),
IPv6 packets can contain headers (like options) before the TCP/UDP/ICMP6 header. pf finds the first TCP/UDP/ICMP6 header to filter by traversing the header chain. In the case where headers are skipped, the protocol checksum verification used the wrong length (included the skipped headers), leading to incorrectly mismatching checksums. Such IPv6 packets with headers were silently dropped. Discovered by: Bernhard Schmidt MFC after: 1 week
2004-12-05 12:15:43 +00:00
IPPROTO_ICMPV6, AF_INET6)) {
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
action = PF_DROP;
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
REASON_SET(&reason, PFRES_PROTCKSUM);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
goto done;
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_test_state_icmp(&s, dir, kif,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
m, off, h, &pd, &reason);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (action == PF_PASS) {
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
#if NPFSYNC
pfsync_update_state(s);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* NPFSYNC */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
r = s->rule.ptr;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
a = s->anchor.ptr;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
log = s->log;
} else if (s == NULL)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#ifdef __FreeBSD__
action = pf_test_icmp(&r, &s, dir, kif,
m, off, h, &pd, &a, &ruleset, NULL);
#else
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_test_icmp(&r, &s, dir, kif,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
m, off, h, &pd, &a, &ruleset, &ip6intrq);
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
}
default:
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_test_state_other(&s, dir, kif, &pd);
if (action == PF_PASS) {
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#if NPFSYNC
pfsync_update_state(s);
#endif /* NPFSYNC */
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
r = s->rule.ptr;
a = s->anchor.ptr;
log = s->log;
} else if (s == NULL)
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#ifdef __FreeBSD__
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
action = pf_test_other(&r, &s, dir, kif, m, off, h,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
&pd, &a, &ruleset, NULL);
#else
action = pf_test_other(&r, &s, dir, kif, m, off, h,
&pd, &a, &ruleset, &ip6intrq);
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
break;
}
done:
From OpenBSD, rev. 1.524, 1.528, 1.529 Deal with IPv6 routing headers (see FreeBSD-SA-07:03.ipv6 for background) Block IPv6 packets with routing headers by default, unless 'allow-opts' is specified. Block RH0 unconditionally. Deal with ip6_plen 0. MFC after: 1 week Discussed with: mlaier
2007-05-21 20:08:59 +00:00
/* 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"));
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if ((s && s->tag) || r->rtableid)
pf_tag_packet(m, pd.pf_mtag, s ? s->tag : 0, r->rtableid);
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
#ifdef ALTQ
if (action == PF_PASS && r->qid) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (pd.tos & IPTOS_LOWDELAY)
pd.pf_mtag->qid = r->pqid;
else
pd.pf_mtag->qid = r->qid;
/* add hints for ecn */
pd.pf_mtag->af = AF_INET6;
pd.pf_mtag->hdr = h;
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
}
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
#endif /* ALTQ */
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
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) &&
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
IN6_IS_ADDR_LOOPBACK(&pd.dst->v6))
pd.pf_mtag->flags |= PF_TAG_TRANSLATE_LOCALHOST;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
if (log) {
struct pf_rule *lr;
if (s != NULL && s->nat_rule.ptr != NULL &&
s->nat_rule.ptr->log & PF_LOG_ALL)
lr = s->nat_rule.ptr;
else
lr = r;
PFLOG_PACKET(kif, h, m, AF_INET6, dir, reason, lr, a, ruleset,
&pd);
}
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
kif->pfik_bytes[1][dir == PF_OUT][action != PF_PASS] += pd.tot_len;
kif->pfik_packets[1][dir == PF_OUT][action != PF_PASS]++;
if (action == PF_PASS || r->action == PF_DROP) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
dirndx = (dir == PF_OUT);
r->packets[dirndx]++;
r->bytes[dirndx] += pd.tot_len;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (a != NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
a->packets[dirndx]++;
a->bytes[dirndx] += pd.tot_len;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
if (s != NULL) {
if (s->nat_rule.ptr != NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
s->nat_rule.ptr->packets[dirndx]++;
s->nat_rule.ptr->bytes[dirndx] += pd.tot_len;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
if (s->src_node != NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
s->src_node->packets[dirndx]++;
s->src_node->bytes[dirndx] += pd.tot_len;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
if (s->nat_src_node != NULL) {
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
s->nat_src_node->packets[dirndx]++;
s->nat_src_node->bytes[dirndx] += pd.tot_len;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
dirndx = (dir == s->direction) ? 0 : 1;
s->packets[dirndx]++;
s->bytes[dirndx] += pd.tot_len;
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
tr = r;
nr = (s != NULL) ? s->nat_rule.ptr : pd.nat_rule;
if (nr != NULL) {
struct pf_addr *x;
/*
* XXX: we need to make sure that the addresses
* passed to pfr_update_stats() are the same than
* the addresses used during matching (pfr_match)
*/
if (r == &pf_default_rule) {
tr = nr;
x = (s == NULL || s->direction == dir) ?
