HardenedBSD/sbin/ipfw/tables.c

1997 lines
43 KiB
C
Raw Normal View History

/*
2014-09-05 16:19:02 +02:00
* Copyright (c) 2014 Yandex LLC
* Copyright (c) 2014 Alexander V. Chernikov
*
* 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.
*
2014-09-05 16:19:02 +02:00
* in-kernel ipfw tables support.
*
* $FreeBSD$
*/
#include <sys/types.h>
#include <sys/param.h>
#include <sys/socket.h>
#include <sys/sysctl.h>
#include <ctype.h>
#include <err.h>
#include <errno.h>
#include <netdb.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <sysexits.h>
#include <net/if.h>
#include <netinet/in.h>
#include <netinet/ip_fw.h>
#include <arpa/inet.h>
#include <netdb.h>
#include "ipfw2.h"
static void table_modify_record(ipfw_obj_header *oh, int ac, char *av[],
int add, int quiet, int update, int atomic);
static int table_flush(ipfw_obj_header *oh);
static int table_destroy(ipfw_obj_header *oh);
static int table_do_create(ipfw_obj_header *oh, ipfw_xtable_info *i);
static int table_do_modify(ipfw_obj_header *oh, ipfw_xtable_info *i);
static int table_do_swap(ipfw_obj_header *oh, char *second);
static void table_create(ipfw_obj_header *oh, int ac, char *av[]);
static void table_modify(ipfw_obj_header *oh, int ac, char *av[]);
static void table_lookup(ipfw_obj_header *oh, int ac, char *av[]);
static void table_lock(ipfw_obj_header *oh, int lock);
static int table_swap(ipfw_obj_header *oh, char *second);
static int table_get_info(ipfw_obj_header *oh, ipfw_xtable_info *i);
static int table_show_info(ipfw_xtable_info *i, void *arg);
static void table_fill_ntlv(ipfw_obj_ntlv *ntlv, const char *name,
uint32_t set, uint16_t uidx);
static int table_flush_one(ipfw_xtable_info *i, void *arg);
static int table_show_one(ipfw_xtable_info *i, void *arg);
static int table_do_get_list(ipfw_xtable_info *i, ipfw_obj_header **poh);
static void table_show_list(ipfw_obj_header *oh, int need_header);
static void table_show_entry(ipfw_xtable_info *i, ipfw_obj_tentry *tent);
static void tentry_fill_key(ipfw_obj_header *oh, ipfw_obj_tentry *tent,
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
char *key, int add, uint8_t *ptype, uint32_t *pvmask, ipfw_xtable_info *xi);
static void tentry_fill_value(ipfw_obj_header *oh, ipfw_obj_tentry *tent,
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
char *arg, uint8_t type, uint32_t vmask);
static void table_show_value(char *buf, size_t bufsize, ipfw_table_value *v,
uint32_t vmask, int print_ip);
typedef int (table_cb_t)(ipfw_xtable_info *i, void *arg);
static int tables_foreach(table_cb_t *f, void *arg, int sort);
#ifndef s6_addr32
#define s6_addr32 __u6_addr.__u6_addr32
#endif
static struct _s_x tabletypes[] = {
{ "addr", IPFW_TABLE_ADDR },
{ "iface", IPFW_TABLE_INTERFACE },
{ "number", IPFW_TABLE_NUMBER },
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
{ "flow", IPFW_TABLE_FLOW },
{ NULL, 0 }
};
static struct _s_x tablevaltypes[] = {
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
{ "skipto", IPFW_VTYPE_SKIPTO },
{ "pipe", IPFW_VTYPE_PIPE },
{ "fib", IPFW_VTYPE_FIB },
{ "nat", IPFW_VTYPE_NAT },
{ "dscp", IPFW_VTYPE_DSCP },
{ "tag", IPFW_VTYPE_TAG },
{ "divert", IPFW_VTYPE_DIVERT },
{ "netgraph", IPFW_VTYPE_NETGRAPH },
{ "limit", IPFW_VTYPE_LIMIT },
{ "ipv4", IPFW_VTYPE_NH4 },
{ "ipv6", IPFW_VTYPE_NH6 },
{ NULL, 0 }
};
static struct _s_x tablecmds[] = {
{ "add", TOK_ADD },
{ "delete", TOK_DEL },
{ "create", TOK_CREATE },
{ "destroy", TOK_DESTROY },
{ "flush", TOK_FLUSH },
{ "modify", TOK_MODIFY },
{ "swap", TOK_SWAP },
{ "info", TOK_INFO },
{ "detail", TOK_DETAIL },
{ "list", TOK_LIST },
{ "lookup", TOK_LOOKUP },
{ "atomic", TOK_ATOMIC },
{ "lock", TOK_LOCK },
{ "unlock", TOK_UNLOCK },
{ NULL, 0 }
};
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);
}
/*
* This one handles all table-related commands
* ipfw table NAME create ...
