NetBSD/usr.sbin/npf/npfctl/npf_build.c

1120 lines
26 KiB
C

/*-
* Copyright (c) 2011-2020 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This material is based upon work partially supported by The
* NetBSD Foundation under a contract with Mindaugas Rasiukevicius.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. 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 NETBSD FOUNDATION, INC. 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 FOUNDATION 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.
*/
/*
* npfctl(8) building of the configuration.
*/
#include <sys/cdefs.h>
__RCSID("$NetBSD: npf_build.c,v 1.56 2023/08/18 14:26:50 tnn Exp $");
#include <sys/types.h>
#define __FAVOR_BSD
#include <netinet/tcp.h>
#include <stdlib.h>
#include <inttypes.h>
#include <string.h>
#include <ctype.h>
#include <unistd.h>
#include <fcntl.h>
#include <errno.h>
#include <err.h>
#include <pcap/pcap.h>
#include "npfctl.h"
#define MAX_RULE_NESTING 16
static nl_config_t * npf_conf = NULL;
static bool npf_debug = false;
static nl_rule_t * the_rule = NULL;
static bool npf_conf_built = false;
static nl_rule_t * defgroup = NULL;
static nl_rule_t * current_group[MAX_RULE_NESTING];
static unsigned rule_nesting_level = 0;
static unsigned npfctl_tid_counter = 0;
static void npfctl_dump_bpf(struct bpf_program *);
void
npfctl_config_init(bool debug)
{
npf_conf = npf_config_create();
if (npf_conf == NULL) {
errx(EXIT_FAILURE, "npf_config_create() failed");
}
memset(current_group, 0, sizeof(current_group));
npf_debug = debug;
npf_conf_built = false;
}
nl_config_t *
npfctl_config_ref(void)
{
return npf_conf;
}
nl_rule_t *
npfctl_rule_ref(void)
{
return the_rule;
}
void
npfctl_config_build(void)
{
/* Run-once. */
if (npf_conf_built) {
return;
}
/*
* The default group is mandatory. Note: npfctl_build_group_end()
* skipped the default rule, since it must be the last one.
*/
if (!defgroup) {
errx(EXIT_FAILURE, "default group was not defined");
}
assert(rule_nesting_level == 0);
npf_rule_insert(npf_conf, NULL, defgroup);
npf_config_build(npf_conf);
npf_conf_built = true;
}
int
npfctl_config_send(int fd)
{
npf_error_t errinfo;
int error = 0;
npfctl_config_build();
error = npf_config_submit(npf_conf, fd, &errinfo);
if (error) {
npfctl_print_error(&errinfo);
}
npf_config_destroy(npf_conf);
return error;
}
void
npfctl_config_save(nl_config_t *ncf, const char *outfile)
{
void *blob;
size_t len;
int fd;
blob = npf_config_export(ncf, &len);
if (!blob) {
err(EXIT_FAILURE, "npf_config_export");
}
if ((fd = open(outfile, O_CREAT | O_TRUNC | O_WRONLY, 0644)) == -1) {
err(EXIT_FAILURE, "could not open %s", outfile);
}
if (write(fd, blob, len) != (ssize_t)len) {
err(EXIT_FAILURE, "write to %s failed", outfile);
}
free(blob);
close(fd);
}
bool
npfctl_debug_addif(const char *ifname)
{
