d/freebsd-dev
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

* Add new "flow" table type to support N=1..5-tuple lookups

Browse Source 
* Add "flow:hash" algorithm

Kernel changes:
* Add O_IP_FLOW_LOOKUP opcode to support "flow" lookups
* Add IPFW_TABLE_FLOW table type
* Add "struct tflow_entry" as strage for 6-tuple flows
* Add "flow:hash" algorithm. Basically it is auto-growing chained hash table.
  Additionally, we store mask of fields we need to compare in each instance/

* Increase ipfw_obj_tentry size by adding struct tflow_entry
* Add per-algorithm stat (ifpw_ta_tinfo) to ipfw_xtable_info
* Increase algoname length: 32 -> 64 (algo options passed there as string)
* Assume every table type can be customized by flags, use u8 to store "tflags" field.
* Simplify ipfw_find_table_entry() by providing @tentry directly to algo callback.
* Fix bug in cidr:chash resize procedure.

Userland changes:
* add "flow table(NAME)" syntax to support n-tuple checking tables.
* make fill_flags() separate function to ease working with _s_x arrays
* change "table info" output to reflect longer "type" fields

Syntax:
ipfw table fl2 create type flow:[src-ip][,proto][,src-port][,dst-ip][dst-port] [algo flow:hash]

Examples:

0:02 [2] zfscurr0# ipfw table fl2 create type flow:src-ip,proto,dst-port algo flow:hash
0:02 [2] zfscurr0# ipfw table fl2 info
+++ table(fl2), set(0) +++
 kindex: 0, type: flow:src-ip,proto,dst-port
 valtype: number, references: 0
 algorithm: flow:hash
 items: 0, size: 280
0:02 [2] zfscurr0# ipfw table fl2 add 2a02:6b8::333,tcp,443 45000
0:02 [2] zfscurr0# ipfw table fl2 add 10.0.0.92,tcp,80 22000
0:02 [2] zfscurr0# ipfw table fl2 list
+++ table(fl2), set(0) +++
2a02:6b8::333,6,443 45000
10.0.0.92,6,80 22000
0:02 [2] zfscurr0# ipfw add 200 count tcp from me to 78.46.89.105 80 flow 'table(fl2)'
00200 count tcp from me to 78.46.89.105 dst-port 80 flow table(fl2)
0:03 [2] zfscurr0# ipfw show
00200   0     0 count tcp from me to 78.46.89.105 dst-port 80 flow table(fl2)
65535 617 59416 allow ip from any to any
0:03 [2] zfscurr0# telnet -s 10.0.0.92 78.46.89.105 80
Trying 78.46.89.105...
..
0:04 [2] zfscurr0# ipfw show
00200   5   272 count tcp from me to 78.46.89.105 dst-port 80 flow table(fl2)
65535 682 66733 allow ip from any to any
This commit is contained in:
Alexander V. Chernikov 2014-07-31 20:08:19 +00:00
parent 8ce7a2bc03
commit 914bffb6ab
Notes: svn2git 2020-12-20 02:59:44 +00:00 
svn path=/projects/ipfw/; revision=269348
9 changed files with 1201 additions and 175 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

152
sbin/ipfw/ipfw2.c
View File

@ -364,6 +364,7 @@ static struct _s_x rule_options[] = {
{ "src-ipv6", TOK_SRCIP6},
{ "src-ip6", TOK_SRCIP6},
{ "lookup", TOK_LOOKUP},
{ "flow", TOK_FLOW},
{ "//", TOK_COMMENT },
{ "not", TOK_NOT }, /* pseudo option */
@ -706,6 +707,54 @@ concat_tokens(char *buf, size_t bufsize, struct _s_x *table, char *delimiter)
return (sz);
}
/*
* helper function to process a set of flags and set bits in the
* appropriate masks.
*/
void
fill_flags(struct _s_x *flags, char *p, uint8_t *set, uint8_t *clear)
{
char *q; /* points to the separator */
int val;
uint8_t *which; /* mask we are working on */
while (p && *p) {
if (*p == '!') {
p++;
which = clear;
} else
which = set;
q = strchr(p, ',');
if (q)
*q++ = '\0';
val = match_token(flags, p);
if (val <= 0)
errx(EX_DATAERR, "invalid flag %s", p);
*which |= (uint8_t)val;
p = q;
}
}
void
print_flags_buffer(char *buf, size_t sz, struct _s_x *list, uint8_t set)
{
char const *comma = "";
int i, l;
for (i = 0; list[i].x != 0; i++) {
if ((set & list[i].x) == 0)
continue;
set &= ~list[i].x;
l = snprintf(buf, sz, "%s%s", comma, list[i].s);
if (l >= sz)
return;
comma = ",";
buf += l;
sz -=l;
}
}
/*
* _substrcmp takes two strings and returns 1 if they do not match,
* and 0 if they match exactly or the first string is a sub-string
@ -1087,6 +1136,7 @@ print_flags(char const *name, ipfw_insn *cmd, struct _s_x *list)
}
}
/*
* Print the ip address contained in a command.
*/
@ -1795,6 +1845,18 @@ show_static_rule(struct cmdline_opts *co, struct format_opts *fo,
break;
}
case O_IP_FLOW_LOOKUP:
{
char *t;
t = table_search_ctlv(fo->tstate, cmd->arg1);
printf(" flow table(%s", t);
if (F_LEN(cmd) == F_INSN_SIZE(ipfw_insn_u32))
printf(",%u",
((ipfw_insn_u32 *)cmd)->d[0]);
printf(")");
break;
}
case O_IPID:
if (F_LEN(cmd) == 1)
printf(" ipid %u", cmd->arg1 );
@ -2660,6 +2722,33 @@ pack_table(struct tidx *tstate, char *name, uint32_t set)
return (ntlv->idx);
}
static void
fill_table(ipfw_insn *cmd, char *av, uint8_t opcode, struct tidx *tstate)
{
uint32_t *d = ((ipfw_insn_u32 *)cmd)->d;
uint16_t uidx;
char *p;
if ((p = strchr(av + 6, ')')) == NULL)
errx(EX_DATAERR, "forgotten parenthesis: '%s'", av);
*p = '\0';
p = strchr(av + 6, ',');
if (p)
*p++ = '\0';
if ((uidx = pack_table(tstate, av + 6, 0)) == 0)
errx(EX_DATAERR, "Invalid table name: %s", av + 6);
cmd->opcode = opcode;
cmd->arg1 = uidx;
if (p) {
cmd->len |= F_INSN_SIZE(ipfw_insn_u32);
d[0] = strtoul(p, NULL, 0);
} else
cmd->len |= F_INSN_SIZE(ipfw_insn);
}
/*
* fills the addr and mask fields in the instruction as appropriate from av.
* Update length as appropriate.
@ -2676,8 +2765,6 @@ fill_ip(ipfw_insn_ip *cmd, char *av, int cblen, struct tidx *tstate)
{
int len = 0;
uint32_t *d = ((ipfw_insn_u32 *)cmd)->d;
uint16_t uidx;
char *p;
cmd->o.len &= ~F_LEN_MASK; /* zero len */
@ -2690,23 +2777,7 @@ fill_ip(ipfw_insn_ip *cmd, char *av, int cblen, struct tidx *tstate)
}
if (strncmp(av, "table(", 6) == 0) {
if ((p = strchr(av + 6, ')')) == NULL)
errx(EX_DATAERR, "forgotten parenthesis: '%s'", av);
*p = '\0';
p = strchr(av + 6, ',');
if (p)
*p++ = '\0';
if ((uidx = pack_table(tstate, av + 6, 0)) == 0)
errx(EX_DATAERR, "Invalid table name: %s", av + 6);
cmd->o.opcode = O_IP_DST_LOOKUP;
cmd->o.arg1 = uidx;
if (p) {
cmd->o.len |= F_INSN_SIZE(ipfw_insn_u32);
d[0] = strtoul(p, NULL, 0);
} else
cmd->o.len |= F_INSN_SIZE(ipfw_insn);
fill_table(&cmd->o, av, O_IP_DST_LOOKUP, tstate);
return;
}
@ -2887,35 +2958,14 @@ n2mask(struct in6_addr *mask, int n)
return;
}
/*
* helper function to process a set of flags and set bits in the
* appropriate masks.
*/
static void
fill_flags(ipfw_insn *cmd, enum ipfw_opcodes opcode,
fill_flags_cmd(ipfw_insn *cmd, enum ipfw_opcodes opcode,
struct _s_x *flags, char *p)
{
uint8_t set=0, clear=0;
uint8_t set = 0, clear = 0;
while (p && *p) {
char *q; /* points to the separator */
int val;
uint8_t *which; /* mask we are working on */
fill_flags(flags, p, &set, &clear);
if (*p == '!') {
p++;
which = &clear;
} else
which = &set;
q = strchr(p, ',');
if (q)
*q++ = '\0';
val = match_token(flags, p);
if (val <= 0)
errx(EX_DATAERR, "invalid flag %s", p);
*which |= (uint8_t)val;
p = q;
}
cmd->opcode = opcode;
cmd->len = (cmd->len & (F_NOT | F_OR)) | 1;
cmd->arg1 = (set & 0xff) | ( (clear & 0xff) << 8);
@ -4087,13 +4137,13 @@ compile_rule(char *av[], uint32_t *rbuf, int *rbufsize, struct tidx *tstate)
case TOK_IPOPTS:
NEED1("missing argument for ipoptions");
fill_flags(cmd, O_IPOPT, f_ipopts, *av);
fill_flags_cmd(cmd, O_IPOPT, f_ipopts, *av);
av++;
break;
case TOK_IPTOS:
NEED1("missing argument for iptos");
fill_flags(cmd, O_IPTOS, f_iptos, *av);
fill_flags_cmd(cmd, O_IPTOS, f_iptos, *av);
av++;
break;
@ -4171,7 +4221,7 @@ compile_rule(char *av[], uint32_t *rbuf, int *rbufsize, struct tidx *tstate)
case TOK_TCPOPTS:
NEED1("missing argument for tcpoptions");
fill_flags(cmd, O_TCPOPTS, f_tcpopts, *av);
fill_flags_cmd(cmd, O_TCPOPTS, f_tcpopts, *av);
av++;
break;
@ -4198,7 +4248,7 @@ compile_rule(char *av[], uint32_t *rbuf, int *rbufsize, struct tidx *tstate)
case TOK_TCPFLAGS:
NEED1("missing argument for tcpflags");
cmd->opcode = O_TCPFLAGS;
fill_flags(cmd, O_TCPFLAGS, f_tcpflags, *av);
fill_flags_cmd(cmd, O_TCPFLAGS, f_tcpflags, *av);
av++;
break;
@ -4407,6 +4457,14 @@ compile_rule(char *av[], uint32_t *rbuf, int *rbufsize, struct tidx *tstate)
av++;
}
break;
case TOK_FLOW:
NEED1("missing table name");
if (strncmp(*av, "table(", 6) != 0)
errx(EX_DATAERR,
"enclose table name into \"table()\"");
fill_table(cmd, *av, O_IP_FLOW_LOOKUP, tstate);
av++;
break;
default:
errx(EX_USAGE, "unrecognised option [%d] %s\n", i, s);