* ipfw table NAME modify ...
* ipfw table NAME destroy
* ipfw table NAME swap NAME
* ipfw table NAME lock
* ipfw table NAME unlock
* ipfw table NAME add addr[/masklen] [value]
* ipfw table NAME add [addr[/masklen] value] [addr[/masklen] value] ..
* ipfw table NAME delete addr[/masklen] [addr[/masklen]] ..
* ipfw table NAME lookup addr
* ipfw table {NAME | all} flush
* ipfw table {NAME | all} list
* ipfw table {NAME | all} info
* ipfw table {NAME | all} detail
*/
void
ipfw_table_handler(int ac, char *av[])
{
int do_add, is_all;
int atomic, error, tcmd;
ipfw_xtable_info i;
ipfw_obj_header oh;
char *tablename;
uint32_t set;
void *arg;
memset(&oh, 0, sizeof(oh));
is_all = 0;
if (co.use_set != 0)
set = co.use_set - 1;
else
set = 0;
ac--; av++;
NEED1("table needs name");
tablename = *av;
if (table_check_name(tablename) == 0) {
table_fill_ntlv(&oh.ntlv, *av, set, 1);
oh.idx = 1;
} else {
if (strcmp(tablename, "all") == 0)
is_all = 1;
else
errx(EX_USAGE, "table name %s is invalid", tablename);
}
ac--; av++;
NEED1("table needs command");
tcmd = get_token(tablecmds, *av, "table command");
/* Check if atomic operation was requested */
atomic = 0;
if (tcmd == TOK_ATOMIC) {
ac--; av++;
NEED1("atomic needs command");
tcmd = get_token(tablecmds, *av, "table command");
switch (tcmd) {
case TOK_ADD:
break;
default:
errx(EX_USAGE, "atomic is not compatible with %s", *av);
}
atomic = 1;
}
switch (tcmd) {
case TOK_LIST:
case TOK_INFO:
case TOK_DETAIL:
case TOK_FLUSH:
break;
default:
if (is_all != 0)
errx(EX_USAGE, "table name required");
}
switch (tcmd) {
case TOK_ADD:
case TOK_DEL:
do_add = **av == 'a';
ac--; av++;
table_modify_record(&oh, ac, av, do_add, co.do_quiet,
co.do_quiet, atomic);
break;
case TOK_CREATE:
ac--; av++;
table_create(&oh, ac, av);
break;
case TOK_MODIFY:
ac--; av++;
table_modify(&oh, ac, av);
break;
case TOK_DESTROY:
if (table_destroy(&oh) != 0)
err(EX_OSERR, "failed to destroy table %s", tablename);
break;
case TOK_FLUSH:
if (is_all == 0) {
if ((error = table_flush(&oh)) != 0)
err(EX_OSERR, "failed to flush table %s info",
tablename);
} else {
error = tables_foreach(table_flush_one, &oh, 1);
if (error != 0)
err(EX_OSERR, "failed to flush tables list");
}
break;
case TOK_SWAP:
ac--; av++;
NEED1("second table name required");
table_swap(&oh, *av);
break;
case TOK_LOCK:
case TOK_UNLOCK:
table_lock(&oh, (tcmd == TOK_LOCK));
break;
case TOK_DETAIL:
case TOK_INFO:
arg = (tcmd == TOK_DETAIL) ? (void *)1 : NULL;
if (is_all == 0) {
if ((error = table_get_info(&oh, &i)) != 0)
err(EX_OSERR, "failed to request table info");
table_show_info(&i, arg);
} else {
error = tables_foreach(table_show_info, arg, 1);
if (error != 0)
err(EX_OSERR, "failed to request tables list");
}
break;
case TOK_LIST:
if (is_all == 0) {
ipfw_xtable_info i;
if ((error = table_get_info(&oh, &i)) != 0)
err(EX_OSERR, "failed to request table info");
table_show_one(&i, NULL);
} else {
error = tables_foreach(table_show_one, NULL, 1);
if (error != 0)
err(EX_OSERR, "failed to request tables list");
}
break;
case TOK_LOOKUP:
ac--; av++;
table_lookup(&oh, ac, av);
break;
}
}
static void
table_fill_ntlv(ipfw_obj_ntlv *ntlv, const char *name, uint32_t set,
uint16_t uidx)
{
ntlv->head.type = IPFW_TLV_TBL_NAME;
