freebsd-dev/sbin/ipfw/ipfw.c
Kelly Yancey ca6e3cb0e4 Make preprocessor support more generic by passing all command-line options
after -p except for the last (the ruleset file to process) to the
preprocessor for interpretation.  This allows command-line options besides
-U and -D to be passed to cpp(1) and m4(1) as well as making it easier to
use other preprocessors.

Sponsored By:	NTT Multimedia Communications Labs
MFC after:	1 week
2002-12-23 20:08:21 +00:00

2668 lines
65 KiB
C

/*
* Copyright (c) 1996 Alex Nash, Paul Traina, Poul-Henning Kamp
* Copyright (c) 1994 Ugen J.S.Antsilevich
*
* Idea and grammar partially left from:
* Copyright (c) 1993 Daniel Boulet
*
* Redistribution and use in source forms, with and without modification,
* are permitted provided that this entire comment appears intact.
*
* Redistribution in binary form may occur without any restrictions.
* Obviously, it would be nice if you gave credit where credit is due
* but requiring it would be too onerous.
*
* This software is provided ``AS IS'' without any warranties of any kind.
*
* NEW command line interface for IP firewall facility
*
*/
#ifndef lint
static const char rcsid[] =
"$FreeBSD$";
#endif /* not lint */
#include <sys/param.h>
#include <sys/mbuf.h>
#include <sys/socket.h>
#include <sys/sockio.h>
#include <sys/sysctl.h>
#include <sys/time.h>
#include <sys/wait.h>
#include <ctype.h>
#include <err.h>
#include <errno.h>
#include <grp.h>
#include <limits.h>
#include <netdb.h>
#include <pwd.h>
#include <signal.h>
#include <stdio.h>
#include <stdlib.h>
#include <stdarg.h>
#include <string.h>
#include <unistd.h>
#include <sysexits.h>
#include <net/if.h>
#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/ip_icmp.h>
#include <netinet/ip_fw.h>
#include <net/route.h> /* def. of struct route */
#include <netinet/ip_dummynet.h>
#include <netinet/tcp.h>
#include <arpa/inet.h>
int s, /* main RAW socket */
do_resolv, /* Would try to resolve all */
do_acct, /* Show packet/byte count */
do_time, /* Show time stamps */
do_quiet, /* Be quiet in add and flush */
do_force, /* Don't ask for confirmation */
do_pipe, /* this cmd refers to a pipe */
do_sort, /* field to sort results (0 = no) */
do_dynamic, /* display dynamic rules */
do_expired, /* display expired dynamic rules */
verbose;
struct icmpcode {
int code;
char *str;
};
static struct icmpcode icmpcodes[] = {
{ ICMP_UNREACH_NET, "net" },
{ ICMP_UNREACH_HOST, "host" },
{ ICMP_UNREACH_PROTOCOL, "protocol" },
{ ICMP_UNREACH_PORT, "port" },
{ ICMP_UNREACH_NEEDFRAG, "needfrag" },
{ ICMP_UNREACH_SRCFAIL, "srcfail" },
{ ICMP_UNREACH_NET_UNKNOWN, "net-unknown" },
{ ICMP_UNREACH_HOST_UNKNOWN, "host-unknown" },
{ ICMP_UNREACH_ISOLATED, "isolated" },
{ ICMP_UNREACH_NET_PROHIB, "net-prohib" },
{ ICMP_UNREACH_HOST_PROHIB, "host-prohib" },
{ ICMP_UNREACH_TOSNET, "tosnet" },
{ ICMP_UNREACH_TOSHOST, "toshost" },
{ ICMP_UNREACH_FILTER_PROHIB, "filter-prohib" },
{ ICMP_UNREACH_HOST_PRECEDENCE, "host-precedence" },
{ ICMP_UNREACH_PRECEDENCE_CUTOFF, "precedence-cutoff" },
{ 0, NULL }
};
/*
* structure to hold flag names and associated values to be
* set in the appropriate masks.
* The last element has value=0 and the string is the error message.
*/
struct _flaglist {
char * name;
u_char value;
};
static struct _flaglist f_tcpflags[] = {
{ "syn", TH_SYN },
{ "fin", TH_FIN },
{ "ack", TH_ACK },
{ "psh", TH_PUSH },
{ "rst", TH_RST },
{ "urg", TH_URG },
{ "tcp flag", 0 }
};
static struct _flaglist f_tcpopts[] = {
{ "mss", IP_FW_TCPOPT_MSS },
{ "window", IP_FW_TCPOPT_WINDOW },
{ "sack", IP_FW_TCPOPT_SACK },
{ "ts", IP_FW_TCPOPT_TS },
{ "cc", IP_FW_TCPOPT_CC },
{ "tcp option", 0 }
};
static struct _flaglist f_ipopts[] = {
{ "ssrr", IP_FW_IPOPT_SSRR},
{ "lsrr", IP_FW_IPOPT_LSRR},
{ "rr", IP_FW_IPOPT_RR},
{ "ts", IP_FW_IPOPT_TS},
{ "ip option", 0 }
};
static struct _flaglist f_iptos[] = {
{ "lowdelay", IPTOS_LOWDELAY},
{ "throughput", IPTOS_THROUGHPUT},
{ "reliability", IPTOS_RELIABILITY},
{ "mincost", IPTOS_MINCOST},
{ "congestion", IPTOS_CE},
#if 0 /* conflicting */
{ "ecntransport", IPTOS_ECT},
#endif
{ "ip tos option", 0},
};
/**
* _s_x holds a string-int pair for various lookups. Same as _flaglist.
*/
struct _s_x {
char *s;
int x;
};
static struct _s_x limit_masks[] = {
{"src-addr", DYN_SRC_ADDR},
{"src-port", DYN_SRC_PORT},
{"dst-addr", DYN_DST_ADDR},
{"dst-port", DYN_DST_PORT},
{NULL, 0} };
static struct _s_x ether_types[] = {
/*
* Note, we cannot use "-:&/" in the names because they are field
* separators in the type specifications. Also, we use s = NULL as
* end-delimiter, because a type of 0 can be legal.
*/
{ "ip", 0x0800 },
{ "ipv4", 0x0800 },
{ "ipv6", 0x86dd },
{ "arp", 0x0806 },
{ "rarp", 0x8035 },
{ "vlan", 0x8100 },
{ "loop", 0x9000 },
{ "trail", 0x1000 },
{ "at", 0x809b },
{ "atalk", 0x809b },
{ "aarp", 0x80f3 },
{ "pppoe_disc", 0x8863 },
{ "pppoe_sess", 0x8864 },
{ "ipx_8022", 0x00E0 },
{ "ipx_8023", 0x0000 },
{ "ipx_ii", 0x8137 },
{ "ipx_snap", 0x8137 },
{ "ipx", 0x8137 },
{ "ns", 0x0600 },
{ NULL, 0 }
};
static void show_usage(void);
/*
* print the arrays of ports. The first two entries can be
* a range (a-b) or a port:mask pair, and they are processed
* accordingly if one of the two arguments range,mask is set.
* (they are not supposed to be both set).
*/
static void
print_ports(u_char prot, int n, u_short *ports, int range, int mask)
{
int i=0;
char comma = ' ';
if (mask) {
printf(" %04x:%04x", ports[0], ports[1]);
i=2;
comma = ',';
}
for (; i < n; i++) {
struct servent *se = NULL;
if (do_resolv) {
struct protoent *pe = getprotobynumber(prot);
se = getservbyport(htons(ports[i]),
pe ? pe->p_name : NULL);
}
if (se)
printf("%c%s", comma, se->s_name);
else
printf("%c%d", comma, ports[i]);
if (i == 0 && range)
comma = '-';
else
comma = ',';
}
}
static void
print_iface(char *key, union ip_fw_if *un, int byname)
{
char ifnb[FW_IFNLEN+1];
if (byname) {
strncpy(ifnb, un->fu_via_if.name, FW_IFNLEN);
ifnb[FW_IFNLEN] = '\0';
if (un->fu_via_if.unit == -1)
printf(" %s %s*", key, ifnb);
else
printf(" %s %s%d", key, ifnb, un->fu_via_if.unit);
} else if (un->fu_via_ip.s_addr != 0) {
printf(" %s %s", key, inet_ntoa(un->fu_via_ip));
} else
printf(" %s any", key);
}
static void
print_reject_code(int code)
{
struct icmpcode *ic;
for (ic = icmpcodes; ic->str; ic++)
if (ic->code == code) {
printf("%s", ic->str);
return;
}
printf("%u", code);
}
/*
* Returns the number of bits set (from left) in a contiguous bitmask,
* or -1 if the mask is not contiguous.
* This effectively works on masks in big-endian (network) format.
* First bit is bit 7 of the first byte -- note, for MAC addresses,
* the first bit on the wire is bit 0 of the first byte.
* len is the max length in bits.
*/
static int
contigmask(u_char *p, int len)
{
int i, n;
for (i=0; i<len ; i++)
if ( (p[i/8] & (1 << (7 - (i%8)))) == 0) /* first bit unset */
break;
for (n=i+1; n < len; n++)
if ( (p[n/8] & (1 << (7 - (n%8)))) != 0) /* mask is not contiguous */
return -1;
return i;
}
/*
* print options set/clear in the two bitmasks passed as parameters.