const char tname[] = "npftest";
const size_t tnamelen = sizeof(tname) - 1;
if (npf_debug) {
_npf_debug_addif(npf_conf, ifname);
return strncmp(ifname, tname, tnamelen) == 0;
}
return 0;
}
nl_table_t *
npfctl_table_getbyname(nl_config_t *ncf, const char *name)
{
nl_iter_t i = NPF_ITER_BEGIN;
nl_table_t *tl;
/* XXX dynamic ruleset */
if (!ncf) {
return NULL;
}
while ((tl = npf_table_iterate(ncf, &i)) != NULL) {
const char *tname = npf_table_getname(tl);
if (strcmp(tname, name) == 0) {
break;
}
}
return tl;
}
unsigned
npfctl_table_getid(const char *name)
{
nl_table_t *tl;
tl = npfctl_table_getbyname(npf_conf, name);
return tl ? npf_table_getid(tl) : (unsigned)-1;
}
const char *
npfctl_table_getname(nl_config_t *ncf, unsigned tid, bool *ifaddr)
{
const char *name = NULL;
nl_iter_t i = NPF_ITER_BEGIN;
nl_table_t *tl;
while ((tl = npf_table_iterate(ncf, &i)) != NULL) {
if (npf_table_getid(tl) == tid) {
name = npf_table_getname(tl);
break;
}
}
if (!name) {
return NULL;
}
if (!strncmp(name, NPF_IFNET_TABLE_PREF, NPF_IFNET_TABLE_PREFLEN)) {
name += NPF_IFNET_TABLE_PREFLEN;
*ifaddr = true;
} else {
*ifaddr = false;
}
return name;
}
static in_port_t
npfctl_get_singleport(const npfvar_t *vp)
{
port_range_t *pr;
in_port_t *port;
if (npfvar_get_count(vp) > 1) {
yyerror("multiple ports are not valid");
}
pr = npfvar_get_data(vp, NPFVAR_PORT_RANGE, 0);
if (pr->pr_start != pr->pr_end) {
yyerror("port range is not valid");
}
port = &pr->pr_start;
return *port;
}
static fam_addr_mask_t *
npfctl_get_singlefam(const npfvar_t *vp)
{
fam_addr_mask_t *am;
if (npfvar_get_type(vp, 0) != NPFVAR_FAM) {
yyerror("map segment must be an address or network");
}
if (npfvar_get_count(vp) > 1) {
yyerror("map segment cannot have multiple static addresses");
}
am = npfvar_get_data(vp, NPFVAR_FAM, 0);
if (am == NULL) {
yyerror("invalid map segment");
}
return am;
}
static unsigned
npfctl_get_singletable(const npfvar_t *vp)
{
unsigned *tid;
if (npfvar_get_count(vp) > 1) {
yyerror("invalid use of multiple tables");
}
tid = npfvar_get_data(vp, NPFVAR_TABLE, 0);
assert(tid != NULL);
return *tid;
}
static bool
npfctl_build_fam(npf_bpf_t *ctx, sa_family_t family,
fam_addr_mask_t *fam, unsigned opts)
{
/*
* If family is specified, address does not match it and the
* address is extracted from the interface, then simply ignore.
* Otherwise, address of invalid family was passed manually.
*/
if (family != AF_UNSPEC && family != fam->fam_family) {
if (!fam->fam_ifindex) {
yyerror("specified address is not of the required "
"family %d", family);
}
return false;
}
family = fam->fam_family;
if (family != AF_INET && family != AF_INET6) {
yyerror("family %d is not supported", family);
}
/*
* Optimise 0.0.0.0/0 case to be NOP. Otherwise, address with
* zero mask would never match and therefore is not valid.