ntlv->head.length = sizeof(ipfw_obj_ntlv);
ntlv->idx = uidx;
ntlv->set = set;
strlcpy(ntlv->name, name, sizeof(ntlv->name));
}
static void
table_fill_objheader(ipfw_obj_header *oh, ipfw_xtable_info *i)
{
oh->idx = 1;
table_fill_ntlv(&oh->ntlv, i->tablename, i->set, 1);
}
static struct _s_x tablenewcmds[] = {
{ "type", TOK_TYPE },
{ "valtype", TOK_VALTYPE },
{ "algo", TOK_ALGO },
{ "limit", TOK_LIMIT },
{ "locked", TOK_LOCK },
{ NULL, 0 }
};
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
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)
{
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
uint32_t fset, fclear;
char *e;
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
/* Parse type options */
switch(ttype) {
case IPFW_TABLE_FLOW:
fset = fclear = 0;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
if (fill_flags(flowtypecmds, p, &e, &fset, &fclear) != 0)
errx(EX_USAGE,
"unable to parse flow option %s", e);
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
*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
*
* ipfw table NAME create [ type { addr | iface | number | flow } ]
* [ algo algoname ]
*/
static void
table_create(ipfw_obj_header *oh, int ac, char *av[])
{
ipfw_xtable_info xi;
int error, tcmd, val;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
uint32_t fset, fclear;
char *e, *p;
char tbuf[128];
memset(&xi, 0, sizeof(xi));
while (ac > 0) {
tcmd = get_token(tablenewcmds, *av, "option");
ac--; av++;
switch (tcmd) {
case TOK_LIMIT:
NEED1("limit value required");
xi.limit = strtol(*av, NULL, 10);
ac--; av++;
break;
case TOK_TYPE:
NEED1("table type required");
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
/* Type may have suboptions after ':' */
if ((p = strchr(*av, ':')) != NULL)
*p++ = '\0';
val = match_token(tabletypes, *av);
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
if (val == -1) {
concat_tokens(tbuf, sizeof(tbuf), tabletypes,
", ");
errx(EX_USAGE,
"Unknown tabletype: %s. Supported: %s",
*av, tbuf);
}
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
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");
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
fset = fclear = 0;
val = fill_flags(tablevaltypes, *av, &e, &fset, &fclear);
if (val != -1) {
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
xi.vmask = fset;
ac--; av++;
break;
}
concat_tokens(tbuf, sizeof(tbuf), tablevaltypes, ", ");
errx(EX_USAGE, "Unknown value type: %s. Supported: %s",
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
e, tbuf);
break;
case TOK_ALGO:
NEED1("table algorithm name required");
if (strlen(*av) > sizeof(xi.algoname))
errx(EX_USAGE, "algorithm name too long");
strlcpy(xi.algoname, *av, sizeof(xi.algoname));
ac--; av++;
break;
case TOK_LOCK:
xi.flags |= IPFW_TGFLAGS_LOCKED;
break;
}
}
/* Set some defaults to preserve compatibility. */
if (xi.algoname[0] == '\0' && xi.type == 0)
xi.type = IPFW_TABLE_ADDR;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
if (xi.vmask == 0)
xi.vmask = IPFW_VTYPE_LEGACY;
if ((error = table_do_create(oh, &xi)) != 0)
err(EX_OSERR, "Table creation failed");
}
/*
* Creates new table
*
* Request: [ ipfw_obj_header ipfw_xtable_info ]
*
* Returns 0 on success.