*/
static void
printopts(char *name, u_char set, u_char clear, struct _flaglist *list)
{
char *comma="";
int i;
printf(" %s ", name);
for (i=0; list[i].value != 0; i++) {
if (set & list[i].value) {
printf("%s%s", comma, list[i].name);
comma = ",";
}
if (clear & list[i].value) {
printf("%s!%s", comma, list[i].name);
comma = ",";
}
}
}
static void
print_ip(struct in_addr addr, struct in_addr mask)
{
struct hostent *he = NULL;
int mb = contigmask((u_char *)&mask, 32);
if (mb == 32 && do_resolv)
he = gethostbyaddr((char *)&(addr.s_addr),
sizeof(u_long), AF_INET);
if (he != NULL) /* resolved to name */
printf("%s", he->h_name);
else if (mb == 0) /* any */
printf("any");
else { /* numeric IP followed by some kind of mask */
printf("%s", inet_ntoa(addr));
if (mb < 0)
printf(":%s", inet_ntoa(mask));
else if (mb < 32)
printf("/%d", mb);
}
}
/*
* prints a MAC address/mask pair
*/
static void
print_mac(u_char *addr, u_char *mask)
{
int l = contigmask(mask, 48);
if (l == 0)
printf(" any");
else {
printf(" %02x:%02x:%02x:%02x:%02x:%02x",
addr[0], addr[1], addr[2], addr[3], addr[4], addr[5]);
if (l == -1)
printf("&%02x:%02x:%02x:%02x:%02x:%02x",
mask[0], mask[1], mask[2],
mask[3], mask[4], mask[5]);
else if (l < 48)
printf("/%d", l);
}
}
static void
show_ipfw(struct ip_fw *chain)
{
struct protoent *pe;
int nsp = IP_FW_GETNSRCP(chain);
int ndp = IP_FW_GETNDSTP(chain);
if (do_resolv)
setservent(1/*stay open*/);
printf("%05u ", chain->fw_number);
if (do_acct)
printf("%10qu %10qu ", chain->fw_pcnt, chain->fw_bcnt);
if (do_time) {
if (chain->timestamp) {
char timestr[30];
time_t t = _long_to_time(chain->timestamp);
strcpy(timestr, ctime(&t));
*strchr(timestr, '\n') = '\0';
printf("%s ", timestr);
} else {
printf(" ");
}
}
if (chain->fw_flg == IP_FW_F_CHECK_S) {
printf("check-state\n");
goto done;
}
if (chain->fw_flg & IP_FW_F_RND_MATCH) {
double d = 1.0 * chain->dont_match_prob;
d = 1 - (d / 0x7fffffff);
printf("prob %f ", d);
}
switch (chain->fw_flg & IP_FW_F_COMMAND) {
case IP_FW_F_ACCEPT:
printf("allow");
break;
case IP_FW_F_DENY:
printf("deny");
break;
case IP_FW_F_COUNT:
printf("count");
break;
case IP_FW_F_DIVERT:
printf("divert %u", chain->fw_divert_port);
break;
case IP_FW_F_TEE:
printf("tee %u", chain->fw_divert_port);
break;
case IP_FW_F_SKIPTO:
printf("skipto %u", chain->fw_skipto_rule);
break;
case IP_FW_F_PIPE:
printf("pipe %u", chain->fw_skipto_rule);
break;
case IP_FW_F_QUEUE:
printf("queue %u", chain->fw_skipto_rule);
break;
case IP_FW_F_REJECT:
if (chain->fw_reject_code == IP_FW_REJECT_RST)
printf("reset");
else {
printf("unreach ");
print_reject_code(chain->fw_reject_code);
}
break;
case IP_FW_F_FWD:
printf("fwd %s", inet_ntoa(chain->fw_fwd_ip.sin_addr));
if (chain->fw_fwd_ip.sin_port)
printf(",%d", chain->fw_fwd_ip.sin_port);
break;
default:
errx(EX_OSERR, "impossible");
}
if (chain->fw_flg & IP_FW_F_PRN) {
printf(" log");
if (chain->fw_logamount)
printf(" logamount %d", chain->fw_logamount);
}
if (chain->fw_flg & IP_FW_F_MAC) {
u_char *addr, *mask;
u_short *type, *typemask;
int l;
printf(" MAC");
addr = (u_char *)&(chain->fw_mac_hdr);
mask = (u_char *)&(chain->fw_mac_mask);
print_mac(addr, mask); /* destination */
addr += 6;
mask += 6;
print_mac(addr, mask); /* source */
type = (u_short *)&(chain->fw_mac_type);
typemask = (u_short *)&(chain->fw_mac_mask_type);
/* type is in net format for all cases but range */
if (chain->fw_flg & IP_FW_F_SRNG)
printf(" %04x-%04x", *type, *typemask);
else if (ntohs(*typemask) == 0)
printf(" any");
else if (ntohs(*typemask) != 0xffff)
printf(" %04x&%04x", ntohs(*type), ntohs(*typemask));
else {
struct _s_x *p = NULL;
u_int16_t i = ntohs(*type);
if (do_resolv)
for (p = ether_types ; p->s != NULL ; p++)
if (p->x == i)
break;
if (p && p->s != NULL)
printf(" %s", p->s);
else
printf(" %04x", i);
}
goto do_options;
}
pe = getprotobynumber(chain->fw_prot);
if (pe)
printf(" %s", pe->p_name);
else
printf(" %u", chain->fw_prot);
printf(" from %s", chain->fw_flg & IP_FW_F_INVSRC ? "not " : "");
if (chain->fw_flg & IP_FW_F_SME)
printf("me");
else
print_ip(chain->fw_src, chain->fw_smsk);
if (chain->fw_prot == IPPROTO_TCP || chain->fw_prot == IPPROTO_UDP)
print_ports(chain->fw_prot, nsp, chain->fw_uar.fw_pts,
chain->fw_flg & IP_FW_F_SRNG,
chain->fw_flg & IP_FW_F_SMSK );
printf(" to %s", chain->fw_flg & IP_FW_F_INVDST ? "not " : "");
if (chain->fw_flg & IP_FW_F_DME)
printf("me");
else
print_ip(chain->fw_dst, chain->fw_dmsk);
if (chain->fw_prot == IPPROTO_TCP || chain->fw_prot == IPPROTO_UDP)
print_ports(chain->fw_prot, ndp, chain->fw_uar.fw_pts+nsp,
chain->fw_flg & IP_FW_F_DRNG,
chain->fw_flg & IP_FW_F_DMSK );
do_options:
if (chain->fw_flg & IP_FW_F_UID) {
struct passwd *pwd = getpwuid(chain->fw_uid);
if (pwd)
printf(" uid %s", pwd->pw_name);
else
printf(" uid %u", chain->fw_uid);
}
if (chain->fw_flg & IP_FW_F_GID) {
struct group *grp = getgrgid(chain->fw_gid);
if (grp)
printf(" gid %s", grp->gr_name);
else
printf(" gid %u", chain->fw_gid);
}
if (chain->fw_flg & IP_FW_F_KEEP_S) {
struct _s_x *p = limit_masks;
switch(chain->dyn_type) {
default:
printf(" *** unknown type ***");
break ;
case DYN_KEEP_STATE:
printf(" keep-state");
break;
case DYN_LIMIT:
printf(" limit");
for ( ; p->x != 0 ; p++)
if (chain->limit_mask & p->x)
printf(" %s", p->s);
printf(" %d", chain->conn_limit);
break ;
}
}
/* Direction */
if (chain->fw_flg & IP_FW_BRIDGED)
printf(" bridged");
if ((chain->fw_flg & IP_FW_F_IN) && !(chain->fw_flg & IP_FW_F_OUT))
printf(" in");
if (!(chain->fw_flg & IP_FW_F_IN) && (chain->fw_flg & IP_FW_F_OUT))
printf(" out");
/* Handle hack for "via" backwards compatibility */
if ((chain->fw_flg & IF_FW_F_VIAHACK) == IF_FW_F_VIAHACK) {
print_iface("via", &chain->fw_in_if,
chain->fw_flg & IP_FW_F_IIFNAME);
} else {
/* Receive interface specified */
if (chain->fw_flg & IP_FW_F_IIFACE)
print_iface("recv", &chain->fw_in_if,
chain->fw_flg & IP_FW_F_IIFNAME);
/* Transmit interface specified */
if (chain->fw_flg & IP_FW_F_OIFACE)
print_iface("xmit", &chain->fw_out_if,
chain->fw_flg & IP_FW_F_OIFNAME);
}
if (chain->fw_flg & IP_FW_F_FRAG)
printf(" frag");
if (chain->fw_ipflg & IP_FW_IF_IPOPT)
printopts("ipopt", chain->fw_ipopt, chain->fw_ipnopt,
f_ipopts);
if (chain->fw_ipflg & IP_FW_IF_IPLEN)
printf(" iplen %u", chain->fw_iplen);
if (chain->fw_ipflg & IP_FW_IF_IPID)
printf(" ipid %#x", chain->fw_ipid);
if (chain->fw_ipflg & IP_FW_IF_IPPRE)
printf(" ipprecedence %u", (chain->fw_iptos & 0xe0) >> 5);
if (chain->fw_ipflg & IP_FW_IF_IPTOS)
printopts("iptos", chain->fw_iptos, chain->fw_ipntos,
f_iptos);
if (chain->fw_ipflg & IP_FW_IF_IPTTL)
printf(" ipttl %u", chain->fw_ipttl);
if (chain->fw_ipflg & IP_FW_IF_IPVER)
printf(" ipversion %u", chain->fw_ipver);
if (chain->fw_ipflg & IP_FW_IF_TCPEST)
printf(" established");
else if (chain->fw_tcpf == TH_SYN &&
chain->fw_tcpnf == TH_ACK)
printf(" setup");
else if (chain->fw_ipflg & IP_FW_IF_TCPFLG)
printopts("tcpflags", chain->fw_tcpf, chain->fw_tcpnf,
f_tcpflags);
if (chain->fw_ipflg & IP_FW_IF_TCPOPT)
printopts("tcpoptions", chain->fw_tcpopt, chain->fw_tcpnopt,
f_tcpopts);
if (chain->fw_ipflg & IP_FW_IF_TCPSEQ)
printf(" tcpseq %lu", (u_long)ntohl(chain->fw_tcpseq));
if (chain->fw_ipflg & IP_FW_IF_TCPACK)
printf(" tcpack %lu", (u_long)ntohl(chain->fw_tcpack));
if (chain->fw_ipflg & IP_FW_IF_TCPWIN)
printf(" tcpwin %hu", ntohs(chain->fw_tcpwin));
if (chain->fw_flg & IP_FW_F_ICMPBIT) {
int i, first = 1;
unsigned j;
printf(" icmptype");
for (i = 0; i < IP_FW_ICMPTYPES_DIM; ++i)
for (j = 0; j < sizeof(unsigned) * 8; ++j)
if (chain->fw_uar.fw_icmptypes[i] & (1 << j)) {
printf("%c%d", first ? ' ' : ',',
i * sizeof(unsigned) * 8 + j);
first = 0;
}
}
printf("\n");
done:
if (do_resolv)
endservent();
}
static void
show_dyn_ipfw(struct ipfw_dyn_rule *d)
{
struct protoent *pe;
struct in_addr a;
if (!d->expire && !do_expired)
return;
printf("%05d %qu %qu (T %ds, slot %d)",
(int)(d->rule), d->pcnt, d->bcnt, d->expire, d->bucket);
switch (d->dyn_type) {
case DYN_LIMIT_PARENT:
printf(" PARENT %d", d->count);
break;
case DYN_LIMIT:
printf(" LIMIT");
break;
case DYN_KEEP_STATE: /* bidir, no mask */
printf(" <->");
break;
}
if (do_resolv && (pe = getprotobynumber(d->id.proto)) != NULL)
printf(" %s,", pe->p_name);
else
printf(" %u,", d->id.proto);
a.s_addr = htonl(d->id.src_ip);
printf(" %s %d", inet_ntoa(a), d->id.src_port);
a.s_addr = htonl(d->id.dst_ip);
printf("<-> %s %d", inet_ntoa(a), d->id.dst_port);
printf("\n");
}
int
sort_q(const void *pa, const void *pb)
{
int rev = (do_sort < 0);
int field = rev ? -do_sort : do_sort;
long long res = 0;
const struct dn_flow_queue *a = pa;
const struct dn_flow_queue *b = pb;
switch (field) {
case 1: /* pkts */
res = a->len - b->len;
break;
case 2: /* bytes */
res = a->len_bytes - b->len_bytes;
break;
case 3: /* tot pkts */
res = a->tot_pkts - b->tot_pkts;
break;
case 4: /* tot bytes */
res = a->tot_bytes - b->tot_bytes;
break;
}
if (res < 0)
res = -1;
if (res > 0)
res = 1;
return (int)(rev ? res : -res);
}
static void
list_queues(struct dn_flow_set *fs, struct dn_flow_queue *q)
{
int l;
printf(" mask: 0x%02x 0x%08x/0x%04x -> 0x%08x/0x%04x\n",
fs->flow_mask.proto,
fs->flow_mask.src_ip, fs->flow_mask.src_port,
fs->flow_mask.dst_ip, fs->flow_mask.dst_port);
if (fs->rq_elements == 0)
return;
printf("BKT Prot ___Source IP/port____ "
"____Dest. IP/port____ Tot_pkt/bytes Pkt/Byte Drp\n");
if (do_sort != 0)
heapsort(q, fs->rq_elements, sizeof *q, sort_q);
for (l = 0; l < fs->rq_elements; l++) {
struct in_addr ina;
struct protoent *pe;
ina.s_addr = htonl(q[l].id.src_ip);
printf("%3d ", q[l].hash_slot);
pe = getprotobynumber(q[l].id.proto);
if (pe)
printf("%-4s ", pe->p_name);
else
printf("%4u ", q[l].id.proto);
printf("%15s/%-5d ",
inet_ntoa(ina), q[l].id.src_port);
ina.s_addr = htonl(q[l].id.dst_ip);
printf("%15s/%-5d ",
inet_ntoa(ina), q[l].id.dst_port);
printf("%4qu %8qu %2u %4u %3u\n",
q[l].tot_pkts, q[l].tot_bytes,
q[l].len, q[l].len_bytes, q[l].drops);
if (verbose)
printf(" S %20qd F %20qd\n",
q[l].S, q[l].F);
}
}
static void
print_flowset_parms(struct dn_flow_set *fs, char *prefix)
{
int l;
char qs[30];
char plr[30];
char red[90]; /* Display RED parameters */
l = fs->qsize;
if (fs->flags_fs & DN_QSIZE_IS_BYTES) {
if (l >= 8192)
sprintf(qs, "%d KB", l / 1024);
else
sprintf(qs, "%d B", l);
} else
sprintf(qs, "%3d sl.", l);
if (fs->plr)
sprintf(plr, "plr %f", 1.0 * fs->plr / (double)(0x7fffffff));
else
plr[0] = '\0';
if (fs->flags_fs & DN_IS_RED) /* RED parameters */
sprintf(red,
"\n\t %cRED w_q %f min_th %d max_th %d max_p %f",
(fs->flags_fs & DN_IS_GENTLE_RED) ? 'G' : ' ',
1.0 * fs->w_q / (double)(1 << SCALE_RED),
SCALE_VAL(fs->min_th),
SCALE_VAL(fs->max_th),
1.0 * fs->max_p / (double)(1 << SCALE_RED));
else
sprintf(red, "droptail");
printf("%s %s%s %d queues (%d buckets) %s\n",
prefix, qs, plr, fs->rq_elements, fs->rq_size, red);
}
static void
list(int ac, char *av[])
{
struct ip_fw *rules, *r;
struct ipfw_dyn_rule *dynrules, *d;
struct dn_pipe *pipes;
void *data = NULL;
int n, nbytes, nstat, ndyn;
int exitval = EX_OK;
int lac;
char **lav;
u_long rnum;
char *endptr;
int seen = 0;
/* get rules or pipes from kernel, resizing array as necessary */
{
const int unit = do_pipe ? sizeof(*pipes) : sizeof(*rules);
const int ocmd = do_pipe ? IP_DUMMYNET_GET : IP_FW_GET;
int nalloc = unit;
nbytes = nalloc;
while (nbytes >= nalloc) {
nalloc = nalloc * 2 + 200;
nbytes = nalloc;
if ((data = realloc(data, nbytes)) == NULL)
err(EX_OSERR, "realloc");
if (getsockopt(s, IPPROTO_IP, ocmd, data, &nbytes) < 0)
err(EX_OSERR, "getsockopt(IP_%s_GET)",
do_pipe ? "DUMMYNET" : "FW");
}
}
/* display requested pipes */
if (do_pipe) {
u_long rulenum;
void *next = data;
struct dn_pipe *p = (struct dn_pipe *) data;
struct dn_flow_set *fs;
struct dn_flow_queue *q;
int l;
if (ac > 0)
rulenum = strtoul(*av++, NULL, 10);
else
rulenum = 0;
for (; nbytes >= sizeof *p; p = (struct dn_pipe *)next) {
double b = p->bandwidth;
char buf[30];
char prefix[80];
if (p->next != (struct dn_pipe *)DN_IS_PIPE)
break;
l = sizeof(*p) + p->fs.rq_elements * sizeof(*q);
next = (void *)p + l;
nbytes -= l;
q = (struct dn_flow_queue *)(p+1);
if (rulenum != 0 && rulenum != p->pipe_nr)
continue;
if (p->if_name[0] != '\0')
sprintf(buf, "%s", p->if_name);
else if (b == 0)
sprintf(buf, "unlimited");
else if (b >= 1000000)
sprintf(buf, "%7.3f Mbit/s", b/1000000);
else if (b >= 1000)
sprintf(buf, "%7.3f Kbit/s", b/1000);
else
sprintf(buf, "%7.3f bit/s ", b);
sprintf(prefix, "%05d: %s %4d ms ",
p->pipe_nr, buf, p->delay);
print_flowset_parms(&(p->fs), prefix);
if (verbose)
printf(" V %20qd\n", p->V >> MY_M);
list_queues(&(p->fs), q);
}
fs = (struct dn_flow_set *) next;
for (; nbytes >= sizeof *fs; fs = (struct dn_flow_set *)next) {
char prefix[80];
if (fs->next != (struct dn_flow_set *)DN_IS_QUEUE)
break;
l = sizeof(*fs) + fs->rq_elements * sizeof(*q);
next = (void *)fs + l;
nbytes -= l;
q = (struct dn_flow_queue *)(fs+1);
sprintf(prefix, "q%05d: weight %d pipe %d ",
fs->fs_nr, fs->weight, fs->parent_nr);
print_flowset_parms(fs, prefix);
list_queues(fs, q);
}
free(data);
return;
}
rules = (struct ip_fw *)data;
for (nstat = 0; rules[nstat].fw_number < 65535; ++nstat)
/* nothing */ ;
nstat++; /* counting starts from 0 ... */
dynrules = (struct ipfw_dyn_rule *)&rules[nstat];
ndyn = (nbytes - (nstat * sizeof *rules)) / sizeof *dynrules;
/* if no rule numbers were specified, list all rules */
if (ac == 0) {
for (n = 0; n < nstat; n++)
show_ipfw(&rules[n]);
if (do_dynamic && ndyn) {
printf("## Dynamic rules:\n");
for (n = 0, d = dynrules; n < ndyn; n++, d++)
show_dyn_ipfw(d);
}
free(data);
return;
}
/* display specific rules requested on command line */
for (lac = ac, lav = av; lac != 0; lac--) {
/* convert command line rule # */
rnum = strtoul(*lav++, &endptr, 10);
if (*endptr) {
exitval = EX_USAGE;
warnx("invalid rule number: %s", *(lav - 1));
continue;
}
for (n = seen = 0, r = rules; n < nstat; n++, r++) {
if (r->fw_number > rnum)
break;
if (r->fw_number == rnum) {
show_ipfw(r);
seen = 1;
}
}
if (!seen) {
/* give precedence to other error(s) */
if (exitval == EX_OK)
exitval = EX_UNAVAILABLE;
warnx("rule %lu does not exist", rnum);
}
}
printf("## Dynamic rules:\n");
if (do_dynamic && ndyn) {
for (lac = ac, lav = av; lac != 0; lac--) {
rnum = strtoul(*lav++, &endptr, 10);
if (*endptr)
/* already warned */
continue;
for (n = 0, d = dynrules; n < ndyn; n++, d++) {
if ((int)(d->rule) > rnum)
break;
if ((int)(d->rule) == rnum)
show_dyn_ipfw(d);
}
}
}
ac = 0;
free(data);
if (exitval != EX_OK)
exit(exitval);
}
static void
show_usage(void)
{
fprintf(stderr, "usage: ipfw [options]\n"
" add [number] rule\n"
" pipe number config [pipeconfig]\n"
" queue number config [queueconfig]\n"
" [pipe] flush\n"
" [pipe] delete number ...\n"
" [pipe] {list|show} [number ...]\n"
" {zero|resetlog} [number ...]\n"
" rule: [prob <match_probability>] action (mac-hdr|ip-hdr) options...\n"
" action:\n"
" {allow|permit|accept|pass|deny|drop|reject|unreach code|\n"
" reset|count|skipto num|divert port|tee port|fwd ip|\n"
" pipe num} [log [logamount count]]\n"
" proto: {ip|tcp|udp|icmp|<number>}\n"
" src: from [not] {me|any|ip[{/bits|:mask}]} [{port[-port]}, [port], ...]\n"
" dst: to [not] {me|any|ip[{/bits|:mask}]} [{port[-port]}, [port], ...]\n"
" extras:\n"
" uid {user id}\n"
" gid {group id}\n"
" fragment (may not be used with ports or tcpflags)\n"
" in\n"
" out\n"
" {xmit|recv|via} {iface|ip|any}\n"
" {established|setup}\n"
" tcpflags [!]{syn|fin|rst|ack|psh|urg}, ...\n"
" ipoptions [!]{ssrr|lsrr|rr|ts}, ...\n"
" iplen {length}\n"
" ipid {identification number}\n"
" ipprecedence {precedence}\n"
" iptos [!]{lowdelay|throughput|reliability|mincost|congestion}, ...\n"
" ipttl {time to live}\n"
" ipversion {version number}\n"
" tcpoptions [!]{mss|window|sack|ts|cc}, ...\n"
" tcpseq {sequence number}\n"
" tcpack {acknowledgement number}\n"
" tcpwin {window size}\n"
" icmptypes {type[, type]}...\n"
" keep-state [method]\n"