5
sbin/ipfw/ipfw2.h
View File

@ -217,6 +217,7 @@ enum tokens {
TOK_DEL,
TOK_VALTYPE,
TOK_ALGO,
TOK_FLOW,
};
/*
* the following macro returns an error message if we run out of
@ -253,6 +254,10 @@ int match_token(struct _s_x *table, char *string);
char const *match_value(struct _s_x *p, int value);
size_t concat_tokens(char *buf, size_t bufsize, struct _s_x *table,
char *delimiter);
void fill_flags(struct _s_x *flags, char *p, uint8_t *set, uint8_t *clear);
void print_flags(char const *name, struct _s_x *list, uint8_t set,
uint8_t clear);
void print_flags_buffer(char *buf, size_t sz, struct _s_x *list, uint8_t set);
struct _ip_fw3_opheader;
int do_cmd(int optname, void *optval, uintptr_t optlen);

296
sbin/ipfw/tables.c
View File

@ -83,6 +83,7 @@ static struct _s_x tabletypes[] = {
{ "cidr", IPFW_TABLE_CIDR },
{ "iface", IPFW_TABLE_INTERFACE },
{ "number", IPFW_TABLE_NUMBER },
{ "flow", IPFW_TABLE_FLOW },
{ NULL, 0 }
};
@ -256,6 +257,59 @@ static struct _s_x tablenewcmds[] = {
{ NULL, 0 }
};
static struct _s_x flowtypecmds[] = {
{ "src-ip", IPFW_TFFLAG_SRCIP },
{ "proto", IPFW_TFFLAG_PROTO },
{ "src-port", IPFW_TFFLAG_SRCPORT },
{ "dst-ip", IPFW_TFFLAG_DSTIP },
{ "dst-port", IPFW_TFFLAG_DSTPORT },
{ NULL, 0 }
};
int
table_parse_type(uint8_t ttype, char *p, uint8_t *tflags)
{
uint8_t fset, fclear;
/* Parse type options */
switch(ttype) {
case IPFW_TABLE_FLOW:
fset = fclear = 0;
fill_flags(flowtypecmds, p, &fset,
&fclear);
*tflags = fset;
break;
default:
return (EX_USAGE);
}
return (0);
}
void
table_print_type(char *tbuf, size_t size, uint8_t type, uint8_t tflags)
{
const char *tname;
int l;
if ((tname = match_value(tabletypes, type)) == NULL)
tname = "unknown";
l = snprintf(tbuf, size, "%s", tname);
tbuf += l;
size -= l;
switch(type) {
case IPFW_TABLE_FLOW:
if (tflags != 0) {
*tbuf++ = ':';
l--;
print_flags_buffer(tbuf, size, flowtypecmds, tflags);
}
break;
}
}
/*
* Creates new table
*
@ -271,6 +325,7 @@ table_create(ipfw_obj_header *oh, int ac, char *av[])
ipfw_xtable_info xi;
int error, tcmd, val;
size_t sz;
char *p;
char tbuf[128];
sz = sizeof(tbuf);
@ -288,15 +343,25 @@ table_create(ipfw_obj_header *oh, int ac, char *av[])
switch (tcmd) {
case TOK_TYPE:
NEED1("table type required");
/* Type may have suboptions after ':' */
if ((p = strchr(*av, ':')) != NULL)
*p++ = '\0';
val = match_token(tabletypes, *av);
if (val != -1) {
xi.type = val;
ac--; av++;
break;
if (val == -1) {
concat_tokens(tbuf, sizeof(tbuf), tabletypes,
", ");
errx(EX_USAGE,
"Unknown tabletype: %s. Supported: %s",
*av, tbuf);
}
concat_tokens(tbuf, sizeof(tbuf), tabletypes, ", ");
errx(EX_USAGE, "Unknown tabletype: %s. Supported: %s",
*av, tbuf);
xi.type = val;
if (p != NULL) {
error = table_parse_type(val, p, &xi.tflags);
if (error != 0)
errx(EX_USAGE,
"Unsupported suboptions: %s", p);
}
ac--; av++;
break;
case TOK_VALTYPE:
NEED1("table value type required");
@ -408,15 +473,15 @@ table_get_info(ipfw_obj_header *oh, ipfw_xtable_info *i)
static int
table_show_info(ipfw_xtable_info *i, void *arg)
{
const char *ttype, *vtype;
const char *vtype;
char ttype[64];
printf("--- table(%s), set(%u) ---\n", i->tablename, i->set);
if ((ttype = match_value(tabletypes, i->type)) == NULL)
ttype = "unknown";
table_print_type(ttype, sizeof(ttype), i->type, i->tflags);
if ((vtype = match_value(tablevaltypes, i->vtype)) == NULL)
vtype = "unknown";
printf(" type: %s, kindex: %d\n", ttype, i->kidx);
printf("--- table(%s), set(%u) ---\n", i->tablename, i->set);
printf(" kindex: %d, type: %s\n", i->kidx, ttype);
printf(" valtype: %s, references: %u\n", vtype, i->refcnt);
printf(" algorithm: %s\n", i->algoname);
printf(" items: %u, size: %u\n", i->count, i->size);
@ -575,12 +640,15 @@ table_lookup(ipfw_obj_header *oh, int ac, char *av[])
{
ipfw_obj_tentry xtent;
ipfw_xtable_info xi;
char key[64];
int error;
if (ac == 0)
errx(EX_USAGE, "address required");
error = table_do_lookup(oh, *av, &xi, &xtent);
strlcpy(key, *av, sizeof(key));
error = table_do_lookup(oh, key, &xi, &xtent);
switch (error) {
case 0:
@ -600,12 +668,17 @@ table_lookup(ipfw_obj_header *oh, int ac, char *av[])
}
static void
tentry_fill_key_type(char *arg, ipfw_obj_tentry *tentry, uint8_t type)
tentry_fill_key_type(char *arg, ipfw_obj_tentry *tentry, uint8_t type,
uint8_t tflags)
{
char *p;
char *p, *pp;
int mask, af;
struct in6_addr *paddr;
struct in6_addr *paddr, tmp;
struct tflow_entry *tfe;
uint32_t key, *pkey;
uint16_t port;
struct protoent *pent;
struct servent *sent;
int masklen;
masklen = 0;
@ -664,6 +737,117 @@ tentry_fill_key_type(char *arg, ipfw_obj_tentry *tentry, uint8_t type)
*pkey = key;
masklen = 32;
break;
case IPFW_TABLE_FLOW:
/* Assume [src-ip][,proto][,src-port][,dst-ip][,dst-port] */
tfe = &tentry->k.flow;
af = 0;
/* Handle <ipv4|ipv6>*/
if ((tflags & IPFW_TFFLAG_SRCIP) != 0) {
if ((p = strchr(arg, ',')) != NULL)
*p++ = '\0';