*/
if (fam->fam_mask == 0) {
if (!npfctl_addr_iszero(&fam->fam_addr)) {
yyerror("filter criterion would never match");
}
return false;
}
npfctl_bpf_cidr(ctx, opts, family, &fam->fam_addr, fam->fam_mask);
return true;
}
static void
npfctl_build_vars(npf_bpf_t *ctx, sa_family_t family, npfvar_t *vars, int opts)
{
npfctl_bpf_group_enter(ctx, (opts & MATCH_INVERT) != 0);
for (unsigned i = 0; i < npfvar_get_count(vars); i++) {
const unsigned type = npfvar_get_type(vars, i);
void *data = npfvar_get_data(vars, type, i);
assert(data != NULL);
switch (type) {
case NPFVAR_FAM: {
fam_addr_mask_t *fam = data;
npfctl_build_fam(ctx, family, fam, opts);
break;
}
case NPFVAR_PORT_RANGE: {
port_range_t *pr = data;
npfctl_bpf_ports(ctx, opts, pr->pr_start, pr->pr_end);
break;
}
case NPFVAR_TABLE: {
unsigned tid;
memcpy(&tid, data, sizeof(unsigned));
npfctl_bpf_table(ctx, opts, tid);
break;
}
default:
yyerror("unexpected %s", npfvar_type(type));
}
}
npfctl_bpf_group_exit(ctx);
}
static void
npfctl_build_proto_block(npf_bpf_t *ctx, const opt_proto_t *op, bool multiple)
{
const unsigned proto = op->op_proto;
npfvar_t *popts = op->op_opts;
if (multiple && popts) {
yyerror("multiple protocol options with protocol filters "
"are not yet supported");
}
/* Build the protocol filter. */
npfctl_bpf_proto(ctx, proto);
switch (proto) {
case IPPROTO_TCP:
/* Build TCP flags matching (optional). */
if (popts) {
uint8_t *tf, *tf_mask;
assert(npfvar_get_count(popts) == 2);
tf = npfvar_get_data(popts, NPFVAR_TCPFLAG, 0);
tf_mask = npfvar_get_data(popts, NPFVAR_TCPFLAG, 1);
npfctl_bpf_tcpfl(ctx, *tf, *tf_mask);
}
break;
case IPPROTO_ICMP:
case IPPROTO_ICMPV6:
/* Build ICMP/ICMPv6 type and/or code matching. */
if (popts) {
int *icmp_type, *icmp_code;
assert(npfvar_get_count(popts) == 2);
icmp_type = npfvar_get_data(popts, NPFVAR_ICMP, 0);
icmp_code = npfvar_get_data(popts, NPFVAR_ICMP, 1);
npfctl_bpf_icmp(ctx, *icmp_type, *icmp_code);
}
break;
default:
/* No options for other protocols. */
break;
}
}
static void
npfctl_build_proto(npf_bpf_t *ctx, const npfvar_t *vars)
{
const unsigned count = npfvar_get_count(vars);
/*
* XXX: For now, just do not support multiple protocol
* blocks with options; this is because npfctl_bpf_tcpfl()
* and npfctl_bpf_icmp() will not work correctly in a group.
*/
if (count == 1) {
const opt_proto_t *op = npfvar_get_data(vars, NPFVAR_PROTO, 0);
npfctl_build_proto_block(ctx, op, false);
return;
}
npfctl_bpf_group_enter(ctx, false);
for (unsigned i = 0; i < count; i++) {
const opt_proto_t *op = npfvar_get_data(vars, NPFVAR_PROTO, i);
npfctl_build_proto_block(ctx, op, true);
}
npfctl_bpf_group_exit(ctx);
}
static bool
npfctl_check_proto(const npfvar_t *vars, bool *non_tcpudp, bool *tcp_with_nofl)
{
unsigned count;
*non_tcpudp = false;
*tcp_with_nofl = false;
if (vars == NULL) {
return false;
}
count = npfvar_get_count(vars);
for (unsigned i = 0; i < count; i++) {
const opt_proto_t *op = npfvar_get_data(vars, NPFVAR_PROTO, i);
switch (op->op_proto) {
case IPPROTO_TCP:
*tcp_with_nofl = op->op_opts == NULL;
break;
case IPPROTO_UDP:
case -1:
break;
default:
*non_tcpudp = true;
break;
}
}
return count != 0;
}
static bool
npfctl_build_code(nl_rule_t *rl, sa_family_t family, const npfvar_t *popts,
const filt_opts_t *fopts)
{
const addr_port_t *apfrom = &fopts->fo_from;
const addr_port_t *apto = &fopts->fo_to;
bool any_proto, any_addrs, any_ports, stateful;
bool any_l4proto, non_tcpudp, tcp_with_nofl;
npf_bpf_t *bc;
unsigned opts;
size_t len;
/*
* Gather some information about the protocol options, if any.
* Check the filter criteria in general -- if none specified,
* then no byte-code.
*/
any_l4proto = npfctl_check_proto(popts, &non_tcpudp, &tcp_with_nofl);
any_proto = (family != AF_UNSPEC) || any_l4proto;
any_addrs = apfrom->ap_netaddr || apto->ap_netaddr;
any_ports = apfrom->ap_portrange || apto->ap_portrange;
stateful = (npf_rule_getattr(rl) & NPF_RULE_STATEFUL) != 0;
if (!any_proto && !any_addrs && !any_ports && !stateful) {
return false;
}
/*
* Sanity check: ports can only be used with TCP or UDP protocol.