" pipeconfig:\n"
" {bw|bandwidth} <number>{bit/s|Kbit/s|Mbit/s|Bytes/s|KBytes/s|MBytes/s}\n"
" {bw|bandwidth} interface_name\n"
" delay <milliseconds>\n"
" queue <size>{packets|Bytes|KBytes}\n"
" plr <fraction>\n"
" mask {all| [dst-ip|src-ip|dst-port|src-port|proto] <number>}\n"
" buckets <number>}\n"
" {red|gred} <fraction>/<number>/<number>/<fraction>\n"
" droptail\n"
);
exit(EX_USAGE);
}
static int
lookup_host (char *host, struct in_addr *ipaddr)
{
struct hostent *he;
if (!inet_aton(host, ipaddr)) {
if ((he = gethostbyname(host)) == NULL)
return(-1);
*ipaddr = *(struct in_addr *)he->h_addr_list[0];
}
return(0);
}
static void
fill_ip(struct in_addr *ipno, struct in_addr *mask, int *acp, char ***avp)
{
int ac = *acp;
char **av = *avp;
char *p = 0, md = 0;
if (ac && !strncmp(*av, "any", strlen(*av))) {
ipno->s_addr = mask->s_addr = 0; av++; ac--;
} else {
u_int32_t i;
p = strchr(*av, '/');
if (!p)
p = strchr(*av, ':');
if (p) {
md = *p;
*p++ = '\0';
}
if (lookup_host(*av, ipno) != 0)
errx(EX_NOHOST, "hostname ``%s'' unknown", *av);
switch (md) {
case ':':
if (!inet_aton(p, mask))
errx(EX_DATAERR, "bad netmask ``%s''", p);
break;
case '/':
i = atoi(p);
if (i == 0)
mask->s_addr = 0;
else if (i > 32)
errx(EX_DATAERR, "bad width ``%s''", p);
else
mask->s_addr = htonl(~0 << (32 - i));
break;
default:
mask->s_addr = htonl(~0);
break;
}
ipno->s_addr &= mask->s_addr;
av++;
ac--;
}
*acp = ac;
*avp = av;
}
static void
fill_reject_code(u_short *codep, char *str)
{
struct icmpcode *ic;
u_long val;
char *s;
if (str == '\0')
errx(EX_DATAERR, "missing unreachable code");
val = strtoul(str, &s, 0);
if (s != str && *s == '\0' && val < 0x100) {
*codep = val;
return;
}
for (ic = icmpcodes; ic->str; ic++)
if (!strcasecmp(str, ic->str)) {
*codep = ic->code;
return;
}
errx(EX_DATAERR, "unknown ICMP unreachable code ``%s''", str);
}
static void
add_port(u_short *cnt, u_short *ptr, u_short off, u_short port)
{
if (off + *cnt >= IP_FW_MAX_PORTS)
errx(EX_USAGE, "too many ports (max is %d)", IP_FW_MAX_PORTS);
ptr[off+*cnt] = port;
(*cnt)++;
}
static int
lookup_port(const char *arg, int proto, int test, int nodash)
{
int val;
char *earg, buf[32];
struct servent *s;
char *p, *q;
snprintf(buf, sizeof(buf), "%s", arg);
for (p = q = buf; *p; *q++ = *p++) {
if (*p == '\\') {
if (*(p+1))
p++;
} else {
if (*p == ',' || (nodash && *p == '-'))
break;
}
}
*q = '\0';
val = (int) strtoul(buf, &earg, 0);
if (!*buf || *earg) {
char *protocol = NULL;
if (proto != 0) {
struct protoent *pe = getprotobynumber(proto);
if (pe)
protocol = pe->p_name;
}
setservent(1);
s = getservbyname(buf, protocol);
if (s != NULL)
val = htons(s->s_port);
else {
if (!test)
errx(EX_DATAERR, "unknown port ``%s''", buf);
val = -1;
}
} else {
if (val < 0 || val > 0xffff) {
if (!test)
errx(EX_DATAERR,
"port ``%s'' out of range", buf);
val = -1;
}
}
return(val);
}
/*
* return: 0 normally, 1 if first pair is a range,
* 2 if first pair is a port+mask
*/
static int
fill_port(u_short *cnt, u_short *ptr, u_short off, char *arg, int proto)
{
char *s;
int initial_range = 0;
for (s = arg; *s && *s != ',' && *s != '-' && *s != ':'; s++) {
if (*s == '\\' && *(s+1))
s++;
}
if (*s == ':') {
*s++ = '\0';
if (strchr(arg, ','))
errx(EX_USAGE, "port/mask must be first in list");
add_port(cnt, ptr, off,
*arg ? lookup_port(arg, proto, 0, 0) : 0x0000);
arg = s;
s = strchr(arg, ',');
if (s)
*s++ = '\0';
add_port(cnt, ptr, off,
*arg ? lookup_port(arg, proto, 0, 0) : 0xffff);
arg = s;
initial_range = 2;
} else if (*s == '-') {
*s++ = '\0';
if (strchr(arg, ','))
errx(EX_USAGE, "port range must be first in list");
add_port(cnt, ptr, off,
*arg ? lookup_port(arg, proto, 0, 0) : 0x0000);
arg = s;
s = strchr(arg, ',');
if (s)
*s++ = '\0';
add_port(cnt, ptr, off,
*arg ? lookup_port(arg, proto, 0, 0) : 0xffff);
arg = s;
initial_range = 1;
}
while (arg != NULL) {
s = strchr(arg, ',');
if (s)
*s++ = '\0';
add_port(cnt, ptr, off, lookup_port(arg, proto, 0, 0));
arg = s;
}
return initial_range;
}
/*
* helper function to process a set of flags and set bits in the
* appropriate masks.
*/
static void
fill_flags(u_char *set, u_char *reset, struct _flaglist *flags, char **vp)
{
char *p = *vp, *q; /* parameter */
u_char *d; /* which mask we are working on */
while (p && *p) {
int i;
if (*p == '!') {
p++;
d = reset;
} else
d = set;
q = strchr(p, ',');
if (q)
*q++ = '\0';
for (i = 0; flags[i].value != 0; ++i)
if (!strncmp(p, flags[i].name, strlen(p))) {
*d |= flags[i].value;
break;
}
if (flags[i].value == 0)
errx(EX_DATAERR, "invalid %s ``%s''", flags[i].name, p);
p = q;
}
}
static void
fill_icmptypes(unsigned *types, char **vp, u_int *fw_flg)
{
unsigned long icmptype;
char *c = *vp;
while (*c) {
if (*c == ',')
++c;
icmptype = strtoul(c, &c, 0);
if (*c != ',' && *c != '\0')
errx(EX_DATAERR, "invalid ICMP type");
if (icmptype >= IP_FW_ICMPTYPES_DIM * sizeof(unsigned) * 8)
errx(EX_DATAERR, "ICMP type out of range");
types[icmptype / (sizeof(unsigned) * 8)] |=
1 << (icmptype % (sizeof(unsigned) * 8));
*fw_flg |= IP_FW_F_ICMPBIT;
}
}
static void
delete(int ac, char *av[])
{
struct ip_fw rule;
struct dn_pipe pipe;
int i;
int exitval = EX_OK;
memset(&rule, 0, sizeof rule);
memset(&pipe, 0, sizeof pipe);
av++; ac--;
/* Rule number */
while (ac && isdigit(**av)) {
i = atoi(*av); av++; ac--;
if (do_pipe) {
if (do_pipe == 1)
pipe.pipe_nr = i;
else
pipe.fs.fs_nr = i;
i = setsockopt(s, IPPROTO_IP, IP_DUMMYNET_DEL,
&pipe, sizeof pipe);
if (i) {
exitval = 1;
warn("rule %u: setsockopt(IP_DUMMYNET_DEL)",
do_pipe == 1 ? pipe.pipe_nr :
pipe.fs.fs_nr);
}
} else {
rule.fw_number = i;
i = setsockopt(s, IPPROTO_IP, IP_FW_DEL, &rule,
sizeof rule);
if (i) {
exitval = EX_UNAVAILABLE;
warn("rule %u: setsockopt(IP_FW_DEL)",
rule.fw_number);
}
}
}
if (exitval != EX_OK)
exit(exitval);
}
static void
verify_interface(union ip_fw_if *ifu)
{
struct ifreq ifr;
/*
* If a unit was specified, check for that exact interface.
* If a wildcard was specified, check for unit 0.
*/
snprintf(ifr.ifr_name, sizeof(ifr.ifr_name), "%s%d",
ifu->fu_via_if.name,
ifu->fu_via_if.unit == -1 ? 0 : ifu->fu_via_if.unit);
if (ioctl(s, SIOCGIFFLAGS, &ifr) < 0)
warnx("warning: interface ``%s'' does not exist",
ifr.ifr_name);
}
static void
fill_iface(char *which, union ip_fw_if *ifu, int *byname, int ac, char *arg)
{
if (!ac)
errx(EX_USAGE, "missing argument for ``%s''", which);
/* Parse the interface or address */
if (!strcmp(arg, "any")) {
ifu->fu_via_ip.s_addr = 0;
*byname = 0;
} else if (!isdigit(*arg)) {
char *q;
*byname = 1;
strncpy(ifu->fu_via_if.name, arg,
sizeof(ifu->fu_via_if.name));
ifu->fu_via_if.name[sizeof(ifu->fu_via_if.name) - 1] = '\0';
for (q = ifu->fu_via_if.name;
*q && !isdigit(*q) && *q != '*'; q++)
continue;
ifu->fu_via_if.unit = (*q == '*') ? -1 : atoi(q);
*q = '\0';
verify_interface(ifu);
} else if (!inet_aton(arg, &ifu->fu_via_ip)) {
errx(EX_DATAERR, "bad ip address ``%s''", arg);
} else
*byname = 0;
}
static void
config_pipe(int ac, char **av)
{
struct dn_pipe pipe;
int i;
char *end;
memset(&pipe, 0, sizeof pipe);
av++; ac--;
/* Pipe number */
if (ac && isdigit(**av)) {
i = atoi(*av); av++; ac--;
if (do_pipe == 1)
pipe.pipe_nr = i;
else
pipe.fs.fs_nr = i;
}
while (ac > 1) {
if (!strncmp(*av, "plr", strlen(*av))) {
double d = strtod(av[1], NULL);
if (d > 1)
d = 1;
else if (d < 0)
d = 0;
pipe.fs.plr = (int)(d*0x7fffffff);
av += 2;
ac -= 2;
} else if (!strncmp(*av, "queue", strlen(*av))) {
end = NULL;
pipe.fs.qsize = strtoul(av[1], &end, 0);
if (*end == 'K' || *end == 'k') {
pipe.fs.flags_fs |= DN_QSIZE_IS_BYTES;
pipe.fs.qsize *= 1024;
} else if (*end == 'B' || !strncmp(end, "by", 2)) {
pipe.fs.flags_fs |= DN_QSIZE_IS_BYTES;
}
av += 2;
ac -= 2;
} else if (!strncmp(*av, "buckets", strlen(*av))) {
pipe.fs.rq_size = strtoul(av[1], NULL, 0);
av += 2;
ac -= 2;
} else if (!strncmp(*av, "mask", strlen(*av))) {
/* per-flow queue, mask is dst_ip, dst_port,
* src_ip, src_port, proto measured in bits
*/
u_int32_t a;
void *par = NULL;
pipe.fs.flow_mask.dst_ip = 0;