*/
static int
table_do_create(ipfw_obj_header *oh, ipfw_xtable_info *i)
{
char tbuf[sizeof(ipfw_obj_header) + sizeof(ipfw_xtable_info)];
int error;
memcpy(tbuf, oh, sizeof(*oh));
memcpy(tbuf + sizeof(*oh), i, sizeof(*i));
oh = (ipfw_obj_header *)tbuf;
error = do_set3(IP_FW_TABLE_XCREATE, &oh->opheader, sizeof(tbuf));
return (error);
}
/*
* Modifies existing table
*
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
* ipfw table NAME modify [ limit number ]
*/
static void
table_modify(ipfw_obj_header *oh, int ac, char *av[])
{
ipfw_xtable_info xi;
int tcmd;
memset(&xi, 0, sizeof(xi));
while (ac > 0) {
tcmd = get_token(tablenewcmds, *av, "option");
ac--; av++;
switch (tcmd) {
case TOK_LIMIT:
NEED1("limit value required");
xi.limit = strtol(*av, NULL, 10);
xi.mflags |= IPFW_TMFLAGS_LIMIT;
ac--; av++;
break;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
default:
errx(EX_USAGE, "cmd is not supported for modificatiob");
}
}
if (table_do_modify(oh, &xi) != 0)
err(EX_OSERR, "Table modification failed");
}
/*
* Modifies existing table.
*
* Request: [ ipfw_obj_header ipfw_xtable_info ]
*
* Returns 0 on success.
*/
static int
table_do_modify(ipfw_obj_header *oh, ipfw_xtable_info *i)
{
char tbuf[sizeof(ipfw_obj_header) + sizeof(ipfw_xtable_info)];
int error;
memcpy(tbuf, oh, sizeof(*oh));
memcpy(tbuf + sizeof(*oh), i, sizeof(*i));
oh = (ipfw_obj_header *)tbuf;
error = do_set3(IP_FW_TABLE_XMODIFY, &oh->opheader, sizeof(tbuf));
return (error);
}
/*
* Locks or unlocks given table
*/
static void
table_lock(ipfw_obj_header *oh, int lock)
{
ipfw_xtable_info xi;
memset(&xi, 0, sizeof(xi));
xi.mflags |= IPFW_TMFLAGS_LOCK;
xi.flags |= (lock != 0) ? IPFW_TGFLAGS_LOCKED : 0;
if (table_do_modify(oh, &xi) != 0)
err(EX_OSERR, "Table %s failed", lock != 0 ? "lock" : "unlock");
}
/*
* Destroys given table specified by @oh->ntlv.
* Returns 0 on success.
*/
static int
table_destroy(ipfw_obj_header *oh)
{
if (do_set3(IP_FW_TABLE_XDESTROY, &oh->opheader, sizeof(*oh)) != 0)
return (-1);
return (0);
}
/*
* Flushes given table specified by @oh->ntlv.
* Returns 0 on success.