*/
if (any_ports && non_tcpudp) {
yyerror("invalid filter options for given the protocol(s)");
}
bc = npfctl_bpf_create();
/* Build layer 3 and 4 protocol blocks. */
if (family != AF_UNSPEC) {
npfctl_bpf_ipver(bc, family);
}
if (any_l4proto) {
npfctl_build_proto(bc, popts);
}
/*
* If this is a stateful rule and TCP flags are not specified,
* then add "flags S/SAFR" filter for TCP protocol case.
*/
if (stateful && (!any_l4proto || tcp_with_nofl)) {
npfctl_bpf_tcpfl(bc, TH_SYN, TH_SYN | TH_ACK | TH_FIN | TH_RST);
}
/* Build IP address blocks. */
opts = MATCH_SRC | (fopts->fo_finvert ? MATCH_INVERT : 0);
npfctl_build_vars(bc, family, apfrom->ap_netaddr, opts);
opts = MATCH_DST | (fopts->fo_tinvert ? MATCH_INVERT : 0);
npfctl_build_vars(bc, family, apto->ap_netaddr, opts);
/*
* Build the port-range blocks. If no protocol is specified,
* then we implicitly filter for the TCP / UDP protocols.
*/
if (any_ports && !any_l4proto) {
npfctl_bpf_group_enter(bc, false);
npfctl_bpf_proto(bc, IPPROTO_TCP);
npfctl_bpf_proto(bc, IPPROTO_UDP);
npfctl_bpf_group_exit(bc);
}
npfctl_build_vars(bc, family, apfrom->ap_portrange, MATCH_SRC);
npfctl_build_vars(bc, family, apto->ap_portrange, MATCH_DST);
/* Set the byte-code marks, if any. */
const void *bmarks = npfctl_bpf_bmarks(bc, &len);
if (bmarks && npf_rule_setinfo(rl, bmarks, len) != 0) {
errx(EXIT_FAILURE, "npf_rule_setinfo");
}
/* Complete BPF byte-code and pass to the rule. */
struct bpf_program *bf = npfctl_bpf_complete(bc);
if (bf == NULL) {
npfctl_bpf_destroy(bc);
return true;
}
len = bf->bf_len * sizeof(struct bpf_insn);
if (npf_rule_setcode(rl, NPF_CODE_BPF, bf->bf_insns, len) != 0) {
errx(EXIT_FAILURE, "npf_rule_setcode");
}
npfctl_dump_bpf(bf);
npfctl_bpf_destroy(bc);
return true;
}
static void
npfctl_build_pcap(nl_rule_t *rl, const char *filter)
{
const size_t maxsnaplen = 64 * 1024;
struct bpf_program bf;
size_t len;
pcap_t *pd;
pd = pcap_open_dead(DLT_RAW, maxsnaplen);
if (pd == NULL) {
err(EXIT_FAILURE, "pcap_open_dead");
}
if (pcap_compile(pd, &bf,
filter, 1, PCAP_NETMASK_UNKNOWN) == -1) {
yyerror("invalid pcap-filter(7) syntax");
}
len = bf.bf_len * sizeof(struct bpf_insn);
if (npf_rule_setcode(rl, NPF_CODE_BPF, bf.bf_insns, len) != 0) {
errx(EXIT_FAILURE, "npf_rule_setcode failed");
}
npfctl_dump_bpf(&bf);
pcap_freecode(&bf);
pcap_close(pd);
}
static void
npfctl_build_rpcall(nl_rproc_t *rp, const char *name, npfvar_t *args)
{
npf_extmod_t *extmod;
nl_ext_t *extcall;
int error;
extmod = npf_extmod_get(name, &extcall);
if (extmod == NULL) {
yyerror("unknown rule procedure '%s'", name);
}
for (size_t i = 0; i < npfvar_get_count(args); i++) {
const char *param, *value;
proc_param_t *p;
p = npfvar_get_data(args, NPFVAR_PROC_PARAM, i);
param = p->pp_param;
value = p->pp_value;
error = npf_extmod_param(extmod, extcall, param, value);
switch (error) {
case EINVAL:
yyerror("invalid parameter '%s'", param);
default:
break;
}
}
error = npf_rproc_extcall(rp, extcall);
if (error) {
yyerror(error == EEXIST ?