pipe.fs.flow_mask.src_ip = 0;
pipe.fs.flow_mask.dst_port = 0;
pipe.fs.flow_mask.src_port = 0;
pipe.fs.flow_mask.proto = 0;
end = NULL;
av++; ac--;
if (ac >= 1 && !strncmp(*av, "all", strlen(*av))) {
/* special case -- all bits are significant */
pipe.fs.flow_mask.dst_ip = ~0;
pipe.fs.flow_mask.src_ip = ~0;
pipe.fs.flow_mask.dst_port = ~0;
pipe.fs.flow_mask.src_port = ~0;
pipe.fs.flow_mask.proto = ~0;
pipe.fs.flags_fs |= DN_HAVE_FLOW_MASK;
av++;
ac--;
continue;
}
while (ac >= 1) {
int len = strlen(*av);
if (!strncmp(*av, "dst-ip", len))
par = &pipe.fs.flow_mask.dst_ip;
else if (!strncmp(*av, "src-ip", len))
par = &pipe.fs.flow_mask.src_ip;
else if (!strncmp(*av, "dst-port", len))
par = &pipe.fs.flow_mask.dst_port;
else if (!strncmp(*av, "src-port", len))
par = &pipe.fs.flow_mask.src_port;
else if (!strncmp(*av, "proto", len))
par = &pipe.fs.flow_mask.proto;
else
break;
if (ac < 2)
errx(EX_USAGE, "mask: %s value"
" missing", *av);
if (*av[1] == '/') {
a = strtoul(av[1]+1, &end, 0);
if (a == 32) /* special case... */
a = ~0;
else
a = (1 << a) - 1;
} else {
a = strtoul(av[1], &end, 0);
}
if (par == &pipe.fs.flow_mask.src_port
|| par == &pipe.fs.flow_mask.dst_port) {
if (a >= (1 << 16))
errx(EX_DATAERR, "mask: %s"
" must be 16 bit, not"
" 0x%08x", *av, a);
*((u_int16_t *)par) = (u_int16_t)a;
} else if (par == &pipe.fs.flow_mask.proto) {
if (a >= (1 << 8))
errx(EX_DATAERR, "mask: %s"
" must be"
" 8 bit, not 0x%08x",
*av, a);
*((u_int8_t *)par) = (u_int8_t)a;
} else
*((u_int32_t *)par) = a;
if (a != 0)
pipe.fs.flags_fs |= DN_HAVE_FLOW_MASK;
av += 2;
ac -= 2;
} /* end for */
} else if (!strncmp(*av, "red", strlen(*av))
|| !strncmp(*av, "gred", strlen(*av))) {
/* RED enabled */
pipe.fs.flags_fs |= DN_IS_RED;
if (*av[0] == 'g')
pipe.fs.flags_fs |= DN_IS_GENTLE_RED;
if ((end = strsep(&av[1], "/"))) {
double w_q = strtod(end, NULL);
if (w_q > 1 || w_q <= 0)
errx(EX_DATAERR, "w_q %f must be "
"0 < x <= 1", w_q);
pipe.fs.w_q = (int) (w_q * (1 << SCALE_RED));
}
if ((end = strsep(&av[1], "/"))) {
pipe.fs.min_th = strtoul(end, &end, 0);
if (*end == 'K' || *end == 'k')
pipe.fs.min_th *= 1024;
}
if ((end = strsep(&av[1], "/"))) {
pipe.fs.max_th = strtoul(end, &end, 0);
if (*end == 'K' || *end == 'k')
pipe.fs.max_th *= 1024;
}
if ((end = strsep(&av[1], "/"))) {
double max_p = strtod(end, NULL);
if (max_p > 1 || max_p <= 0)
errx(EX_DATAERR, "max_p %f must be "
"0 < x <= 1", max_p);
pipe.fs.max_p =
(int)(max_p * (1 << SCALE_RED));
}
av += 2;
ac -= 2;
} else if (!strncmp(*av, "droptail", strlen(*av))) {
/* DROPTAIL */
pipe.fs.flags_fs &= ~(DN_IS_RED|DN_IS_GENTLE_RED);
av += 1;
ac -= 1;
} else {
int len = strlen(*av);
if (do_pipe == 1) {
/* some commands are only good for pipes. */
if (!strncmp(*av, "bw", len)
|| !strncmp(*av, "bandwidth", len)) {
if (av[1][0] >= 'a'
&& av[1][0] <= 'z') {
int l = sizeof(pipe.if_name)-1;
/* interface name */
strncpy(pipe.if_name, av[1], l);
pipe.if_name[l] = '\0';
pipe.bandwidth = 0;
} else {
pipe.if_name[0] = '\0';
pipe.bandwidth =
strtoul(av[1], &end, 0);
if (*end == 'K'
|| *end == 'k') {
end++;
pipe.bandwidth *=
1000;
} else if (*end == 'M') {
end++;
pipe.bandwidth *=
1000000;
}
if (*end == 'B'
|| !strncmp(end, "by", 2))
pipe.bandwidth *= 8;
}
if (pipe.bandwidth < 0)
errx(EX_DATAERR,
"bandwidth too large");
av += 2;
ac -= 2;
} else if (!strncmp(*av, "delay", len)) {
pipe.delay = strtoul(av[1], NULL, 0);
av += 2;
ac -= 2;
} else {
errx(EX_DATAERR, "unrecognised pipe"
" option ``%s''", *av);
}
} else { /* this refers to a queue */
if (!strncmp(*av, "weight", len)) {
pipe.fs.weight =
strtoul(av[1], &end, 0);
av += 2;
ac -= 2;
} else if (!strncmp(*av, "pipe", len)) {
pipe.fs.parent_nr =
strtoul(av[1], &end, 0);
av += 2;
ac -= 2;
} else {
errx(EX_DATAERR, "unrecognised option "
"``%s''", *av);
}
}
}
}
if (do_pipe == 1) {
if (pipe.pipe_nr == 0)
errx(EX_DATAERR, "pipe_nr %d must be > 0",
pipe.pipe_nr);
if (pipe.delay > 10000)
errx(EX_DATAERR, "delay %d must be < 10000",
pipe.delay);
} else { /* do_pipe == 2, queue */
if (pipe.fs.parent_nr == 0)
errx(EX_DATAERR, "pipe %d must be > 0",
pipe.fs.parent_nr);
if (pipe.fs.weight >100)
errx(EX_DATAERR, "weight %d must be <= 100",
pipe.fs.weight);
}
if (pipe.fs.flags_fs & DN_QSIZE_IS_BYTES) {
if (pipe.fs.qsize > 1024*1024)
errx(EX_DATAERR, "queue size %d, must be < 1MB",
pipe.fs.qsize);
} else {
if (pipe.fs.qsize > 100)
errx(EX_DATAERR, "queue size %d, must be"
" 2 <= x <= 100", pipe.fs.qsize);
}
if (pipe.fs.flags_fs & DN_IS_RED) {
size_t len;
int lookup_depth, avg_pkt_size;
double s, idle, weight, w_q;
struct clockinfo clock;
int t;
if (pipe.fs.min_th >= pipe.fs.max_th)
errx(EX_DATAERR, "min_th %d must be < than max_th %d",
pipe.fs.min_th, pipe.fs.max_th);
if (pipe.fs.max_th == 0)
errx(EX_DATAERR, "max_th must be > 0");
len = sizeof(int);
if (sysctlbyname("net.inet.ip.dummynet.red_lookup_depth",
&lookup_depth, &len, NULL, 0) == -1)
errx(1, "sysctlbyname(\"%s\")",
"net.inet.ip.dummynet.red_lookup_depth");
if (lookup_depth == 0)
errx(EX_DATAERR, "net.inet.ip.dummynet.red_lookup_depth"
" must be greater than zero");
len = sizeof(int);
if (sysctlbyname("net.inet.ip.dummynet.red_avg_pkt_size",
&avg_pkt_size, &len, NULL, 0) == -1)
errx(1, "sysctlbyname(\"%s\")",
"net.inet.ip.dummynet.red_avg_pkt_size");
if (avg_pkt_size == 0)
errx(EX_DATAERR,
"net.inet.ip.dummynet.red_avg_pkt_size must"
" be greater than zero");
len = sizeof(struct clockinfo);
if (sysctlbyname("kern.clockrate", &clock, &len, NULL, 0) == -1)
errx(1, "sysctlbyname(\"%s\")", "kern.clockrate");
/*
* Ticks needed for sending a medium-sized packet.
* Unfortunately, when we are configuring a WF2Q+ queue, we
* do not have bandwidth information, because that is stored
* in the parent pipe, and also we have multiple queues
* competing for it. So we set s=0, which is not very
* correct. But on the other hand, why do we want RED with
* WF2Q+ ?
*/
if (pipe.bandwidth==0) /* this is a WF2Q+ queue */
s = 0;
else
s = clock.hz * avg_pkt_size * 8 / pipe.bandwidth;
/*
* max idle time (in ticks) before avg queue size becomes 0.
* NOTA: (3/w_q) is approx the value x so that
* (1-w_q)^x < 10^-3.
*/
w_q = ((double)pipe.fs.w_q) / (1 << SCALE_RED);
idle = s * 3. / w_q;
pipe.fs.lookup_step = (int)idle / lookup_depth;
if (!pipe.fs.lookup_step)
pipe.fs.lookup_step = 1;
weight = 1 - w_q;
for (t = pipe.fs.lookup_step; t > 0; --t)
weight *= weight;
pipe.fs.lookup_weight = (int)(weight * (1 << SCALE_RED));
}
i = setsockopt(s, IPPROTO_IP, IP_DUMMYNET_CONFIGURE, &pipe,
sizeof pipe);
if (i)
err(1, "setsockopt(%s)", "IP_DUMMYNET_CONFIGURE");
}
static void
get_mac_addr_mask(char *p, u_char *addr, u_char *mask)
{
int i, l;
for (i=0; i<6; i++)
addr[i] = mask[i] = 0;
if (!strcmp(p, "any"))
return;
for (i=0; *p && i<6;i++, p++) {
addr[i] = strtol(p, &p, 16);
if (*p != ':') /* we start with the mask */
break;
}
if (*p == '/') { /* mask len */
l = strtol(p+1, &p, 0);
for (i=0; l>0; l -=8, i++)
mask[i] = (l >=8) ? 0xff : (~0) << (8-l);
} else if (*p == '&') { /* mask */
for (i=0, p++; *p && i<6;i++, p++) {
mask[i] = strtol(p, &p, 16);
if (*p != ':')
break;
}
} else if (*p == '\0') {
for (i=0; i<6; i++)
mask[i] = 0xff;
}
for (i=0; i<6; i++)
addr[i] &= mask[i];
}
/*
* fetch and add the MAC address and type, with masks
*/
static void
add_mac(struct ip_fw *rule, int ac, char *av[])
{
u_char *addr, *mask;
u_short *type, *typemask;
int i;
char *p;
struct _s_x *pt;
if (ac <3)
errx(EX_DATAERR, "MAC dst src type");
addr = (u_char *)&(rule->fw_mac_hdr);
mask = (u_char *)&(rule->fw_mac_mask);
get_mac_addr_mask(av[0], addr, mask);
addr += 6;
mask += 6;
av++;
get_mac_addr_mask(av[0], addr, mask);
av++;
type = (u_short *)&(rule->fw_mac_type);
typemask = (u_short *)&(rule->fw_mac_mask_type);
rule->fw_flg |= IP_FW_F_MAC;
if (!strcmp(av[0], "any")) {
*type = *typemask = htons(0);
return;
}
/*
* the match length is the string up to the first separator
* we know, i.e. any of "\0:/&". Note, we use bcmp instead of
* strcmp as we want an exact match.