*/
static int
table_flush(ipfw_obj_header *oh)
{
if (do_set3(IP_FW_TABLE_XFLUSH, &oh->opheader, sizeof(*oh)) != 0)
return (-1);
return (0);
}
static int
table_do_swap(ipfw_obj_header *oh, char *second)
{
char tbuf[sizeof(ipfw_obj_header) + sizeof(ipfw_obj_ntlv)];
int error;
memset(tbuf, 0, sizeof(tbuf));
memcpy(tbuf, oh, sizeof(*oh));
oh = (ipfw_obj_header *)tbuf;
table_fill_ntlv((ipfw_obj_ntlv *)(oh + 1), second, oh->ntlv.set, 1);
error = do_set3(IP_FW_TABLE_XSWAP, &oh->opheader, sizeof(tbuf));
return (error);
}
/*
* Swaps given table with @second one.
*/
static int
table_swap(ipfw_obj_header *oh, char *second)
{
int error;
if (table_check_name(second) != 0)
errx(EX_USAGE, "table name %s is invalid", second);
error = table_do_swap(oh, second);
switch (error) {
case EINVAL:
errx(EX_USAGE, "Unable to swap table: check types");
case EFBIG:
errx(EX_USAGE, "Unable to swap table: check limits");
}
return (0);
}
/*
* Retrieves table in given table specified by @oh->ntlv.
* it inside @i.
* Returns 0 on success.
*/
static int
table_get_info(ipfw_obj_header *oh, ipfw_xtable_info *i)
{
char tbuf[sizeof(ipfw_obj_header) + sizeof(ipfw_xtable_info)];
size_t sz;
sz = sizeof(tbuf);
memset(tbuf, 0, sizeof(tbuf));
memcpy(tbuf, oh, sizeof(*oh));
oh = (ipfw_obj_header *)tbuf;
if (do_get3(IP_FW_TABLE_XINFO, &oh->opheader, &sz) != 0)
return (errno);
if (sz < sizeof(tbuf))
return (EINVAL);
*i = *(ipfw_xtable_info *)(oh + 1);
return (0);
}
static struct _s_x tablealgoclass[] = {
{ "hash", IPFW_TACLASS_HASH },
{ "array", IPFW_TACLASS_ARRAY },
{ "radix", IPFW_TACLASS_RADIX },
{ NULL, 0 }
};
struct ta_cldata {
uint8_t taclass;
uint8_t spare4;
uint16_t itemsize;
uint16_t itemsize6;
uint32_t size;
uint32_t count;
};
/*
* Print global/per-AF table @i algorithm info.
*/
static void
table_show_tainfo(ipfw_xtable_info *i, struct ta_cldata *d,
const char *af, const char *taclass)
{
switch (d->taclass) {
case IPFW_TACLASS_HASH:
case IPFW_TACLASS_ARRAY:
printf(" %salgorithm %s info\n", af, taclass);
if (d->itemsize == d->itemsize6)
printf(" size: %u items: %u itemsize: %u\n",
d->size, d->count, d->itemsize);
else
printf(" size: %u items: %u "
"itemsize4: %u itemsize6: %u\n",
d->size, d->count,
d->itemsize, d->itemsize6);
break;
case IPFW_TACLASS_RADIX:
printf(" %salgorithm %s info\n", af, taclass);
if (d->itemsize == d->itemsize6)
printf(" items: %u itemsize: %u\n",
d->count, d->itemsize);
else
printf(" items: %u "
"itemsize4: %u itemsize6: %u\n",
d->count, d->itemsize, d->itemsize6);
break;
default:
printf(" algo class: %s\n", taclass);
}
}
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
static void
table_print_valheader(char *buf, size_t bufsize, uint32_t vmask)
{
if (vmask == IPFW_VTYPE_LEGACY) {
snprintf(buf, bufsize, "legacy");
return;
}
memset(buf, 0, bufsize);
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
print_flags_buffer(buf, bufsize, tablevaltypes, vmask);
}
/*
* Prints table info struct @i in human-readable form.