/* Determine family using temporary storage */
if (inet_pton(AF_INET, arg, &tmp) == 1) {
if (af != 0 && af != AF_INET)
errx(EX_DATAERR,
"Inconsistent address family\n");
af = AF_INET;
memcpy(&tfe->a.a4.sip, &tmp, 4);
} else if (inet_pton(AF_INET6, arg, &tmp) == 1) {
if (af != 0 && af != AF_INET6)
errx(EX_DATAERR,
"Inconsistent address family\n");
af = AF_INET6;
memcpy(&tfe->a.a6.sip6, &tmp, 16);
}
arg = p;
}
/* Handle <proto-num|proto-name> */
if ((tflags & IPFW_TFFLAG_PROTO) != 0) {
if ((p = strchr(arg, ',')) != NULL)
*p++ = '\0';
key = strtol(arg, &pp, 10);
if (*pp != '\0') {
if ((pent = getprotobyname(arg)) == NULL)
errx(EX_DATAERR, "Unknown proto: %s",
arg);
else
key = pent->p_proto;
}
if (key > 255)
errx(EX_DATAERR, "Bad protocol number: %u",key);
tfe->proto = key;
arg = p;
}
/* Handle <port-num|service-name> */
if ((tflags & IPFW_TFFLAG_SRCPORT) != 0) {
if ((p = strchr(arg, ',')) != NULL)
*p++ = '\0';
if ((port = htons(strtol(arg, NULL, 10))) == 0) {
if ((sent = getservbyname(arg, NULL)) == NULL)
errx(EX_DATAERR, "Unknown service: %s",
arg);
else
key = sent->s_port;
}
tfe->sport = port;
arg = p;
}
/* Handle <ipv4|ipv6>*/
if ((tflags & IPFW_TFFLAG_DSTIP) != 0) {
if ((p = strchr(arg, ',')) != NULL)
*p++ = '\0';
/* Determine family using temporary storage */
if (inet_pton(AF_INET, arg, &tmp) == 1) {
if (af != 0 && af != AF_INET)
errx(EX_DATAERR,
"Inconsistent address family");
af = AF_INET;
memcpy(&tfe->a.a4.dip, &tmp, 4);
} else if (inet_pton(AF_INET6, arg, &tmp) == 1) {
if (af != 0 && af != AF_INET6)
errx(EX_DATAERR,
"Inconsistent address family");
af = AF_INET6;
memcpy(&tfe->a.a6.dip6, &tmp, 16);
}
arg = p;
}
/* Handle <port-num|service-name> */
if ((tflags & IPFW_TFFLAG_DSTPORT) != 0) {
if ((p = strchr(arg, ',')) != NULL)
*p++ = '\0';
if ((port = htons(strtol(arg, NULL, 10))) == 0) {
if ((sent = getservbyname(arg, NULL)) == NULL)
errx(EX_DATAERR, "Unknown service: %s",
arg);
else
key = sent->s_port;
}
tfe->dport = port;
arg = p;
}
tfe->af = af;
break;
default:
errx(EX_DATAERR, "Unsupported table type: %d", type);
}
@ -676,11 +860,12 @@ static void
tentry_fill_key(ipfw_obj_header *oh, ipfw_obj_tentry *tent, char *key,
uint8_t *ptype, uint8_t *pvtype, ipfw_xtable_info *xi)
{
uint8_t type, vtype;
uint8_t type, tflags, vtype;
int error;
char *del;
type = 0;
tflags = 0;
vtype = 0;
error = table_get_info(oh, xi);
@ -688,6 +873,7 @@ tentry_fill_key(ipfw_obj_header *oh, ipfw_obj_tentry *tent, char *key,
if (error == 0) {
/* Table found. */
type = xi->type;
tflags = xi->tflags;
vtype = xi->vtype;
} else {
if (error != ESRCH)
@ -718,7 +904,7 @@ tentry_fill_key(ipfw_obj_header *oh, ipfw_obj_tentry *tent, char *key,
*del = '/';
}
tentry_fill_key_type(key, tent, type);
tentry_fill_key_type(key, tent, type, tflags);
*ptype = type;
*pvtype = vtype;
@ -874,41 +1060,75 @@ table_show_list(ipfw_obj_header *oh, int need_header)
static void
table_show_entry(ipfw_xtable_info *i, ipfw_obj_tentry *tent)
{
char tbuf[128];
char *comma, tbuf[128], pval[32];
void *paddr;
uint32_t tval;
struct tflow_entry *tfe;
tval = tent->value;
if (co.do_value_as_ip) {
tval = htonl(tval);
inet_ntop(AF_INET, &tval, pval, sizeof(pval));
} else
snprintf(pval, sizeof(pval), "%u", tval);
switch (i->type) {
case IPFW_TABLE_CIDR:
/* IPv4 or IPv6 prefixes */
inet_ntop(tent->subtype, &tent->k, tbuf, sizeof(tbuf));
if (co.do_value_as_ip) {
tval = htonl(tval);
printf("%s/%u %s\n", tbuf, tent->masklen,
inet_ntoa(*(struct in_addr *)&tval));
} else
printf("%s/%u %u\n", tbuf, tent->masklen, tval);
printf("%s/%u %s\n", tbuf, tent->masklen, pval);
break;
case IPFW_TABLE_INTERFACE:
/* Interface names */
if (co.do_value_as_ip) {
tval = htonl(tval);
printf("%s %s\n", tent->k.iface,
inet_ntoa(*(struct in_addr *)&tval));
} else
printf("%s %u\n", tent->k.iface, tval);
printf("%s %s\n", tent->k.iface, pval);
break;
case IPFW_TABLE_NUMBER:
/* numbers */
if (co.do_value_as_ip) {
tval = htonl(tval);
printf("%u %s\n", tent->k.key,
inet_ntoa(*(struct in_addr *)&tval));
} else
printf("%u %u\n", tent->k.key, tval);
printf("%u %s\n", tent->k.key, pval);
break;
case IPFW_TABLE_FLOW:
/* flows */
tfe = &tent->k.flow;
comma = "";
if ((i->tflags & IPFW_TFFLAG_SRCIP) != 0) {
if (tfe->af == AF_INET)
paddr = &tfe->a.a4.sip;
else
paddr = &tfe->a.a6.sip6;
inet_ntop(tfe->af, paddr, tbuf, sizeof(tbuf));
printf("%s%s", comma, tbuf);
comma = ",";
}
if ((i->tflags & IPFW_TFFLAG_PROTO) != 0) {
printf("%s%d", comma, tfe->proto);
comma = ",";
}
if ((i->tflags & IPFW_TFFLAG_SRCPORT) != 0) {
printf("%s%d", comma, ntohs(tfe->sport));
comma = ",";
}
if ((i->tflags & IPFW_TFFLAG_DSTIP) != 0) {
if (tfe->af == AF_INET)
paddr = &tfe->a.a4.dip;
else
paddr = &tfe->a.a6.dip6;
inet_ntop(tfe->af, paddr, tbuf, sizeof(tbuf));
printf("%s%s", comma, tbuf);
comma = ",";
}
if ((i->tflags & IPFW_TFFLAG_DSTPORT) != 0) {
printf("%s%d", comma, ntohs(tfe->dport));
comma = ",";
}
printf(" %s\n", pval);
}
}