*/
p = strpbrk(av[0], "-:/&");
if (p == NULL)
i = strlen(av[0]);
else
i = p - av[0];
for (pt = ether_types ; i && pt->s != NULL ; pt++)
if (strlen(pt->s) == i && !bcmp(*av, pt->s, i))
break;
/* store type in network format for all cases but range */
if (pt->s != NULL) {
*type = htons(pt->x);
p = av[0] + i;
} else
*type = htons( strtol(av[0], &p, 16) );
*typemask = htons(0xffff); /* default */
if (*p == '-') {
rule->fw_flg |= IP_FW_F_SRNG;
*type = ntohs(*type); /* revert to host format */
p++;
i = strlen(p);
for (pt = ether_types ; i && pt->s != NULL ; pt++)
if (strlen(pt->s) == i && !bcmp(p, pt->s, i))
break;
if (pt->s != NULL) {
*typemask = pt->x;
p += i;
} else
*typemask = strtol(p, &p, 16);
} else if (*p == '/') {
i = strtol(p+1, &p, 10);
if (i > 16)
errx(EX_DATAERR, "MAC: bad type %s\n", av[0]);
*typemask = htons( (~0) << (16 - i) );
*type &= *typemask;
} else if (*p == ':') {
*typemask = htons( strtol(p+1, &p, 16) );
*type &= *typemask;
}
if (*p != '\0')
errx(EX_DATAERR, "MAC: bad end type %s\n", av[0]);
}
/*
* the following macro returns an error message if we run out of
* arguments.
*/
#define NEED1(msg) {if (!ac) errx(EX_USAGE, msg);}
static void
add(int ac, char *av[])
{
struct ip_fw rule;
int i;
u_char proto;
struct protoent *pe;
int saw_xmrc = 0, saw_via = 0;
memset(&rule, 0, sizeof rule);
av++; ac--;
/* [rule N] -- Rule number optional */
if (ac && isdigit(**av)) {
rule.fw_number = atoi(*av); av++; ac--;
}
/* [prob D] -- match probability, optional */
if (ac > 1 && !strncmp(*av, "prob", strlen(*av))) {
double d = strtod(av[1], NULL);
if (d <= 0 || d > 1)
errx(EX_DATAERR, "illegal match prob. %s", av[1]);
if (d != 1) { /* 1 means always match */
rule.fw_flg |= IP_FW_F_RND_MATCH;
rule.dont_match_prob = (long)((1 - d) * 0x7fffffff);
}
av += 2; ac -= 2;
}
/* action -- mandatory */
NEED1("missing action");
if (!strncmp(*av, "accept", strlen(*av))
|| !strncmp(*av, "pass", strlen(*av))
|| !strncmp(*av, "allow", strlen(*av))
|| !strncmp(*av, "permit", strlen(*av))) {
rule.fw_flg |= IP_FW_F_ACCEPT; av++; ac--;
} else if (!strncmp(*av, "count", strlen(*av))) {
rule.fw_flg |= IP_FW_F_COUNT; av++; ac--;
} else if (!strncmp(*av, "pipe", strlen(*av))) {
rule.fw_flg |= IP_FW_F_PIPE; av++; ac--;
NEED1("missing pipe number");
rule.fw_divert_port = strtoul(*av, NULL, 0); av++; ac--;
} else if (!strncmp(*av, "queue", strlen(*av))) {
rule.fw_flg |= IP_FW_F_QUEUE; av++; ac--;
NEED1("missing queue number");
rule.fw_divert_port = strtoul(*av, NULL, 0); av++; ac--;
} else if (!strncmp(*av, "divert", strlen(*av))) {
rule.fw_flg |= IP_FW_F_DIVERT; av++; ac--;
NEED1("missing divert port");
rule.fw_divert_port = strtoul(*av, NULL, 0); av++; ac--;
if (rule.fw_divert_port == 0) {
struct servent *s;
setservent(1);
s = getservbyname(av[-1], "divert");
if (s != NULL)
rule.fw_divert_port = ntohs(s->s_port);
else
errx(EX_DATAERR, "illegal %s port", "divert");
}
} else if (!strncmp(*av, "tee", strlen(*av))) {
rule.fw_flg |= IP_FW_F_TEE; av++; ac--;
NEED1("missing tee divert port");
rule.fw_divert_port = strtoul(*av, NULL, 0); av++; ac--;
if (rule.fw_divert_port == 0) {
struct servent *s;
setservent(1);
s = getservbyname(av[-1], "divert");
if (s != NULL)
rule.fw_divert_port = ntohs(s->s_port);
else
errx(EX_DATAERR, "illegal %s port",
"tee divert");
}
} else if (!strncmp(*av, "fwd", strlen(*av))
|| !strncmp(*av, "forward", strlen(*av))) {
struct in_addr dummyip;
char *pp;
rule.fw_flg |= IP_FW_F_FWD; av++; ac--;
NEED1("missing forwarding IP address");
rule.fw_fwd_ip.sin_len = sizeof(struct sockaddr_in);
rule.fw_fwd_ip.sin_family = AF_INET;
rule.fw_fwd_ip.sin_port = 0;
pp = strchr(*av, ':');
if( pp == NULL)
pp = strchr(*av, ',');
if (pp != NULL) {
*(pp++) = '\0';
i = lookup_port(pp, 0, 1, 0);
if (i == -1)
errx(EX_DATAERR, "illegal forwarding"
" port ``%s''", pp);
else
rule.fw_fwd_ip.sin_port = (u_short)i;
}
fill_ip(&(rule.fw_fwd_ip.sin_addr), &dummyip, &ac, &av);
if (rule.fw_fwd_ip.sin_addr.s_addr == 0)
errx(EX_DATAERR, "illegal forwarding IP address");
} else if (!strncmp(*av, "skipto", strlen(*av))) {
rule.fw_flg |= IP_FW_F_SKIPTO; av++; ac--;
NEED1("missing skipto rule number");
rule.fw_skipto_rule = strtoul(*av, NULL, 10); av++; ac--;
} else if ((!strncmp(*av, "deny", strlen(*av))
|| !strncmp(*av, "drop", strlen(*av)))) {
rule.fw_flg |= IP_FW_F_DENY; av++; ac--;
} else if (!strncmp(*av, "reject", strlen(*av))) {
rule.fw_flg |= IP_FW_F_REJECT; av++; ac--;
rule.fw_reject_code = ICMP_UNREACH_HOST;
} else if (!strncmp(*av, "reset", strlen(*av))) {
rule.fw_flg |= IP_FW_F_REJECT; av++; ac--;
rule.fw_reject_code = IP_FW_REJECT_RST; /* check TCP later */
} else if (!strncmp(*av, "unreach", strlen(*av))) {
rule.fw_flg |= IP_FW_F_REJECT; av++; ac--;
fill_reject_code(&rule.fw_reject_code, *av); av++; ac--;
} else if (!strncmp(*av, "check-state", strlen(*av))) {
rule.fw_flg |= IP_FW_F_CHECK_S; av++; ac--;
goto done;
} else {
errx(EX_DATAERR, "invalid action ``%s''", *av);
}
/* [log [logamount N]] -- log, optional */
if (ac && !strncmp(*av, "log", strlen(*av))) {
rule.fw_flg |= IP_FW_F_PRN; av++; ac--;
if (ac && !strncmp(*av, "logamount", strlen(*av))) {
ac--; av++;
NEED1("``logamount'' requires argument");
rule.fw_logamount = atoi(*av);
if (rule.fw_logamount < 0)
errx(EX_DATAERR, "``logamount'' argument must be positive");
if (rule.fw_logamount == 0)
rule.fw_logamount = -1;
ac--; av++;
}
}
/* protocol -- mandatory */
NEED1("missing protocol");
if (!strncmp(*av, "MAC", strlen(*av))) {
ac--;
av++;
/* need exactly 3 fields */
add_mac(&rule, ac, av); /* exits in case of errors */
ac -= 3;
av += 3;
goto do_options;
} else if ((proto = atoi(*av)) > 0) {
rule.fw_prot = proto; av++; ac--;
} else if (!strncmp(*av, "all", strlen(*av))) {
rule.fw_prot = IPPROTO_IP; av++; ac--;
} else if ((pe = getprotobyname(*av)) != NULL) {
rule.fw_prot = pe->p_proto; av++; ac--;
} else {
errx(EX_DATAERR, "invalid protocol ``%s''", *av);
}
if (rule.fw_prot != IPPROTO_TCP
&& (rule.fw_flg & IP_FW_F_COMMAND) == IP_FW_F_REJECT
&& rule.fw_reject_code == IP_FW_REJECT_RST)
errx(EX_DATAERR, "``reset'' is only valid for tcp packets");
/* from -- mandatory */
if (!ac || strncmp(*av, "from", strlen(*av)))
errx(EX_USAGE, "missing ``from''");
av++; ac--;
/* not -- optional */
if (ac && !strncmp(*av, "not", strlen(*av))) {
rule.fw_flg |= IP_FW_F_INVSRC;
av++; ac--;
}
NEED1("missing source address");
/* source -- mandatory */
if (!strncmp(*av, "me", strlen(*av))) {
rule.fw_flg |= IP_FW_F_SME;
av++; ac--;
} else {
fill_ip(&rule.fw_src, &rule.fw_smsk, &ac, &av);
}
/* ports -- optional */
if (ac && (isdigit(**av)
|| lookup_port(*av, rule.fw_prot, 1, 1) >= 0)) {
u_short nports = 0;
int retval;
retval = fill_port(&nports, rule.fw_uar.fw_pts,
0, *av, rule.fw_prot);
if (retval == 1)
rule.fw_flg |= IP_FW_F_SRNG;
else if (retval == 2)
rule.fw_flg |= IP_FW_F_SMSK;
IP_FW_SETNSRCP(&rule, nports);
av++; ac--;
}
/* to -- mandatory */
if (!ac || strncmp(*av, "to", strlen(*av)))
errx(EX_USAGE, "missing ``to''");
av++; ac--;
/* not -- optional */
if (ac && !strncmp(*av, "not", strlen(*av))) {
rule.fw_flg |= IP_FW_F_INVDST;
av++; ac--;
}
NEED1("missing dst address");
/* destination -- mandatory */
if (!strncmp(*av, "me", strlen(*av))) {
rule.fw_flg |= IP_FW_F_DME;
av++; ac--;
} else {
fill_ip(&rule.fw_dst, &rule.fw_dmsk, &ac, &av);
}
/* dest.ports -- optional */
if (ac && (isdigit(**av)
|| lookup_port(*av, rule.fw_prot, 1, 1) >= 0)) {