*/
static int
table_show_info(ipfw_xtable_info *i, void *arg)
{
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
const char *vtype;
ipfw_ta_tinfo *tainfo;
int afdata, afitem;
struct ta_cldata d;
char ttype[64], tvtype[64];
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
table_print_type(ttype, sizeof(ttype), i->type, i->tflags);
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
table_print_valheader(tvtype, sizeof(tvtype), i->vmask);
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
printf("--- table(%s), set(%u) ---\n", i->tablename, i->set);
if ((i->flags & IPFW_TGFLAGS_LOCKED) != 0)
printf(" kindex: %d, type: %s, locked\n", i->kidx, ttype);
else
printf(" kindex: %d, type: %s\n", i->kidx, ttype);
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
printf(" references: %u, valtype: %s\n", i->refcnt, tvtype);
printf(" algorithm: %s\n", i->algoname);
printf(" items: %u, size: %u\n", i->count, i->size);
if (i->limit > 0)
printf(" limit: %u\n", i->limit);
/* Print algo-specific info if requested & set */
if (arg == NULL)
return (0);
if ((i->ta_info.flags & IPFW_TATFLAGS_DATA) == 0)
return (0);
tainfo = &i->ta_info;
afdata = 0;
afitem = 0;
if (tainfo->flags & IPFW_TATFLAGS_AFDATA)
afdata = 1;
if (tainfo->flags & IPFW_TATFLAGS_AFITEM)
afitem = 1;
memset(&d, 0, sizeof(d));
d.taclass = tainfo->taclass4;
d.size = tainfo->size4;
d.count = tainfo->count4;
d.itemsize = tainfo->itemsize4;
if (afdata == 0 && afitem != 0)
d.itemsize6 = tainfo->itemsize6;
else
d.itemsize6 = d.itemsize;
if ((vtype = match_value(tablealgoclass, d.taclass)) == NULL)
vtype = "unknown";
if (afdata == 0) {
table_show_tainfo(i, &d, "", vtype);
} else {
table_show_tainfo(i, &d, "IPv4 ", vtype);
memset(&d, 0, sizeof(d));
d.taclass = tainfo->taclass6;
if ((vtype = match_value(tablealgoclass, d.taclass)) == NULL)
vtype = "unknown";
d.size = tainfo->size6;
d.count = tainfo->count6;
d.itemsize = tainfo->itemsize6;
d.itemsize6 = d.itemsize;
table_show_tainfo(i, &d, "IPv6 ", vtype);
}
return (0);
}
/*
* Function wrappers which can be used either
* as is or as foreach function parameter.
*/
static int
table_show_one(ipfw_xtable_info *i, void *arg)
{
ipfw_obj_header *oh;
int error;
if ((error = table_do_get_list(i, &oh)) != 0) {
err(EX_OSERR, "Error requesting table %s list", i->tablename);
return (error);
}
table_show_list(oh, 1);
free(oh);
return (0);
}
static int
table_flush_one(ipfw_xtable_info *i, void *arg)
{
ipfw_obj_header *oh;
oh = (ipfw_obj_header *)arg;
table_fill_ntlv(&oh->ntlv, i->tablename, i->set, 1);
return (table_flush(oh));
}
static int
table_do_modify_record(int cmd, ipfw_obj_header *oh,
ipfw_obj_tentry *tent, int count, int atomic)
{
ipfw_obj_ctlv *ctlv;
ipfw_obj_tentry *tent_base;
caddr_t pbuf;
char xbuf[sizeof(*oh) + sizeof(ipfw_obj_ctlv) + sizeof(*tent)];