"duplicate procedure call" : "unexpected error");
}
}
/*
* npfctl_build_rproc: create and insert a rule procedure.
*/
void
npfctl_build_rproc(const char *name, npfvar_t *procs)
{
nl_rproc_t *rp;
size_t i;
rp = npf_rproc_create(name);
if (rp == NULL) {
errx(EXIT_FAILURE, "%s failed", __func__);
}
for (i = 0; i < npfvar_get_count(procs); i++) {
proc_call_t *pc = npfvar_get_data(procs, NPFVAR_PROC, i);
npfctl_build_rpcall(rp, pc->pc_name, pc->pc_opts);
}
npf_rproc_insert(npf_conf, rp);
}
/*
* npfctl_build_maprset: create and insert a NAT ruleset.
*/
void
npfctl_build_maprset(const char *name, int attr, const char *ifname)
{
const int attr_di = (NPF_RULE_IN | NPF_RULE_OUT);
nl_rule_t *rl;
bool natset;
int err;
/* Validate the prefix. */
err = npfctl_nat_ruleset_p(name, &natset);
if (!natset) {
yyerror("NAT ruleset names must be prefixed with `"
NPF_RULESET_MAP_PREF "`");
}
if (err) {
yyerror("NAT ruleset is missing a name (only prefix found)");
}
/* If no direction is not specified, then both. */
if ((attr & attr_di) == 0) {
attr |= attr_di;
}
/* Allow only "in/out" attributes. */
attr = NPF_RULE_GROUP | NPF_RULE_DYNAMIC | (attr & attr_di);
rl = npf_rule_create(name, attr, ifname);
npf_rule_setprio(rl, NPF_PRI_LAST);
npf_nat_insert(npf_conf, rl);
}
/*
* npfctl_build_group: create a group, update the current group pointer
* and increase the nesting level.
*/
void
npfctl_build_group(const char *name, int attr, const char *ifname, bool def)
{
const int attr_di = (NPF_RULE_IN | NPF_RULE_OUT);
nl_rule_t *rl;
if (def || (attr & attr_di) == 0) {
attr |= attr_di;
}
rl = npf_rule_create(name, attr | NPF_RULE_GROUP, ifname);
npf_rule_setprio(rl, NPF_PRI_LAST);
if (def) {
if (defgroup) {
yyerror("multiple default groups are not valid");
}
if (rule_nesting_level) {
yyerror("default group can only be at the top level");
}
defgroup = rl;
}
/* Set the current group and increase the nesting level. */
if (rule_nesting_level >= MAX_RULE_NESTING) {
yyerror("rule nesting limit reached");
}
current_group[++rule_nesting_level] = rl;
}
void
npfctl_build_group_end(void)
{
nl_rule_t *parent, *group;
assert(rule_nesting_level > 0);
parent = current_group[rule_nesting_level - 1];
group = current_group[rule_nesting_level];
current_group[rule_nesting_level--] = NULL;
/*
* Note:
* - If the parent is NULL, then it is a global rule.
* - The default rule must be the last, so it is inserted later.
*/
if (group == defgroup) {
assert(parent == NULL);
return;
}
npf_rule_insert(npf_conf, parent, group);
}
/*
* npfctl_build_rule: create a rule, build byte-code from filter options,
* if any, and insert into the ruleset of current group, or set the rule.