&pd.baddr : &pd.naddr;
} else {
x = (s == NULL || s->direction == dir) ?
&pd.naddr : &pd.baddr;
}
if (x == &pd.baddr || s == NULL) {
if (dir == PF_OUT)
pd.src = x;
else
pd.dst = x;
}
}
if (tr->src.addr.type == PF_ADDR_TABLE)
pfr_update_stats(tr->src.addr.p.tbl, (s == NULL ||
s->direction == dir) ? pd.src : pd.dst, pd.af,
pd.tot_len, dir == PF_OUT, r->action == PF_PASS,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
tr->src.neg);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
if (tr->dst.addr.type == PF_ADDR_TABLE)
pfr_update_stats(tr->dst.addr.p.tbl, (s == NULL ||
s->direction == dir) ? pd.dst : pd.src, pd.af,
pd.tot_len, dir == PF_OUT, r->action == PF_PASS,
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
tr->dst.neg);
Commit pf version 3.5 and link additional files to the kernel build. Version 3.5 brings: - Atomic commits of ruleset changes (reduce the chance of ending up in an inconsistent state). - A 30% reduction in the size of state table entries. - Source-tracking (limit number of clients and states per client). - Sticky-address (the flexibility of round-robin with the benefits of source-hash). - Significant improvements to interface handling. - and many more ...
2004-06-16 23:24:02 +00:00
}
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
if (action == PF_SYNPROXY_DROP) {
m_freem(*m0);
*m0 = NULL;
action = PF_PASS;
} else if (r->rt)
/* pf_route6 can free the mbuf causing *m0 to become NULL */
Commit resolved import of OpenBSD 4.1 pf from perforce. Approved by: re (kensmith)
2007-07-03 12:16:07 +00:00
pf_route6(m0, r, dir, ifp, s, &pd);
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
Style(9) round for the pf kernel parts. Mostly #if defined() -> #ifdef Also set HOOK_HACK to true (remove the related #ifdef's) as we have the hooks in the kernel this was missed during the merge from the port. Noticed by: Amir S. (for the HOOK_HACK part) Approved by: bms(mentor)
2004-03-17 21:11:02 +00:00
#ifdef __FreeBSD__
Bring diff from the security/pf port. This has code been tested as a port for a long time and is run in production use. This is the code present in portversion 2.03 with some additional tweaks. The rather extensive diff accounts for: - locking (to enable pf to work with a giant-free netstack) - byte order difference between OpenBSD and FreeBSD for ip_len/ip_off - conversion from pool(9) to zone(9) - api differences etc. Approved by: bms(mentor) (in general)
2004-02-26 02:34:12 +00:00
PF_UNLOCK();
#endif
Vendor import of OpenBSD's packet filter (pf) as of OpenBSD 3.4 Approved by: bms(mentor), core (in general)
2004-02-26 02:04:28 +00:00
return (action);
}
#endif /* INET6 */
Resolve conflicts created during the import of pf 3.7 Some features are missing and will be implemented in a second step. This is functional as is. Tested by: freebsd-pf, pfsense.org Obtained from: OpenBSD X-MFC after: never (breaks API/ABI)
2005-05-03 16:43:32 +00:00
int
pf_check_congestion(struct ifqueue *ifq)
{
#ifdef __FreeBSD__
/* XXX_IMPORT: later */
return (0);
#else
if (ifq->ifq_congestion)
return (1);
else
return (0);
#endif
}
Reference in New Issue Copy Permalink