int error, i;
size_t sz;
sz = sizeof(*ctlv) + sizeof(*tent) * count;
if (count == 1) {
memset(xbuf, 0, sizeof(xbuf));
pbuf = xbuf;
} else {
if ((pbuf = calloc(1, sizeof(*oh) + sz)) == NULL)
return (ENOMEM);
}
memcpy(pbuf, oh, sizeof(*oh));
oh = (ipfw_obj_header *)pbuf;
oh->opheader.version = 1;
ctlv = (ipfw_obj_ctlv *)(oh + 1);
ctlv->count = count;
ctlv->head.length = sz;
if (atomic != 0)
ctlv->flags |= IPFW_CTF_ATOMIC;
tent_base = tent;
memcpy(ctlv + 1, tent, sizeof(*tent) * count);
tent = (ipfw_obj_tentry *)(ctlv + 1);
for (i = 0; i < count; i++, tent++) {
tent->head.length = sizeof(ipfw_obj_tentry);
tent->idx = oh->idx;
}
sz += sizeof(*oh);
error = do_get3(cmd, &oh->opheader, &sz);
tent = (ipfw_obj_tentry *)(ctlv + 1);
/* Copy result back to provided buffer */
memcpy(tent_base, ctlv + 1, sizeof(*tent) * count);
if (pbuf != xbuf)
free(pbuf);
return (error);
}
static void
table_modify_record(ipfw_obj_header *oh, int ac, char *av[], int add,
int quiet, int update, int atomic)
{
ipfw_obj_tentry *ptent, tent, *tent_buf;
ipfw_xtable_info xi;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
uint8_t type;
uint32_t vmask;
int cmd, count, error, i, ignored;
char *texterr, *etxt, *px;
if (ac == 0)
errx(EX_USAGE, "address required");
if (add != 0) {
cmd = IP_FW_TABLE_XADD;
texterr = "Adding record failed";
} else {
cmd = IP_FW_TABLE_XDEL;
texterr = "Deleting record failed";
}
/*
* Calculate number of entries:
* Assume [key val] x N for add
* and
* key x N for delete
*/
count = (add != 0) ? ac / 2 + 1 : ac;
if (count <= 1) {
/* Adding single entry with/without value */
memset(&tent, 0, sizeof(tent));
tent_buf = &tent;
} else {
if ((tent_buf = calloc(count, sizeof(tent))) == NULL)
errx(EX_OSERR,
"Unable to allocate memory for all entries");
}
ptent = tent_buf;
memset(&xi, 0, sizeof(xi));
count = 0;
while (ac > 0) {
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
tentry_fill_key(oh, ptent, *av, add, &type, &vmask, &xi);
/*
* Compatibility layer: auto-create table if not exists.
*/
if (xi.tablename[0] == '\0') {
xi.type = type;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
xi.vmask = vmask;
strlcpy(xi.tablename, oh->ntlv.name,
sizeof(xi.tablename));
fprintf(stderr, "DEPRECATED: inserting data into "
"non-existent table %s. (auto-created)\n",
xi.tablename);
table_do_create(oh, &xi);
}
oh->ntlv.type = type;
ac--; av++;
if (add != 0 && ac > 0) {
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
tentry_fill_value(oh, ptent, *av, type, vmask);
ac--; av++;
}
if (update != 0)
ptent->head.flags |= IPFW_TF_UPDATE;
count++;
ptent++;
}
error = table_do_modify_record(cmd, oh, tent_buf, count, atomic);
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
quiet = 0;
/*
* Compatibility stuff: do not yell on duplicate keys or
* failed deletions.