u_short nports = 0;
int retval;
retval = fill_port(&nports, rule.fw_uar.fw_pts,
IP_FW_GETNSRCP(&rule), *av, rule.fw_prot);
if (retval == 1)
rule.fw_flg |= IP_FW_F_DRNG;
else if (retval == 2)
rule.fw_flg |= IP_FW_F_DMSK;
IP_FW_SETNDSTP(&rule, nports);
av++; ac--;
}
if ((rule.fw_prot != IPPROTO_TCP) && (rule.fw_prot != IPPROTO_UDP)
&& (IP_FW_GETNSRCP(&rule) || IP_FW_GETNDSTP(&rule))) {
errx(EX_USAGE, "only TCP and UDP protocols are valid"
" with port specifications");
}
do_options:
while (ac) {
if (!strncmp(*av, "uid", strlen(*av))) {
struct passwd *pwd;
char *end;
uid_t uid;
rule.fw_flg |= IP_FW_F_UID;
ac--; av++;
NEED1("``uid'' requires argument");
uid = strtoul(*av, &end, 0);
if (*end == '\0')
pwd = getpwuid(uid);
else
pwd = getpwnam(*av);
if (pwd == NULL)
errx(EX_DATAERR, "uid \"%s\" is"
" nonexistent", *av);
rule.fw_uid = pwd->pw_uid;
ac--; av++;
} else if (!strncmp(*av, "gid", strlen(*av))) {
struct group *grp;
char *end;
gid_t gid;
rule.fw_flg |= IP_FW_F_GID;
ac--; av++;
NEED1("``gid'' requires argument");
gid = strtoul(*av, &end, 0);
if (*end == '\0')
grp = getgrgid(gid);
else
grp = getgrnam(*av);
if (grp == NULL)
errx(EX_DATAERR, "gid \"%s\" is"
" nonexistent", *av);
rule.fw_gid = grp->gr_gid;
ac--; av++;
} else if (!strncmp(*av, "in", strlen(*av))) {
rule.fw_flg |= IP_FW_F_IN;
av++; ac--;
} else if (!strncmp(*av,"limit",strlen(*av))) {
/* dyn. rule used to limit number of connections. */
rule.fw_flg |= IP_FW_F_KEEP_S;
rule.dyn_type = DYN_LIMIT ;
rule.limit_mask = 0 ;
av++; ac--;
for (; ac >1 ;) {
struct _s_x *p = limit_masks;
for ( ; p->x != 0 ; p++)
if (!strncmp(*av, p->s, strlen(*av))) {
rule.limit_mask |= p->x ;
av++; ac-- ;
break ;
}
if (p->s == NULL)
break ;
}
NEED1("limit needs mask and # of connections");
rule.conn_limit = atoi(*av);
if (rule.conn_limit == 0)
errx(EX_USAGE, "limit: limit must be >0");
if (rule.limit_mask == 0)
errx(EX_USAGE, "missing limit mask");
av++; ac--;
} else if (!strncmp(*av, "keep-state", strlen(*av))) {
u_long type;
rule.fw_flg |= IP_FW_F_KEEP_S;
av++; ac--;
if (ac > 0 && (type = atoi(*av)) != 0) {
rule.dyn_type = type;
av++; ac--;
}
} else if (!strncmp(*av, "bridged", strlen(*av))) {
rule.fw_flg |= IP_FW_BRIDGED;
av++; ac--;
} else if (!strncmp(*av, "out", strlen(*av))) {
rule.fw_flg |= IP_FW_F_OUT;
av++; ac--;
} else if (!strncmp(*av, "xmit", strlen(*av))) {
union ip_fw_if ifu;
int byname;
if (saw_via) {
badviacombo:
errx(EX_USAGE, "``via'' is incompatible"
" with ``xmit'' and ``recv''");
}
saw_xmrc = 1;
av++; ac--;
fill_iface("xmit", &ifu, &byname, ac, *av);
rule.fw_out_if = ifu;
rule.fw_flg |= IP_FW_F_OIFACE;
if (byname)
rule.fw_flg |= IP_FW_F_OIFNAME;
av++; ac--;
} else if (!strncmp(*av, "recv", strlen(*av))) {
union ip_fw_if ifu;
int byname;
if (saw_via)
goto badviacombo;
saw_xmrc = 1;
av++; ac--;
fill_iface("recv", &ifu, &byname, ac, *av);
rule.fw_in_if = ifu;
rule.fw_flg |= IP_FW_F_IIFACE;
if (byname)
rule.fw_flg |= IP_FW_F_IIFNAME;
av++; ac--;
} else if (!strncmp(*av, "via", strlen(*av))) {
union ip_fw_if ifu;
int byname = 0;
if (saw_xmrc)
goto badviacombo;
saw_via = 1;
av++; ac--;
fill_iface("via", &ifu, &byname, ac, *av);
rule.fw_out_if = rule.fw_in_if = ifu;
if (byname)
rule.fw_flg |=
(IP_FW_F_IIFNAME | IP_FW_F_OIFNAME);
av++; ac--;
} else if (!strncmp(*av, "fragment", strlen(*av))) {
rule.fw_flg |= IP_FW_F_FRAG;
av++; ac--;
} else if (!strncmp(*av, "ipoptions", strlen(*av))
|| !strncmp(*av, "ipopts", strlen(*av))) {
av++; ac--;
NEED1("missing argument for ``ipoptions''");
rule.fw_ipflg |= IP_FW_IF_IPOPT;
fill_flags(&rule.fw_ipopt, &rule.fw_ipnopt,
f_ipopts, av);
av++; ac--;
} else if (!strncmp(*av, "iplen", strlen(*av))) {
av++; ac--;
NEED1("missing argument for ``iplen''");
rule.fw_ipflg |= IP_FW_IF_IPLEN;
rule.fw_iplen = (u_short)strtoul(*av, NULL, 0);
av++; ac--;
} else if (!strncmp(*av, "ipid", strlen(*av))) {
unsigned long ipid;
char *c;
av++; ac--;
NEED1("missing argument for ``ipid''");
ipid = strtoul(*av, &c, 0);
if (*c != '\0')
errx(EX_DATAERR, "argument to ipid must"
" be numeric");
if (ipid > 65535)
errx(EX_DATAERR, "argument to ipid out"
" of range");
rule.fw_ipflg |= IP_FW_IF_IPID;
rule.fw_ipid = (u_short)ipid;
av++; ac--;
} else if (!strncmp(*av, "ipprecedence", strlen(*av))) {
u_long ippre;
char *c;
av++; ac--;
NEED1("missing argument for ``ipprecedence''");
ippre = strtoul(*av, &c, 0);
if (*c != '\0')
errx(EX_DATAERR, "argument to ipprecedence"
" must be numeric");
if (ippre > 7)
errx(EX_DATAERR, "argument to ipprecedence"
" out of range");
rule.fw_ipflg |= IP_FW_IF_IPPRE;
rule.fw_iptos |= (u_short)(ippre << 5);
av++; ac--;
} else if (!strncmp(*av, "iptos", strlen(*av))) {
av++; ac--;
NEED1("missing argument for ``iptos''");
rule.fw_ipflg |= IP_FW_IF_IPTOS;
fill_flags(&rule.fw_iptos, &rule.fw_ipntos,
f_iptos, av);
av++; ac--;
} else if (!strncmp(*av, "ipttl", strlen(*av))) {
av++; ac--;
NEED1("missing argument for ``ipttl''");
rule.fw_ipflg |= IP_FW_IF_IPTTL;
rule.fw_ipttl = (u_short)strtoul(*av, NULL, 0);
av++; ac--;
} else if (!strncmp(*av, "ipversion", strlen(*av))
|| !strncmp(*av, "ipver", strlen(*av))) {
av++; ac--;
NEED1("missing argument for ``ipversion''");
rule.fw_ipflg |= IP_FW_IF_IPVER;
rule.fw_ipver = (u_short)strtoul(*av, NULL, 0);
av++; ac--;
} else if (rule.fw_prot == IPPROTO_TCP) {
if (!strncmp(*av, "established", strlen(*av))) {
rule.fw_ipflg |= IP_FW_IF_TCPEST;
av++; ac--;
} else if (!strncmp(*av, "setup", strlen(*av))) {
rule.fw_tcpf |= TH_SYN;
rule.fw_tcpnf |= TH_ACK;
rule.fw_ipflg |= IP_FW_IF_TCPFLG;
av++; ac--;
} else if (!strncmp(*av, "tcpflags", strlen(*av))
|| !strncmp(*av, "tcpflgs", strlen(*av))) {
av++; ac--;
NEED1("missing argument for ``tcpflags''");
rule.fw_ipflg |= IP_FW_IF_TCPFLG;
fill_flags(&rule.fw_tcpf,
&rule.fw_tcpnf, f_tcpflags, av);
av++; ac--;
} else if (!strncmp(*av, "tcpoptions", strlen(*av))
|| !strncmp(*av, "tcpopts", strlen(*av))) {
av++; ac--;
NEED1("missing argument for ``tcpoptions''");
rule.fw_ipflg |= IP_FW_IF_TCPOPT;
fill_flags(&rule.fw_tcpopt,
&rule.fw_tcpnopt, f_tcpopts, av);
av++; ac--;
} else if (!strncmp(*av, "tcpseq", strlen(*av))) {
av++; ac--;
NEED1("missing argument for ``tcpseq''");
rule.fw_ipflg |= IP_FW_IF_TCPSEQ;
rule.fw_tcpseq =
htonl(strtoul(*av, NULL, 0));
av++; ac--;
} else if (!strncmp(*av, "tcpack", strlen(*av))) {
av++; ac--;
NEED1("missing argument for ``tcpack''");
rule.fw_ipflg |= IP_FW_IF_TCPACK;
rule.fw_tcpack =
htonl(strtoul(*av, NULL, 0));
av++; ac--;
} else if (!strncmp(*av, "tcpwin", strlen(*av))) {
av++; ac--;
NEED1("missing argument for ``tcpwin''");
rule.fw_ipflg |= IP_FW_IF_TCPWIN;
rule.fw_tcpwin =
htons((u_short)strtoul(*av, NULL, 0));
av++; ac--;
} else
errx(EX_USAGE, "unknown or out of order"
" argument ``%s''", *av);
} else if (rule.fw_prot == IPPROTO_ICMP) {
if (!strncmp(*av, "icmptypes", strlen(*av))) {
av++; ac--;
NEED1("missing argument for ``icmptypes''");
fill_icmptypes(rule.fw_uar.fw_icmptypes,
av, &rule.fw_flg);
av++; ac--;
} else
errx(EX_USAGE, "unknown or out of"
" order argument ``%s''", *av);
} else
errx(EX_USAGE, "unknown argument ``%s''", *av);
}
/* No direction specified -> do both directions */
if (!(rule.fw_flg & (IP_FW_F_OUT|IP_FW_F_IN)))
rule.fw_flg |= (IP_FW_F_OUT|IP_FW_F_IN);
/* Sanity check interface check, but handle "via" case separately */
if (saw_via) {
if (rule.fw_flg & IP_FW_F_IN)
rule.fw_flg |= IP_FW_F_IIFACE;
if (rule.fw_flg & IP_FW_F_OUT)
rule.fw_flg |= IP_FW_F_OIFACE;