63
sys/netinet/ip_fw.h
View File

@ -262,6 +262,7 @@ enum ipfw_opcodes { /* arguments (4 byte each) */
O_DSCP, /* 2 u32 = DSCP mask */
O_SETDSCP, /* arg1=DSCP value */
O_IP_FLOW_LOOKUP, /* arg1=table number, u32=value */
O_LAST_OPCODE /* not an opcode! */
};
@ -675,7 +676,8 @@ struct _ipfw_dyn_rule {
#define IPFW_TABLE_CIDR 1 /* Table for holding IPv4/IPv6 prefixes */
#define IPFW_TABLE_INTERFACE 2 /* Table for holding interface names */
#define IPFW_TABLE_NUMBER 3 /* Table for holding ports/uid/gid/etc */
#define IPFW_TABLE_MAXTYPE 3 /* Maximum valid number */
#define IPFW_TABLE_FLOW 4 /* Table for holding flow data */
#define IPFW_TABLE_MAXTYPE 4 /* Maximum valid number */
#define IPFW_VTYPE_U32 1 /* Skipto/tablearg integer */
#define IPFW_VTYPE_IP 2 /* Nexthop IP address */
@ -743,6 +745,25 @@ typedef struct _ipfw_obj_ntlv {
char name[64]; /* Null-terminated name */
} ipfw_obj_ntlv;
/* IPv4/IPv6 L4 flow description */
struct tflow_entry {
uint8_t af;
uint8_t proto;
uint16_t spare;
uint16_t sport;
uint16_t dport;
union {
struct {
struct in_addr sip;
struct in_addr dip;
} a4;
struct {
struct in6_addr sip6;
struct in6_addr dip6;
} a6;
} a;
};
/* Table entry TLV */
typedef struct _ipfw_obj_tentry {
ipfw_obj_tlv head; /* TLV header */
@ -753,10 +774,11 @@ typedef struct _ipfw_obj_tentry {
uint64_t spare;
union {
/* Longest field needs to be aligned by 8-byte boundary */
struct in_addr addr; /* IPv4 address */
uint32_t key; /* uid/gid/port */
struct in6_addr addr6; /* IPv6 address */
struct in_addr addr; /* IPv4 address */
uint32_t key; /* uid/gid/port */
struct in6_addr addr6; /* IPv6 address */
char iface[IF_NAMESIZE]; /* interface name */
struct tflow_entry flow;
} k;
} ipfw_obj_tentry;
#define IPFW_TF_UPDATE 0x01 /* Update record if exists */
@ -776,19 +798,44 @@ typedef struct _ipfw_obj_ctlv {
uint8_t spare;
} ipfw_obj_ctlv;
typedef struct _ifpw_ta_tinfo {
uint32_t flags; /* Format flags */
uint8_t taclass; /* algorithm class */
uint8_t spare0;
uint16_t spare1;
uint32_t rssize4; /* runtime structure size */
uint32_t rcount4; /* number of items in runtime */
uint32_t rsize4; /* item size in runtime */
uint32_t rssize6; /* runtime structure size */
uint32_t rcount6; /* number of items in runtime */
uint32_t rsize6; /* item size in runtime */
} ifpw_ta_tinfo;
#define IPFW_TACLASS_HASH 1 /* algo is based on hash */
#define IPFW_TACLASS_ARRAY 2 /* algo is based on array */
#define IPFW_TACLASS_RADIX 3 /* algo is based on radix tree */
#define IPFW_TATFLAGS_DATA 0x0001 /* Has data filled in */
#define IPFW_TATFLAGS_AF 0x0002 /* Separate data per AF */
typedef struct _ipfw_xtable_info {
uint8_t type; /* table type (cidr,iface,..) */
uint8_t tflags; /* type flags */
uint8_t ftype; /* table value format type */
uint8_t vtype; /* value type */
uint16_t spare0;
uint32_t set; /* set table is in */
uint32_t kidx; /* kernel index */
uint32_t refcnt; /* number of references */
uint32_t count; /* Number of records */
uint32_t size; /* Total size of records */
uint32_t size; /* Total size of records(export)*/
char tablename[64]; /* table name */
char algoname[32]; /* algorithm name */
char algoname[64]; /* algorithm name */
ifpw_ta_tinfo ta_info; /* additional algo stats */
} ipfw_xtable_info;
#define IPFW_TFFLAG_SRCIP 0x01
#define IPFW_TFFLAG_DSTIP 0x02
#define IPFW_TFFLAG_SRCPORT 0x04
#define IPFW_TFFLAG_DSTPORT 0x08
#define IPFW_TFFLAG_PROTO 0x10
typedef struct _ipfw_iface_info {
char ifname[64]; /* interface name */
@ -801,7 +848,7 @@ typedef struct _ipfw_iface_info {
#define IPFW_IFFLAG_RESOLVED 0x01 /* Interface exists */
typedef struct _ipfw_ta_info {
char algoname[32]; /* algorithm name */
char algoname[64]; /* algorithm name */
uint32_t type; /* lookup type */
uint32_t flags;
uint32_t refcnt;

11
sys/netpfil/ipfw/ip_fw2.c
View File

@ -1522,6 +1522,17 @@ do { \
}
break;
case O_IP_FLOW_LOOKUP:
{
uint32_t v = 0;
match = ipfw_lookup_table_extended(chain,
cmd->arg1, 0, &args->f_id, &v);
if (cmdlen == F_INSN_SIZE(ipfw_insn_u32))
match = ((ipfw_insn_u32 *)cmd)->d[0] == v;
if (match)
tablearg = v;
}
break;
case O_IP_SRC_MASK:
case O_IP_DST_MASK:
if (is_ipv4) {

13
sys/netpfil/ipfw/ip_fw_sockopt.c
View File

@ -1011,6 +1011,17 @@ check_ipfw_rule_body(ipfw_insn *cmd, int cmd_len, struct rule_check_info *ci)
goto bad_size;
ci->table_opcodes++;
break;
case O_IP_FLOW_LOOKUP:
if (cmd->arg1 >= V_fw_tables_max) {
printf("ipfw: invalid table number %d\n",
cmd->arg1);
return (EINVAL);
}
if (cmdlen != F_INSN_SIZE(ipfw_insn) &&
cmdlen != F_INSN_SIZE(ipfw_insn_u32))
goto bad_size;
ci->table_opcodes++;
break;
case O_MACADDR2:
if (cmdlen != F_INSN_SIZE(ipfw_insn_mac))
goto bad_size;
@ -1726,7 +1737,7 @@ ipfw_ctl3(struct sockopt *sopt)
size_t bsize_max, size, valsize;
struct ip_fw_chain *chain;
uint32_t opt;
char xbuf[128];
char xbuf[256];
struct sockopt_data sdata;
ip_fw3_opheader *op3 = NULL;