*/
void
npfctl_build_rule(uint32_t attr, const char *ifname, sa_family_t family,
const npfvar_t *popts, const filt_opts_t *fopts,
const char *pcap_filter, const char *rproc)
{
nl_rule_t *rl;
attr |= (npf_conf ? 0 : NPF_RULE_DYNAMIC);
rl = npf_rule_create(NULL, attr, ifname);
if (pcap_filter) {
npfctl_build_pcap(rl, pcap_filter);
} else {
npfctl_build_code(rl, family, popts, fopts);
}
if (rproc) {
npf_rule_setproc(rl, rproc);
}
if (npf_conf) {
nl_rule_t *cg = current_group[rule_nesting_level];
if (rproc && !npf_rproc_exists_p(npf_conf, rproc)) {
yyerror("rule procedure '%s' is not defined", rproc);
}
assert(cg != NULL);
npf_rule_setprio(rl, NPF_PRI_LAST);
npf_rule_insert(npf_conf, cg, rl);
} else {
/* We have parsed a single rule - set it. */
the_rule = rl;
}
}
/*
* npfctl_build_nat: create a single NAT policy of a specified
* type with a given filter options.
*/
static nl_nat_t *
npfctl_build_nat(int type, const char *ifname, const addr_port_t *ap,
const npfvar_t *popts, const filt_opts_t *fopts, unsigned flags)
{
fam_addr_mask_t *am;
sa_family_t family;
in_port_t port;
nl_nat_t *nat;
unsigned tid;
if (ap->ap_portrange) {
/*
* The port forwarding case. In such case, there has to
* be a single port used for translation; we keep the port
* translation on, but disable the port map.
*/
port = npfctl_get_singleport(ap->ap_portrange);
flags = (flags & ~NPF_NAT_PORTMAP) | NPF_NAT_PORTS;
} else {
port = 0;
}
nat = npf_nat_create(type, flags, ifname);
switch (npfvar_get_type(ap->ap_netaddr, 0)) {
case NPFVAR_FAM:
/* Translation address. */
am = npfctl_get_singlefam(ap->ap_netaddr);
family = am->fam_family;
npf_nat_setaddr(nat, family, &am->fam_addr, am->fam_mask);
break;
case NPFVAR_TABLE:
/* Translation table. */
family = AF_UNSPEC;
tid = npfctl_get_singletable(ap->ap_netaddr);
npf_nat_settable(nat, tid);
break;
default:
yyerror("map must have a valid translation address");
abort();
}
npf_nat_setport(nat, port);
npfctl_build_code(nat, family, popts, fopts);
return nat;
}
static void
npfctl_dnat_check(const addr_port_t *ap, const unsigned algo)
{
const unsigned type = npfvar_get_type(ap->ap_netaddr, 0);
fam_addr_mask_t *am;
switch (algo) {
case NPF_ALGO_NETMAP:
if (type == NPFVAR_FAM) {
break;
}
yyerror("translation address using NETMAP must be "
"a network and not a dynamic pool");
break;
case NPF_ALGO_IPHASH:
case NPF_ALGO_RR:
case NPF_ALGO_NONE:
if (type != NPFVAR_FAM) {
break;
}
am = npfctl_get_singlefam(ap->ap_netaddr);
if (am->fam_mask == NPF_NO_NETMASK) {
break;
}
yyerror("translation address, given the specified algorithm, "
"must be a pool or a single address");
break;
default:
yyerror("invalid algorithm specified for dynamic NAT");
}
}
/*
* npfctl_build_natseg: validate and create NAT policies.
*/
void
npfctl_build_natseg(int sd, int type, unsigned mflags, const char *ifname,
const addr_port_t *ap1, const addr_port_t *ap2, const npfvar_t *popts,
const filt_opts_t *fopts, unsigned algo)
{
fam_addr_mask_t *am1 = NULL, *am2 = NULL;
nl_nat_t *nt1 = NULL, *nt2 = NULL;
filt_opts_t imfopts;
uint16_t adj = 0;
unsigned flags;
bool binat;
assert(ifname != NULL);
/*
* Validate that mapping has the translation address(es) set.
*/
if ((type & NPF_NATIN) != 0 && ap1->ap_netaddr == NULL) {
yyerror("inbound network segment is not specified");
}
if ((type & NPF_NATOUT) != 0 && ap2->ap_netaddr == NULL) {
yyerror("outbound network segment is not specified");
}
/*
* Bi-directional NAT is a combination of inbound NAT and outbound
* NAT policies with the translation segments inverted respectively.