} else if ((rule.fw_flg & IP_FW_F_OIFACE)
&& (rule.fw_flg & IP_FW_F_IN)) {
errx(EX_DATAERR, "can't check xmit interface of incoming"
" packets");
}
/* frag may not be used in conjunction with ports or TCP flags */
if (rule.fw_flg & IP_FW_F_FRAG) {
if (rule.fw_tcpf || rule.fw_tcpnf)
errx(EX_DATAERR, "can't mix 'frag' and tcpflags");
if (rule.fw_nports)
errx(EX_DATAERR, "can't mix 'frag' and port"
" specifications");
}
if (rule.fw_flg & IP_FW_F_PRN) {
if (!rule.fw_logamount) {
size_t len = sizeof(int);
if (sysctlbyname("net.inet.ip.fw.verbose_limit",
&rule.fw_logamount, &len, NULL, 0) == -1)
errx(1, "sysctlbyname(\"%s\")",
"net.inet.ip.fw.verbose_limit");
} else if (rule.fw_logamount == -1)
rule.fw_logamount = 0;
rule.fw_loghighest = rule.fw_logamount;
}
done:
i = sizeof(rule);
if (getsockopt(s, IPPROTO_IP, IP_FW_ADD, &rule, &i) == -1)
err(EX_UNAVAILABLE, "getsockopt(%s)", "IP_FW_ADD");
if (!do_quiet)
show_ipfw(&rule);
}
static void
zero (int ac, char *av[])
{
struct ip_fw rule;
int failed = EX_OK;
av++; ac--;
if (!ac) {
/* clear all entries */
if (setsockopt(s, IPPROTO_IP, IP_FW_ZERO, NULL, 0) < 0)
err(EX_UNAVAILABLE, "setsockopt(%s)", "IP_FW_ZERO");
if (!do_quiet)
printf("Accounting cleared.\n");
return;
}
memset(&rule, 0, sizeof rule);
while (ac) {
/* Rule number */
if (isdigit(**av)) {
rule.fw_number = atoi(*av); av++; ac--;
if (setsockopt(s, IPPROTO_IP,
IP_FW_ZERO, &rule, sizeof rule)) {
warn("rule %u: setsockopt(IP_FW_ZERO)",
rule.fw_number);
failed = EX_UNAVAILABLE;
} else if (!do_quiet)
printf("Entry %d cleared\n",
rule.fw_number);
} else {
errx(EX_USAGE, "invalid rule number ``%s''", *av);
}
}
if (failed != EX_OK)
exit(failed);
}
static void
resetlog (int ac, char *av[])
{
struct ip_fw rule;
int failed = EX_OK;
av++; ac--;
if (!ac) {
/* clear all entries */
if (setsockopt(s, IPPROTO_IP, IP_FW_RESETLOG, NULL, 0) < 0)
err(EX_UNAVAILABLE, "setsockopt(IP_FW_RESETLOG)");
if (!do_quiet)
printf("Logging counts reset.\n");
return;
}
memset(&rule, 0, sizeof rule);
while (ac) {
/* Rule number */
if (isdigit(**av)) {
rule.fw_number = atoi(*av); av++; ac--;
if (setsockopt(s, IPPROTO_IP,
IP_FW_RESETLOG, &rule, sizeof rule)) {
warn("rule %u: setsockopt(IP_FW_RESETLOG)",
rule.fw_number);
failed = EX_UNAVAILABLE;
} else if (!do_quiet)
printf("Entry %d logging count reset\n",
rule.fw_number);
} else {
errx(EX_DATAERR, "invalid rule number ``%s''", *av);
}
}
if (failed != EX_OK)
exit(failed);
}
static void
flush()
{
if (!do_force && !do_quiet) { /* need to ask user */
int c;
printf("Are you sure? [yn] ");
fflush(stdout);
do {
c = toupper(getc(stdin));
while (c != '\n' && getc(stdin) != '\n')
if (feof(stdin))
return; /* and do not flush */
} while (c != 'Y' && c != 'N');
printf("\n");
if (c == 'N') /* user said no */
return;
}
if (setsockopt(s, IPPROTO_IP,
do_pipe ? IP_DUMMYNET_FLUSH : IP_FW_FLUSH, NULL, 0) < 0)
err(EX_UNAVAILABLE, "setsockopt(IP_%s_FLUSH)",
do_pipe ? "DUMMYNET" : "FW");
if (!do_quiet)
printf("Flushed all %s.\n", do_pipe ? "pipes" : "rules");
}
static int
ipfw_main(int ac, char **av)
{
int ch;
if (ac == 1)
show_usage();
/* Set the force flag for non-interactive processes */
do_force = !isatty(STDIN_FILENO);
optind = optreset = 1;
while ((ch = getopt(ac, av, "s:adefNqtv")) != -1)
switch (ch) {
case 's': /* sort */
do_sort = atoi(optarg);
break;
case 'a':
do_acct = 1;
break;
case 'd':
do_dynamic = 1;
break;
case 'e':
do_expired = 1;
break;
case 'f':
do_force = 1;
break;
case 'N':
do_resolv = 1;
break;
case 'q':
do_quiet = 1;
break;
case 't':
do_time = 1;
break;
case 'v': /* verbose */
verbose++;
break;
default:
show_usage();
}
ac -= optind;
av += optind;
if (*av == NULL)
errx(EX_USAGE, "bad arguments, for usage summary ``ipfw''");
if (!strncmp(*av, "pipe", strlen(*av))) {
do_pipe = 1;
ac--;
av++;
} else if (!strncmp(*av, "queue", strlen(*av))) {
do_pipe = 2;
ac--;
av++;
}
if (!ac)
errx(EX_USAGE, "pipe requires arguments");
/* allow argument swapping -- pipe config xxx or pipe xxx config */
if (ac > 1 && *av[0] >= '0' && *av[0] <= '9') {
char *p = av[0];
av[0] = av[1];
av[1] = p;
}
if (!strncmp(*av, "add", strlen(*av)))
add(ac, av);
else if (do_pipe && !strncmp(*av, "config", strlen(*av)))
config_pipe(ac, av);
else if (!strncmp(*av, "delete", strlen(*av)))
delete(ac, av);
else if (!strncmp(*av, "flush", strlen(*av)))
flush();
else if (!strncmp(*av, "zero", strlen(*av)))
zero(ac, av);
else if (!strncmp(*av, "resetlog", strlen(*av)))
resetlog(ac, av);
else if (!strncmp(*av, "print", strlen(*av)))
list(--ac, ++av);
else if (!strncmp(*av, "list", strlen(*av)))
list(--ac, ++av);
else if (!strncmp(*av, "show", strlen(*av))) {
do_acct++;
list(--ac, ++av);
} else
errx(EX_USAGE, "bad command `%s'", *av);
return 0;
}
static void
ipfw_readfile(int ac, char *av[])
{
#define MAX_ARGS 32
#define WHITESP " \t\f\v\n\r"
char buf[BUFSIZ];
char *a, *p, *args[MAX_ARGS], *cmd = NULL;
char linename[10];
int i=0, lineno=0, qflag=0, pflag=0, status;
FILE *f = NULL;
pid_t preproc = 0;
int c;
while ((c = getopt(ac, av, "p:q")) != -1) {
switch(c) {
case 'p':
pflag = 1;
cmd = optarg;
args[0] = cmd;
i = 1;
break;
case 'q':
qflag = 1;
break;
default:
errx(EX_USAGE, "bad arguments, for usage"
" summary ``ipfw''");
}
if (pflag)
break;
}
if (pflag) {
/* Pass all but the last argument to the preprocessor. */
while (optind < ac - 1) {
if (i >= MAX_ARGS)
errx(EX_USAGE, "too many preprocessor options");
args[i++] = av[optind++];
}
}
av += optind;
ac -= optind;
if (ac != 1)
errx(EX_USAGE, "extraneous filename arguments");
if ((f = fopen(av[0], "r")) == NULL)
err(EX_UNAVAILABLE, "fopen: %s", av[0]);
if (pflag) {
/* pipe through preprocessor (cpp or m4) */
int pipedes[2];
args[i] = 0;
if (pipe(pipedes) == -1)
err(EX_OSERR, "cannot create pipe");
switch((preproc = fork())) {
case -1:
err(EX_OSERR, "cannot fork");
case 0:
/* child */
if (dup2(fileno(f), 0) == -1
|| dup2(pipedes[1], 1) == -1)
err(EX_OSERR, "dup2()");
fclose(f);
close(pipedes[1]);
close(pipedes[0]);
execvp(cmd, args);
err(EX_OSERR, "execvp(%s) failed", cmd);
default:
/* parent */
fclose(f);
close(pipedes[1]);
if ((f = fdopen(pipedes[0], "r")) == NULL) {
int savederrno = errno;
(void)kill(preproc, SIGTERM);
errno = savederrno;
err(EX_OSERR, "fdopen()");
}
}
}
while (fgets(buf, BUFSIZ, f)) {
lineno++;
sprintf(linename, "Line %d", lineno);
args[0] = linename;
if (*buf == '#')
continue;
if ((p = strchr(buf, '#')) != NULL)
*p = '\0';
i = 1;
if (qflag)
args[i++] = "-q";
for (a = strtok(buf, WHITESP);
a && i < MAX_ARGS; a = strtok(NULL, WHITESP), i++)
args[i] = a;
if (i == (qflag? 2: 1))
continue;
if (i == MAX_ARGS)
errx(EX_USAGE, "%s: too many arguments",
linename);
args[i] = NULL;
ipfw_main(i, args);
}
fclose(f);
if (pflag) {
if (waitpid(preproc, &status, 0) == -1)
errx(EX_OSERR, "waitpid()");
if (WIFEXITED(status) && WEXITSTATUS(status) != EX_OK)
errx(EX_UNAVAILABLE,
"preprocessor exited with status %d",
WEXITSTATUS(status));
else if (WIFSIGNALED(status))
errx(EX_UNAVAILABLE,
"preprocessor exited with signal %d",
WTERMSIG(status));
}
}
int
main(int ac, char *av[])
{
s = socket(AF_INET, SOCK_RAW, IPPROTO_RAW);
if (s < 0)
err(EX_UNAVAILABLE, "socket");
setbuf(stdout, 0);
/*
* Only interpret the last command line argument as a file to
* be preprocessed if it is specified as an absolute pathname.
*/
if (ac > 1 && av[ac - 1][0] == '/' && access(av[ac - 1], R_OK) == 0)
ipfw_readfile(ac, av);
else
ipfw_main(ac, av);
return EX_OK;
}