81
sys/netpfil/ipfw/ip_fw_table.c
View File

@ -77,7 +77,8 @@ struct table_config {
struct named_object no;
uint8_t vtype; /* format table type */
uint8_t linked; /* 1 if already linked */
uint16_t spare;
uint8_t tflags; /* type flags */
uint8_t spare;
uint32_t count; /* Number of records */
uint64_t flags; /* state flags */
char tablename[64]; /* table name */
@ -95,11 +96,12 @@ struct tables_config {
static struct table_config *find_table(struct namedobj_instance *ni,
struct tid_info *ti);
static struct table_config *alloc_table_config(struct ip_fw_chain *ch,
struct tid_info *ti, struct table_algo *ta, char *adata, uint8_t vtype);
struct tid_info *ti, struct table_algo *ta, char *adata, uint8_t tflags,
uint8_t vtype);
static void free_table_config(struct namedobj_instance *ni,
struct table_config *tc);
static int create_table_internal(struct ip_fw_chain *ch, struct tid_info *ti,
char *aname, uint8_t vtype);
char *aname, uint8_t tflags, uint8_t vtype);
static void link_table(struct ip_fw_chain *chain, struct table_config *tc);
static void unlink_table(struct ip_fw_chain *chain, struct table_config *tc);
static void free_table_state(void **state, void **xstate, uint8_t type);
@ -169,7 +171,7 @@ add_table_entry(struct ip_fw_chain *ch, struct tid_info *ti,
if ((tei->flags & TEI_FLAGS_COMPAT) == 0)
return (ESRCH);
error = create_table_internal(ch, ti, NULL, IPFW_VTYPE_U32);
error = create_table_internal(ch, ti, NULL, 0, IPFW_VTYPE_U32);
if (error != 0)
return (error);
@ -533,8 +535,7 @@ ipfw_find_table_entry(struct ip_fw_chain *ch, ip_fw3_opheader *op3,
struct table_algo *ta;
struct table_info *kti;
struct namedobj_instance *ni;
int error, plen;
void *paddr;
int error;
size_t sz;
/* Check minimum header size */
@ -571,41 +572,13 @@ ipfw_find_table_entry(struct ip_fw_chain *ch, ip_fw3_opheader *op3,
return (EINVAL);
}
/* Check lookup key for validness */
plen = 0;
paddr = &tent->k;
switch (ti.type)
{
case IPFW_TABLE_CIDR:
if (tent->subtype == AF_INET)
plen = sizeof(struct in_addr);
else if (tent->subtype == AF_INET6)
plen = sizeof(struct in6_addr);
else {
IPFW_UH_RUNLOCK(ch);
return (EINVAL);
}
break;
case IPFW_TABLE_INTERFACE:
/* Check key first */
plen = sizeof(tent->k.iface);
if (strnlen(tent->k.iface, plen) == plen) {
IPFW_UH_RUNLOCK(ch);
return (EINVAL);
}
case IPFW_TABLE_NUMBER:
plen = sizeof(uint32_t);
break;
break;
default:
IPFW_UH_RUNLOCK(ch);
return (ENOTSUP);
}
kti = KIDX_TO_TI(ch, tc->no.kidx);
ta = tc->ta;
error = ta->find_tentry(tc->astate, kti, paddr, plen, tent);
if (ta->find_tentry == NULL)
return (ENOTSUP);
error = ta->find_tentry(tc->astate, kti, tent);
IPFW_UH_RUNLOCK(ch);
@ -651,9 +624,10 @@ flush_table(struct ip_fw_chain *ch, struct tid_info *ti)
struct table_algo *ta;
struct table_info ti_old, ti_new, *tablestate;
void *astate_old, *astate_new;
char algostate[32], *pstate;
char algostate[64], *pstate;
int error;
uint16_t kidx;
uint8_t tflags;
/*
* Stage 1: save table algoritm.
@ -674,13 +648,14 @@ flush_table(struct ip_fw_chain *ch, struct tid_info *ti)
pstate = algostate;
} else
pstate = NULL;
tflags = tc->tflags;
IPFW_UH_WUNLOCK(ch);
/*
* Stage 2: allocate new table instance using same algo.
*/
memset(&ti_new, 0, sizeof(struct table_info));
if ((error = ta->init(ch, &astate_new, &ti_new, pstate)) != 0) {
if ((error = ta->init(ch, &astate_new, &ti_new, pstate, tflags)) != 0) {
IPFW_UH_WLOCK(ch);
tc->no.refcnt--;
IPFW_UH_WUNLOCK(ch);
@ -1211,7 +1186,7 @@ ipfw_create_table(struct ip_fw_chain *ch, ip_fw3_opheader *op3,
}
IPFW_UH_RUNLOCK(ch);
return (create_table_internal(ch, &ti, aname, i->vtype));
return (create_table_internal(ch, &ti, aname, i->tflags, i->vtype));
}
/*
@ -1224,7 +1199,7 @@ ipfw_create_table(struct ip_fw_chain *ch, ip_fw3_opheader *op3,
*/
static int
create_table_internal(struct ip_fw_chain *ch, struct tid_info *ti,
char *aname, uint8_t vtype)
char *aname, uint8_t tflags, uint8_t vtype)
{
struct namedobj_instance *ni;
struct table_config *tc;
@ -1237,7 +1212,7 @@ create_table_internal(struct ip_fw_chain *ch, struct tid_info *ti,
if (ta == NULL)
return (ENOTSUP);
if ((tc = alloc_table_config(ch, ti, ta, aname, vtype)) == NULL)
if ((tc = alloc_table_config(ch, ti, ta, aname, tflags, vtype)) == NULL)
return (ENOMEM);
IPFW_UH_WLOCK(ch);
@ -1311,6 +1286,7 @@ export_table_info(struct ip_fw_chain *ch, struct table_config *tc,
struct table_info *ti;
i->type = tc->no.type;
i->tflags = tc->tflags;
i->vtype = tc->vtype;
i->set = tc->no.set;
i->kidx = tc->no.kidx;
@ -1605,6 +1581,10 @@ find_table_algo(struct tables_config *tcfg, struct tid_info *ti, char *name)
return (&cidr_radix);
case IPFW_TABLE_INTERFACE:
return (&iface_idx);
case IPFW_TABLE_NUMBER:
return (&number_array);
case IPFW_TABLE_FLOW:
return (&flow_hash);
}
return (NULL);
@ -1776,6 +1756,11 @@ classify_table_opcode(ipfw_insn *cmd, uint16_t *puidx, uint8_t *ptype)
*puidx = cmdif->p.glob;
skip = 0;
break;
case O_IP_FLOW_LOOKUP:
*puidx = cmd->arg1;
*ptype = IPFW_TABLE_FLOW;
skip = 0;
break;
}
return (skip);
@ -1803,6 +1788,9 @@ update_table_opcode(ipfw_insn *cmd, uint16_t idx)
cmdif = (ipfw_insn_if *)cmd;
cmdif->p.glob = idx;
break;
case O_IP_FLOW_LOOKUP:
cmd->arg1 = idx;
break;
}
}
@ -1906,7 +1894,7 @@ find_table(struct namedobj_instance *ni, struct tid_info *ti)
static struct table_config *
alloc_table_config(struct ip_fw_chain *ch, struct tid_info *ti,
struct table_algo *ta, char *aname, uint8_t vtype)
struct table_algo *ta, char *aname, uint8_t tflags, uint8_t vtype)
{
char *name, bname[16];
struct table_config *tc;
@ -1930,6 +1918,7 @@ alloc_table_config(struct ip_fw_chain *ch, struct tid_info *ti,
tc->no.name = tc->tablename;
tc->no.type = ti->type;
tc->no.set = set;
tc->tflags = tflags;
tc->ta = ta;
strlcpy(tc->tablename, name, sizeof(tc->tablename));
/* Set default value type to u32 for compability reasons */
@ -1944,7 +1933,7 @@ alloc_table_config(struct ip_fw_chain *ch, struct tid_info *ti,
}
/* Preallocate data structures for new tables */
error = ta->init(ch, &tc->astate, &tc->ti, aname);
error = ta->init(ch, &tc->astate, &tc->ti, aname, tflags);
if (error != 0) {
free(tc, M_IPFW);
return (NULL);
@ -2285,7 +2274,7 @@ ipfw_rewrite_table_uidx(struct ip_fw_chain *chain,
error = ENOTSUP;
goto free;
}
tc = alloc_table_config(chain, &ti, ta, NULL,
tc = alloc_table_config(chain, &ti, ta, NULL, 0,
IPFW_VTYPE_U32);
if (tc == NULL) {

16
sys/netpfil/ipfw/ip_fw_table.h
View File

@ -63,7 +63,7 @@ struct tentry_info {
#define TEI_FLAGS_COMPAT 0x04 /* Called from old ABI */
typedef int (ta_init)(struct ip_fw_chain *ch, void **ta_state,
struct table_info *ti, char *data);
struct table_info *ti, char *data, uint8_t tflags);
typedef void (ta_destroy)(void *ta_state, struct table_info *ti);
typedef int (ta_prepare_add)(struct ip_fw_chain *ch, struct tentry_info *tei,
void *ta_buf);
@ -92,8 +92,10 @@ typedef void ta_foreach(void *ta_state, struct table_info *ti, ta_foreach_f *f,
void *arg);
typedef int ta_dump_tentry(void *ta_state, struct table_info *ti, void *e,
ipfw_obj_tentry *tent);
typedef int ta_find_tentry(void *ta_state, struct table_info *ti, void *key,
uint32_t keylen, ipfw_obj_tentry *tent);
typedef int ta_find_tentry(void *ta_state, struct table_info *ti,
ipfw_obj_tentry *tent);
typedef int ta_dump_tinfo(void *ta_state, struct table_info *ti,
ifpw_ta_tinfo *tinfo);
struct table_algo {
char name[16];
@ -108,21 +110,23 @@ struct table_algo {
ta_add *add;
ta_del *del;
ta_flush_entry *flush_entry;
ta_find_tentry *find_tentry;
ta_prepare_mod *prepare_mod;
ta_fill_mod *fill_mod;
ta_modify *modify;
ta_flush_mod *flush_mod;
ta_change_ti *change_ti;
ta_foreach *foreach;
ta_dump_tentry *dump_tentry;
ta_print_config *print_config;
ta_find_tentry *find_tentry;
ta_change_ti *change_ti;
ta_dump_tinfo *dump_tinfo;
};
int ipfw_add_table_algo(struct ip_fw_chain *ch, struct table_algo *ta,
size_t size, int *idx);
void ipfw_del_table_algo(struct ip_fw_chain *ch, int idx);
extern struct table_algo cidr_radix, iface_idx;
extern struct table_algo cidr_radix, iface_idx, number_array, flow_hash;
void ipfw_table_algo_init(struct ip_fw_chain *chain);
void ipfw_table_algo_destroy(struct ip_fw_chain *chain);