*/
binat = (NPF_NATIN | NPF_NATOUT) == type;
switch (sd) {
case NPFCTL_NAT_DYNAMIC:
/*
* Dynamic NAT: stateful translation -- traditional NAPT
* is expected. Unless it is bi-directional NAT, perform
* the port mapping.
*/
flags = !binat ? (NPF_NAT_PORTS | NPF_NAT_PORTMAP) : 0;
if (type & NPF_NATIN) {
npfctl_dnat_check(ap1, algo);
}
if (type & NPF_NATOUT) {
npfctl_dnat_check(ap2, algo);
}
break;
case NPFCTL_NAT_STATIC:
/*
* Static NAT: stateless translation.
*/
flags = NPF_NAT_STATIC;
/* Note: translation address/network cannot be a table. */
if (type & NPF_NATIN) {
am1 = npfctl_get_singlefam(ap1->ap_netaddr);
}
if (type & NPF_NATOUT) {
am2 = npfctl_get_singlefam(ap2->ap_netaddr);
}
/* Validate the algorithm. */
switch (algo) {
case NPF_ALGO_NPT66:
if (!binat || am1->fam_mask != am2->fam_mask) {
yyerror("asymmetric NPTv6 is not supported");
}
adj = npfctl_npt66_calcadj(am1->fam_mask,
&am1->fam_addr, &am2->fam_addr);
break;
case NPF_ALGO_NETMAP:
if (binat && am1->fam_mask != am2->fam_mask) {
yyerror("net-to-net mapping using the "
"NETMAP algorithm must be 1:1");
}
break;
case NPF_ALGO_NONE:
if ((am1 && am1->fam_mask != NPF_NO_NETMASK) ||
(am2 && am2->fam_mask != NPF_NO_NETMASK)) {
yyerror("static net-to-net translation "
"must have an algorithm specified");
}
break;
default:
yyerror("invalid algorithm specified for static NAT");
}
break;
default:
abort();
}
/*
* Apply the flag modifications.
*/
if (mflags & NPF_NAT_PORTS) {
flags &= ~(NPF_NAT_PORTS | NPF_NAT_PORTMAP);
}
/*
* If the filter criteria is not specified explicitly, apply implicit
* filtering according to the given network segments.
*
* Note: filled below, depending on the type.
*/
if (__predict_true(!fopts)) {
fopts = &imfopts;
}
if (type & NPF_NATIN) {
memset(&imfopts, 0, sizeof(filt_opts_t));
memcpy(&imfopts.fo_to, ap2, sizeof(addr_port_t));
nt1 = npfctl_build_nat(NPF_NATIN, ifname,
ap1, popts, fopts, flags);
}
if (type & NPF_NATOUT) {
memset(&imfopts, 0, sizeof(filt_opts_t));
memcpy(&imfopts.fo_from, ap1, sizeof(addr_port_t));
nt2 = npfctl_build_nat(NPF_NATOUT, ifname,
ap2, popts, fopts, flags);
}
switch (algo) {
case NPF_ALGO_NONE:
break;
case NPF_ALGO_NPT66:
/*
* NPTv6 is a special case using special adjustment value.
* It is always bidirectional NAT.
*/
assert(nt1 && nt2);
npf_nat_setnpt66(nt1, ~adj);
npf_nat_setnpt66(nt2, adj);
break;
default:
/*
* Set the algorithm.
*/
if (nt1) {
npf_nat_setalgo(nt1, algo);
}
if (nt2) {
npf_nat_setalgo(nt2, algo);
}
}
if (npf_conf) {
if (nt1) {
npf_rule_setprio(nt1, NPF_PRI_LAST);
npf_nat_insert(npf_conf, nt1);
}
if (nt2) {
npf_rule_setprio(nt2, NPF_PRI_LAST);
npf_nat_insert(npf_conf, nt2);
}
} else {
// XXX/TODO: need to refactor a bit to enable this..
if (nt1 && nt2) {
errx(EXIT_FAILURE, "bidirectional NAT is currently "
"not yet supported in the dynamic rules");
}
the_rule = nt1 ? nt1 : nt2;
}
}
/*
* npfctl_fill_table: fill NPF table with entries from a specified file.