739
sys/netpfil/ipfw/ip_fw_table_algo.c
View File

@ -149,7 +149,7 @@ ta_lookup_radix(struct table_info *ti, void *key, uint32_t keylen,
*/
static int
ta_init_radix(struct ip_fw_chain *ch, void **ta_state, struct table_info *ti,
char *data)
char *data, uint8_t tflags)
{
if (!rn_inithead(&ti->state, OFF_LEN_INET))
@ -221,23 +221,23 @@ ta_dump_radix_tentry(void *ta_state, struct table_info *ti, void *e,
}
static int
ta_find_radix_tentry(void *ta_state, struct table_info *ti, void *key,
uint32_t keylen, ipfw_obj_tentry *tent)
ta_find_radix_tentry(void *ta_state, struct table_info *ti,
ipfw_obj_tentry *tent)
{
struct radix_node_head *rnh;
void *e;
e = NULL;
if (keylen == sizeof(in_addr_t)) {
if (tent->subtype == AF_INET) {
struct sockaddr_in sa;
KEY_LEN(sa) = KEY_LEN_INET;
sa.sin_addr.s_addr = *((in_addr_t *)key);
sa.sin_addr.s_addr = tent->k.addr.s_addr;
rnh = (struct radix_node_head *)ti->state;
e = rnh->rnh_matchaddr(&sa, rnh);
} else {
struct sa_in6 sa6;
KEY_LEN(sa6) = KEY_LEN_INET6;
memcpy(&sa6.sin6_addr, key, sizeof(struct in6_addr));
memcpy(&sa6.sin6_addr, &tent->k.addr6, sizeof(struct in6_addr));
rnh = (struct radix_node_head *)ti->xstate;
e = rnh->rnh_matchaddr(&sa6, rnh);
}
@ -817,6 +817,17 @@ ta_print_chash_config(void *ta_state, struct table_info *ti, char *buf,
snprintf(buf, bufsize, "%s", "cidr:hash");
}
static int
log2(uint32_t v)
{
uint32_t r;
r = 0;
while (v >>= 1)
r++;
return (r);
}
/*
* New table.
@ -825,10 +836,10 @@ ta_print_chash_config(void *ta_state, struct table_info *ti, char *buf,
*/
static int
ta_init_chash(struct ip_fw_chain *ch, void **ta_state, struct table_info *ti,
char *data)
char *data, uint8_t tflags)
{
int error, i;
int v4, v6;
uint32_t hsize;
struct chash_cfg *ccfg;
ccfg = malloc(sizeof(struct chash_cfg), M_IPFW, M_WAITOK | M_ZERO);
@ -841,10 +852,8 @@ ta_init_chash(struct ip_fw_chain *ch, void **ta_state, struct table_info *ti,
return (error);
}
v4 = 6;
v6 = 6;
ccfg->size4 = 1 << v4;
ccfg->size6 = 1 << v6;
ccfg->size4 = 128;
ccfg->size6 = 128;
ccfg->head4 = malloc(sizeof(struct chashbhead) * ccfg->size4, M_IPFW,
M_WAITOK | M_ZERO);
@ -861,18 +870,19 @@ ta_init_chash(struct ip_fw_chain *ch, void **ta_state, struct table_info *ti,
ti->xstate = ccfg->head6;
/* Store data depending on v6 mask length */
hsize = log2(ccfg->size4) << 8 | log2(ccfg->size6);
if (ccfg->mask6 == 64) {
ti->data = (32 - ccfg->mask4) << 24 | (128 - ccfg->mask6) << 16 |
v4 << 8 | v6;
ti->data = (32 - ccfg->mask4) << 24 | (128 - ccfg->mask6) << 16|
hsize;
ti->lookup = ta_lookup_chash_64;
} else if ((ccfg->mask6 % 8) == 0) {
ti->data = (32 - ccfg->mask4) << 24 |
ccfg->mask6 << 13 | v4 << 8 | v6;
ccfg->mask6 << 13 | hsize;
ti->lookup = ta_lookup_chash_aligned;
} else {
/* don't do that! */
ti->data = (32 - ccfg->mask4) << 24 |
ccfg->mask6 << 16 | v4 << 8 | v6;
ccfg->mask6 << 16 | hsize;
ti->lookup = ta_lookup_chash_slow;
}
@ -986,8 +996,8 @@ tei_to_chash_ent(struct tentry_info *tei, struct chashentry *ent)
static int
ta_find_chash_tentry(void *ta_state, struct table_info *ti, void *key,
uint32_t keylen, ipfw_obj_tentry *tent)
ta_find_chash_tentry(void *ta_state, struct table_info *ti,
ipfw_obj_tentry *tent)
{
struct chash_cfg *ccfg;
struct chashbhead *head;
@ -1001,8 +1011,8 @@ ta_find_chash_tentry(void *ta_state, struct table_info *ti, void *key,
memset(&ent, 0, sizeof(ent));
memset(&tei, 0, sizeof(tei));
if (keylen == sizeof(in_addr_t)) {
tei.paddr = key;
if (tent->subtype == AF_INET) {
tei.paddr = &tent->k.addr;
tei.masklen = ccfg->mask4;
tei.subtype = AF_INET;
@ -1020,7 +1030,7 @@ ta_find_chash_tentry(void *ta_state, struct table_info *ti, void *key,
return (0);
}
} else {
tei.paddr = key;
tei.paddr = &tent->k.addr6;
tei.masklen = ccfg->mask6;
tei.subtype = AF_INET6;
@ -1330,6 +1340,9 @@ ta_modify_chash(void *ta_state, struct table_info *ti, void *ta_buf,
ccfg->size6 = mi->size;
}
ti->data = (ti->data & 0xFFFFFFFF00000000) | log2(ccfg->size4) << 8 |
log2(ccfg->size6);
mi->main_ptr = old_head;
return (0);
@ -1539,7 +1552,7 @@ ta_lookup_ifidx(struct table_info *ti, void *key, uint32_t keylen,
static int
ta_init_ifidx(struct ip_fw_chain *ch, void **ta_state, struct table_info *ti,
char *data)
char *data, uint8_t tflags)
{
struct iftable_cfg *icfg;
@ -1949,15 +1962,15 @@ ta_dump_ifidx_tentry(void *ta_state, struct table_info *ti, void *e,
}
static int
ta_find_ifidx_tentry(void *ta_state, struct table_info *ti, void *key,
uint32_t keylen, ipfw_obj_tentry *tent)
ta_find_ifidx_tentry(void *ta_state, struct table_info *ti,
ipfw_obj_tentry *tent)
{
struct iftable_cfg *icfg;
struct ifentry *ife;
char *ifname;
icfg = (struct iftable_cfg *)ta_state;
ifname = (char *)key;
ifname = tent->k.iface;
if (strnlen(ifname, IF_NAMESIZE) == IF_NAMESIZE)
return (EINVAL);
@ -2098,7 +2111,7 @@ ta_lookup_numarray(struct table_info *ti, void *key, uint32_t keylen,
static int
ta_init_numarray(struct ip_fw_chain *ch, void **ta_state, struct table_info *ti,
char *data)
char *data, uint8_t tflags)
{
struct numarray_cfg *cfg;
@ -2335,15 +2348,15 @@ ta_dump_numarray_tentry(void *ta_state, struct table_info *ti, void *e,
}
static int
ta_find_numarray_tentry(void *ta_state, struct table_info *ti, void *key,
uint32_t keylen, ipfw_obj_tentry *tent)
ta_find_numarray_tentry(void *ta_state, struct table_info *ti,
ipfw_obj_tentry *tent)
{
struct numarray_cfg *cfg;
struct numarray *ri;
cfg = (struct numarray_cfg *)ta_state;
ri = numarray_find(ti, key);
ri = numarray_find(ti, &tent->k.key);
if (ri != NULL) {
ta_dump_numarray_tentry(ta_state, ti, ri, tent);
@ -2387,6 +2400,672 @@ struct table_algo number_array = {
.flush_mod = ta_flush_mod_numarray,
};
/*
* flow:hash cmds
*
*
* ti->data:
* [inv.mask4][inv.mask6][log2hsize4][log2hsize6]
* [ 8][ 8[ 8][ 8]
*
* inv.mask4: 32 - mask
* inv.mask6:
* 1) _slow lookup: mask
* 2) _aligned: (128 - mask) / 8
* 3) _64: 8
*
*
* pflags:
* [v4=1/v6=0][hsize]
* [ 32][ 32]
*/
struct fhashentry;
SLIST_HEAD(fhashbhead, fhashentry);
/*
struct tflow_entry {
uint8_t af;
uint8_t proto;
uint16_t spare;
uint16_t dport;
uint16_t sport;
union {
struct {
struct in_addr sip;
struct in_addr dip;
} v4;
struct {
struct in6_addr sip6;
struct in6_addr dip6;
} v6;
} a;
};
*/
struct fhashentry {
SLIST_ENTRY(fhashentry) next;
uint8_t af;
uint8_t proto;
uint16_t spare0;
uint16_t dport;
uint16_t sport;
uint32_t value;
uint32_t spare1;
};
struct fhashentry4 {
struct fhashentry e;
struct in_addr dip;
struct in_addr sip;
};
struct fhashentry6 {
struct fhashentry e;
struct in6_addr dip6;
struct in6_addr sip6;
};
struct fhash_cfg {
struct fhashbhead *head;
size_t size;
size_t items;
struct fhashentry4 fe4;
struct fhashentry6 fe6;
};
static __inline int
cmp_flow_ent(struct fhashentry *a, struct fhashentry *b, size_t sz)
{
uint64_t *ka, *kb;
ka = (uint64_t *)(&a->next + 1);
kb = (uint64_t *)(&b->next + 1);
if (*ka == *kb && (memcmp(a + 1, b + 1, sz) == 0))
return (1);
return (0);
}
static __inline uint32_t
hash_flow4(struct fhashentry4 *f, int hsize)
{
uint32_t i;
i = (f->dip.s_addr) ^ (f->sip.s_addr) ^ (f->e.dport) ^ (f->e.sport);
return (i % (hsize - 1));
}
static __inline uint32_t
hash_flow6(struct fhashentry6 *f, int hsize)
{
uint32_t i;
i = (f->dip6.__u6_addr.__u6_addr32[2]) ^
(f->dip6.__u6_addr.__u6_addr32[3]) ^
(f->sip6.__u6_addr.__u6_addr32[2]) ^
(f->sip6.__u6_addr.__u6_addr32[3]) ^
(f->e.dport) ^ (f->e.sport);
return (i % (hsize - 1));
}
static uint32_t
hash_flow_ent(struct fhashentry *ent, uint32_t size)
{
uint32_t hash;
if (ent->af == AF_INET) {
hash = hash_flow4((struct fhashentry4 *)ent, size);
} else {
hash = hash_flow6((struct fhashentry6 *)ent, size);
}
return (hash);
}
static int
ta_lookup_fhash(struct table_info *ti, void *key, uint32_t keylen,
uint32_t *val)
{
struct fhashbhead *head;
struct fhashentry *ent;
struct fhashentry4 *m4;
struct ipfw_flow_id *id;
uint16_t hash, hsize;
id = (struct ipfw_flow_id *)key;
head = (struct fhashbhead *)ti->state;
hsize = ti->data;
m4 = (struct fhashentry4 *)ti->xstate;
if (id->addr_type == 4) {
struct fhashentry4 f;
/* Copy hash mask */
f = *m4;
f.dip.s_addr &= id->dst_ip;
f.sip.s_addr &= id->src_ip;
f.e.dport &= id->dst_port;
f.e.sport &= id->src_port;
f.e.proto &= id->proto;
hash = hash_flow4(&f, hsize);
SLIST_FOREACH(ent, &head[hash], next) {
if (cmp_flow_ent(ent, &f.e, 2 * 4) != 0) {
*val = ent->value;
return (1);
}
}
} else if (id->addr_type == 6) {
struct fhashentry6 f;
uint64_t *fp, *idp;
/* Copy hash mask */
f = *((struct fhashentry6 *)(m4 + 1));
/* Handle lack of __u6_addr.__u6_addr64 */
fp = (uint64_t *)&f.dip6;
idp = (uint64_t *)&id->dst_ip6;
/* src IPv6 is stored after dst IPv6 */
*fp++ &= *idp++;
*fp++ &= *idp++;
*fp++ &= *idp++;
*fp &= *idp;
f.e.dport &= id->dst_port;
f.e.sport &= id->src_port;
f.e.proto &= id->proto;
hash = hash_flow6(&f, hsize);
SLIST_FOREACH(ent, &head[hash], next) {
if (cmp_flow_ent(ent, &f.e, 2 * 16) != 0) {
*val = ent->value;
return (1);
}
}
}
return (0);
}
/*
* New table.
*/
static int
ta_init_fhash(struct ip_fw_chain *ch, void **ta_state, struct table_info *ti,
char *data, uint8_t tflags)
{
int i;
struct fhash_cfg *cfg;
struct fhashentry4 *fe4;
struct fhashentry6 *fe6;
cfg = malloc(sizeof(struct fhash_cfg), M_IPFW, M_WAITOK | M_ZERO);
cfg->size = 512;
cfg->head = malloc(sizeof(struct fhashbhead) * cfg->size, M_IPFW,
M_WAITOK | M_ZERO);
for (i = 0; i < cfg->size; i++)
SLIST_INIT(&cfg->head[i]);
/* Fill in fe masks based on @tflags */
fe4 = &cfg->fe4;
fe6 = &cfg->fe6;
if (tflags & IPFW_TFFLAG_SRCIP) {
memset(&fe4->sip, 0xFF, sizeof(fe4->sip));
memset(&fe6->sip6, 0xFF, sizeof(fe6->sip6));
}
if (tflags & IPFW_TFFLAG_DSTIP) {
memset(&fe4->dip, 0xFF, sizeof(fe4->dip));
memset(&fe6->dip6, 0xFF, sizeof(fe6->dip6));
}
if (tflags & IPFW_TFFLAG_SRCPORT) {
memset(&fe4->e.sport, 0xFF, sizeof(fe4->e.sport));
memset(&fe6->e.sport, 0xFF, sizeof(fe6->e.sport));
}
if (tflags & IPFW_TFFLAG_DSTPORT) {
memset(&fe4->e.dport, 0xFF, sizeof(fe4->e.dport));
memset(&fe6->e.dport, 0xFF, sizeof(fe6->e.dport));
}
if (tflags & IPFW_TFFLAG_PROTO) {
memset(&fe4->e.proto, 0xFF, sizeof(fe4->e.proto));
memset(&fe6->e.proto, 0xFF, sizeof(fe6->e.proto));
}
fe4->e.af = AF_INET;
fe6->e.af = AF_INET6;
*ta_state = cfg;
ti->state = cfg->head;
ti->xstate = &cfg->fe4;
ti->data = cfg->size;
ti->lookup = ta_lookup_fhash;
return (0);
}
static void
ta_destroy_fhash(void *ta_state, struct table_info *ti)
{
struct fhash_cfg *cfg;
struct fhashentry *ent, *ent_next;
int i;
cfg = (struct fhash_cfg *)ta_state;
for (i = 0; i < cfg->size; i++)
SLIST_FOREACH_SAFE(ent, &cfg->head[i], next, ent_next)
free(ent, M_IPFW_TBL);
free(cfg->head, M_IPFW);
free(cfg, M_IPFW);
}
static int
ta_dump_fhash_tentry(void *ta_state, struct table_info *ti, void *e,
ipfw_obj_tentry *tent)
{
struct fhash_cfg *cfg;
struct fhashentry *ent;
struct fhashentry4 *fe4;
struct fhashentry6 *fe6;
struct tflow_entry *tfe;
cfg = (struct fhash_cfg *)ta_state;
ent = (struct fhashentry *)e;
tfe = &tent->k.flow;
tfe->af = ent->af;
tfe->proto = ent->proto;
tfe->dport = htons(ent->dport);
tfe->sport = htons(ent->sport);
tent->value = ent->value;
tent->subtype = ent->af;
if (ent->af == AF_INET) {
fe4 = (struct fhashentry4 *)ent;
tfe->a.a4.sip.s_addr = htonl(fe4->sip.s_addr);
tfe->a.a4.dip.s_addr = htonl(fe4->dip.s_addr);
tent->masklen = 32;
#ifdef INET6
} else {
fe6 = (struct fhashentry6 *)ent;
tfe->a.a6.sip6 = fe6->sip6;
tfe->a.a6.dip6 = fe6->dip6;
tent->masklen = 128;
#endif
}
return (0);
}
static int
tei_to_fhash_ent(struct tentry_info *tei, struct fhashentry *ent)
{
struct fhashentry4 *fe4;
struct fhashentry6 *fe6;
struct tflow_entry *tfe;
tfe = (struct tflow_entry *)tei->paddr;
ent->af = tei->subtype;
ent->proto = tfe->proto;
ent->value = tei->value;
ent->dport = ntohs(tfe->dport);
ent->sport = ntohs(tfe->sport);
if (tei->subtype == AF_INET) {
#ifdef INET
fe4 = (struct fhashentry4 *)ent;
fe4->sip.s_addr = ntohl(tfe->a.a4.sip.s_addr);
fe4->dip.s_addr = ntohl(tfe->a.a4.dip.s_addr);
#endif
#ifdef INET6
} else if (tei->subtype == AF_INET6) {
fe6 = (struct fhashentry6 *)ent;
fe6->sip6 = tfe->a.a6.sip6;
fe6->dip6 = tfe->a.a6.dip6;
#endif
} else {
/* Unknown CIDR type */
return (EINVAL);
}
return (0);
}
static int
ta_find_fhash_tentry(void *ta_state, struct table_info *ti,
ipfw_obj_tentry *tent)
{
struct fhash_cfg *cfg;
struct fhashbhead *head;
struct fhashentry *ent, *tmp;
struct fhashentry6 fe6;
struct tentry_info tei;
int error;
uint32_t hash;
size_t sz;
cfg = (struct fhash_cfg *)ta_state;
ent = &fe6.e;
memset(&fe6, 0, sizeof(fe6));
memset(&tei, 0, sizeof(tei));
tei.paddr = &tent->k.flow;
tei.subtype = tent->subtype;
if ((error = tei_to_fhash_ent(&tei, ent)) != 0)
return (error);
head = cfg->head;
hash = hash_flow_ent(ent, cfg->size);
if (tei.subtype == AF_INET)
sz = 2 * sizeof(struct in_addr);
else
sz = 2 * sizeof(struct in6_addr);