*/
static void
npfctl_fill_table(nl_table_t *tl, unsigned type, const char *fname, FILE *fp)
{
char *buf = NULL;
int l = 0;
size_t n;
if (fp == NULL && (fp = fopen(fname, "r")) == NULL) {
err(EXIT_FAILURE, "open '%s'", fname);
}
while (l++, getline(&buf, &n, fp) != -1) {
fam_addr_mask_t fam;
int alen;
if (*buf == '\n' || *buf == '#') {
continue;
}
if (!npfctl_parse_cidr(buf, &fam, &alen)) {
errx(EXIT_FAILURE,
"%s:%d: invalid table entry", fname, l);
}
if (type != NPF_TABLE_LPM && fam.fam_mask != NPF_NO_NETMASK) {
errx(EXIT_FAILURE, "%s:%d: mask used with the "
"table type other than \"lpm\"", fname, l);
}
npf_table_add_entry(tl, fam.fam_family,
&fam.fam_addr, fam.fam_mask);
}
free(buf);
}
/*
* npfctl_load_table: create an NPF table and fill with contents from a file.
*/
nl_table_t *
npfctl_load_table(const char *tname, int tid, unsigned type,
const char *fname, FILE *fp)
{
nl_table_t *tl;
tl = npf_table_create(tname, tid, type);
if (tl && fname) {
npfctl_fill_table(tl, type, fname, fp);
}
return tl;
}
/*
* npfctl_build_table: create an NPF table, add to the configuration and,
* if required, fill with contents from a file.
*/
void
npfctl_build_table(const char *tname, unsigned type, const char *fname)
{
nl_table_t *tl;
if (type == NPF_TABLE_CONST && !fname) {
yyerror("table type 'const' must be loaded from a file");
}
tl = npfctl_load_table(tname, npfctl_tid_counter++, type, fname, NULL);
assert(tl != NULL);
if (npf_table_insert(npf_conf, tl)) {
yyerror("table '%s' is already defined", tname);
}
}
/*
* npfctl_ifnet_table: get a variable with ifaddr-table; auto-create
* the table on first reference.
*/
npfvar_t *
npfctl_ifnet_table(const char *ifname)
{
char tname[NPF_TABLE_MAXNAMELEN];
nl_table_t *tl;
unsigned tid;
snprintf(tname, sizeof(tname), NPF_IFNET_TABLE_PREF "%s", ifname);
if (!npf_conf) {
errx(EXIT_FAILURE, "expression `ifaddrs(%s)` is currently "
"not yet supported in dynamic rules", ifname);
}
tid = npfctl_table_getid(tname);
if (tid == (unsigned)-1) {
tid = npfctl_tid_counter++;
tl = npf_table_create(tname, tid, NPF_TABLE_IFADDR);
(void)npf_table_insert(npf_conf, tl);
}
return npfvar_create_element(NPFVAR_TABLE, &tid, sizeof(unsigned));
}
/*
* npfctl_build_alg: create an NPF application level gateway and add it
* to the configuration.
*/
void
npfctl_build_alg(const char *al_name)
{
if (npf_alg_load(npf_conf, al_name) != 0) {
yyerror("ALG '%s' is already loaded", al_name);
}
}
void
npfctl_setparam(const char *name, int val)
{
if (strcmp(name, "bpf.jit") == 0) {
npfctl_bpfjit(val != 0);
return;
}
if (npf_param_set(npf_conf, name, val) != 0) {
yyerror("invalid parameter `%s` or its value", name);
}
}
static void
npfctl_dump_bpf(struct bpf_program *bf)
{
if (npf_debug) {
extern char *yytext;
extern int yylineno;
int rule_line = yylineno - (int)(*yytext == '\n');
printf("\nRULE AT LINE %d\n", rule_line);
bpf_dump(bf, 0);
}
}