/* Check for existence */
SLIST_FOREACH(tmp, &head[hash], next) {
if (cmp_flow_ent(tmp, ent, sz) != 0) {
ta_dump_fhash_tentry(ta_state, ti, tmp, tent);
return (0);
}
}
return (ENOENT);
}
static void
ta_foreach_fhash(void *ta_state, struct table_info *ti, ta_foreach_f *f,
void *arg)
{
struct fhash_cfg *cfg;
struct fhashentry *ent, *ent_next;
int i;
cfg = (struct fhash_cfg *)ta_state;
for (i = 0; i < cfg->size; i++)
SLIST_FOREACH_SAFE(ent, &cfg->head[i], next, ent_next)
f(ent, arg);
}
struct ta_buf_fhash
{
void *ent_ptr;
struct fhashentry6 fe6;
};
static int
ta_prepare_add_fhash(struct ip_fw_chain *ch, struct tentry_info *tei,
void *ta_buf)
{
struct ta_buf_fhash *tb;
struct fhashentry *ent;
size_t sz;
int error;
tb = (struct ta_buf_fhash *)ta_buf;
memset(tb, 0, sizeof(struct ta_buf_fhash));
if (tei->subtype == AF_INET)
sz = sizeof(struct fhashentry4);
else if (tei->subtype == AF_INET6)
sz = sizeof(struct fhashentry6);
else
return (EINVAL);
ent = malloc(sz, M_IPFW_TBL, M_WAITOK | M_ZERO);
error = tei_to_fhash_ent(tei, ent);
if (error != 0) {
free(ent, M_IPFW_TBL);
return (error);
}
tb->ent_ptr = ent;
return (0);
}
static int
ta_add_fhash(void *ta_state, struct table_info *ti, struct tentry_info *tei,
void *ta_buf, uint64_t *pflags, uint32_t *pnum)
{
struct fhash_cfg *cfg;
struct fhashbhead *head;
struct fhashentry *ent, *tmp;
struct ta_buf_fhash *tb;
int exists;
uint32_t hash;
size_t sz;
cfg = (struct fhash_cfg *)ta_state;
tb = (struct ta_buf_fhash *)ta_buf;
ent = (struct fhashentry *)tb->ent_ptr;
exists = 0;
head = cfg->head;
hash = hash_flow_ent(ent, cfg->size);
if (tei->subtype == AF_INET)
sz = 2 * sizeof(struct in_addr);
else
sz = 2 * sizeof(struct in6_addr);
/* Check for existence */
SLIST_FOREACH(tmp, &head[hash], next) {
if (cmp_flow_ent(tmp, ent, sz) != 0) {
exists = 1;
break;
}
}
if (exists == 1) {
if ((tei->flags & TEI_FLAGS_UPDATE) == 0)
return (EEXIST);
/* Record already exists. Update value if we're asked to */
tmp->value = tei->value;
/* Indicate that update has happened instead of addition */
tei->flags |= TEI_FLAGS_UPDATED;
*pnum = 0;
} else {
SLIST_INSERT_HEAD(&head[hash], ent, next);
tb->ent_ptr = NULL;
*pnum = 1;
/* Update counters and check if we need to grow hash */
cfg->items++;
if (cfg->items > cfg->size && cfg->size < 65536)
*pflags = cfg->size * 2;
}
return (0);
}
static int
ta_prepare_del_fhash(struct ip_fw_chain *ch, struct tentry_info *tei,
void *ta_buf)
{
struct ta_buf_fhash *tb;
tb = (struct ta_buf_fhash *)ta_buf;
memset(tb, 0, sizeof(struct ta_buf_fhash));
return (tei_to_fhash_ent(tei, &tb->fe6.e));
}
static int
ta_del_fhash(void *ta_state, struct table_info *ti, struct tentry_info *tei,
void *ta_buf, uint64_t *pflags, uint32_t *pnum)
{
struct fhash_cfg *cfg;
struct fhashbhead *head;
struct fhashentry *ent, *tmp;
struct ta_buf_fhash *tb;
uint32_t hash;
size_t sz;
cfg = (struct fhash_cfg *)ta_state;
tb = (struct ta_buf_fhash *)ta_buf;
ent = &tb->fe6.e;
head = cfg->head;
hash = hash_flow_ent(ent, cfg->size);
if (tei->subtype == AF_INET)
sz = 2 * sizeof(struct in_addr);
else
sz = 2 * sizeof(struct in6_addr);
/* Check for existence */
SLIST_FOREACH(tmp, &head[hash], next) {
if (cmp_flow_ent(tmp, ent, sz) != 0) {
SLIST_REMOVE(&head[hash], tmp, fhashentry, next);
*pnum = 1;
cfg->items--;
return (0);
}
}
return (ENOENT);
}
static void
ta_flush_fhash_entry(struct ip_fw_chain *ch, struct tentry_info *tei,
void *ta_buf)
{
struct ta_buf_fhash *tb;
tb = (struct ta_buf_fhash *)ta_buf;
if (tb->ent_ptr != NULL)
free(tb->ent_ptr, M_IPFW_TBL);
}
/*
* Hash growing callbacks.
*/
/*
* Allocate new, larger fhash.
*/
static int
ta_prepare_mod_fhash(void *ta_buf, uint64_t *pflags)
{
struct mod_item *mi;
struct fhashbhead *head;
int i;
mi = (struct mod_item *)ta_buf;
memset(mi, 0, sizeof(struct mod_item));
mi->size = *pflags;
head = malloc(sizeof(struct fhashbhead) * mi->size, M_IPFW,
M_WAITOK | M_ZERO);
for (i = 0; i < mi->size; i++)
SLIST_INIT(&head[i]);
mi->main_ptr = head;
return (0);
}
/*
* Copy data from old runtime array to new one.
*/
static int
ta_fill_mod_fhash(void *ta_state, struct table_info *ti, void *ta_buf,
uint64_t *pflags)
{
/* In is not possible to do rehash if we're not holidng WLOCK. */
return (0);
}
/*
* Switch old & new arrays.
*/
static int
ta_modify_fhash(void *ta_state, struct table_info *ti, void *ta_buf,
uint64_t pflags)
{
struct mod_item *mi;
struct fhash_cfg *cfg;
struct fhashbhead *old_head, *new_head;
struct fhashentry *ent, *ent_next;
int i;
uint32_t nhash;
size_t old_size;
mi = (struct mod_item *)ta_buf;
cfg = (struct fhash_cfg *)ta_state;
/* Check which hash we need to grow and do we still need that */
old_size = cfg->size;
old_head = ti->state;
if (old_size >= mi->size)
return (0);
new_head = (struct fhashbhead *)mi->main_ptr;
for (i = 0; i < old_size; i++) {
SLIST_FOREACH_SAFE(ent, &old_head[i], next, ent_next) {
nhash = hash_flow_ent(ent, mi->size);
SLIST_INSERT_HEAD(&new_head[nhash], ent, next);
}
}
ti->state = new_head;
ti->data = mi->size;
cfg->head = new_head;
cfg->size = mi->size;
mi->main_ptr = old_head;
return (0);
}
/*
* Free unneded array.
*/
static void
ta_flush_mod_fhash(void *ta_buf)
{
struct mod_item *mi;
mi = (struct mod_item *)ta_buf;
if (mi->main_ptr != NULL)
free(mi->main_ptr, M_IPFW);
}
struct table_algo flow_hash = {
.name = "flow:hash",
.type = IPFW_TABLE_FLOW,
.init = ta_init_fhash,
.destroy = ta_destroy_fhash,
.prepare_add = ta_prepare_add_fhash,
.prepare_del = ta_prepare_del_fhash,
.add = ta_add_fhash,
.del = ta_del_fhash,
.flush_entry = ta_flush_fhash_entry,
.foreach = ta_foreach_fhash,
.dump_tentry = ta_dump_fhash_tentry,
.find_tentry = ta_find_fhash_tentry,
.prepare_mod = ta_prepare_mod_fhash,
.fill_mod = ta_fill_mod_fhash,
.modify = ta_modify_fhash,
.flush_mod = ta_flush_mod_fhash,
};
void
ipfw_table_algo_init(struct ip_fw_chain *ch)
{
@ -2400,6 +3079,7 @@ ipfw_table_algo_init(struct ip_fw_chain *ch)
ipfw_add_table_algo(ch, &cidr_hash, sz, &cidr_hash.idx);
ipfw_add_table_algo(ch, &iface_idx, sz, &iface_idx.idx);
ipfw_add_table_algo(ch, &number_array, sz, &number_array.idx);
ipfw_add_table_algo(ch, &flow_hash, sz, &flow_hash.idx);
}
void
@ -2410,6 +3090,7 @@ ipfw_table_algo_destroy(struct ip_fw_chain *ch)
ipfw_del_table_algo(ch, cidr_hash.idx);
ipfw_del_table_algo(ch, iface_idx.idx);
ipfw_del_table_algo(ch, number_array.idx);
ipfw_del_table_algo(ch, flow_hash.idx);
}