*/
if (error == 0 || (error == EEXIST && add != 0) ||
(error == ENOENT && add == 0)) {
if (quiet != 0) {
if (tent_buf != &tent)
free(tent_buf);
return;
}
}
/* Report results back */
ptent = tent_buf;
for (i = 0; i < count; ptent++, i++) {
ignored = 0;
switch (ptent->result) {
case IPFW_TR_ADDED:
px = "added";
break;
case IPFW_TR_DELETED:
px = "deleted";
break;
case IPFW_TR_UPDATED:
px = "updated";
break;
case IPFW_TR_LIMIT:
px = "limit";
ignored = 1;
break;
case IPFW_TR_ERROR:
px = "error";
ignored = 1;
break;
case IPFW_TR_NOTFOUND:
px = "notfound";
ignored = 1;
break;
case IPFW_TR_EXISTS:
px = "exists";
ignored = 1;
break;
case IPFW_TR_IGNORED:
px = "ignored";
ignored = 1;
break;
default:
px = "unknown";
ignored = 1;
}
if (error != 0 && atomic != 0 && ignored == 0)
printf("%s(reverted): ", px);
else
printf("%s: ", px);
table_show_entry(&xi, ptent);
}
if (tent_buf != &tent)
free(tent_buf);
if (error == 0)
return;
/* Get real OS error */
error = errno;
/* Try to provide more human-readable error */
switch (error) {
case EEXIST:
etxt = "record already exists";
break;
case EFBIG:
etxt = "limit hit";
break;
case ESRCH:
etxt = "table not found";
break;
case ENOENT:
etxt = "record not found";
break;
case EACCES:
etxt = "table is locked";
break;
default:
etxt = strerror(error);
}
errx(EX_OSERR, "%s: %s", texterr, etxt);
}
static int
table_do_lookup(ipfw_obj_header *oh, char *key, ipfw_xtable_info *xi,
ipfw_obj_tentry *xtent)
{
char xbuf[sizeof(ipfw_obj_header) + sizeof(ipfw_obj_tentry)];
ipfw_obj_tentry *tent;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
uint8_t type;
uint32_t vmask;
size_t sz;
memcpy(xbuf, oh, sizeof(*oh));
oh = (ipfw_obj_header *)xbuf;
tent = (ipfw_obj_tentry *)(oh + 1);
memset(tent, 0, sizeof(*tent));
tent->head.length = sizeof(*tent);
tent->idx = 1;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
tentry_fill_key(oh, tent, key, 0, &type, &vmask, xi);
oh->ntlv.type = type;
sz = sizeof(xbuf);
if (do_get3(IP_FW_TABLE_XFIND, &oh->opheader, &sz) != 0)
return (errno);
if (sz < sizeof(xbuf))
return (EINVAL);
*xtent = *tent;
return (0);
}
static void
table_lookup(ipfw_obj_header *oh, int ac, char *av[])
{
ipfw_obj_tentry xtent;
ipfw_xtable_info xi;
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
char key[64];
int error;
if (ac == 0)
errx(EX_USAGE, "address required");
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
strlcpy(key, *av, sizeof(key));
memset(&xi, 0, sizeof(xi));
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
error = table_do_lookup(oh, key, &xi, &xtent);
switch (error) {
case 0:
break;
case ESRCH:
errx(EX_UNAVAILABLE, "Table %s not found", oh->ntlv.name);
case ENOENT:
errx(EX_UNAVAILABLE, "Entry %s not found", *av);
case ENOTSUP:
errx(EX_UNAVAILABLE, "Table %s algo does not support "
"\"lookup\" method", oh->ntlv.name);
default:
err(EX_OSERR, "getsockopt(IP_FW_TABLE_XFIND)");
}
table_show_entry(&xi, &xtent);
}
static void
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
tentry_fill_key_type(char *arg, ipfw_obj_tentry *tentry, uint8_t type,
uint8_t tflags)
{
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
char *p, *pp;
int mask, af;
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
struct in6_addr *paddr, tmp;
struct tflow_entry *tfe;
uint32_t key, *pkey;
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
uint16_t port;
struct protoent *pent;
struct servent *sent;
int masklen;
masklen = 0;
af = 0;
paddr = (struct in6_addr *)&tentry->k;
switch (type) {
case IPFW_TABLE_ADDR:
/* Remove / if exists */
if ((p = strchr(arg, '/')) != NULL) {
*p = '\0';
mask = atoi(p + 1);
}
if (inet_pton(AF_INET, arg, paddr) == 1) {
if (p != NULL && mask > 32)
errx(EX_DATAERR, "bad IPv4 mask width: %s",
p + 1);
masklen = p ? mask : 32;
af = AF_INET;
} else if (inet_pton(AF_INET6, arg, paddr) == 1) {
if (IN6_IS_ADDR_V4COMPAT(paddr))
errx(EX_DATAERR,
"Use IPv4 instead of v4-compatible");
if (p != NULL && mask > 128)
errx(EX_DATAERR, "bad IPv6 mask width: %s",
p + 1);
masklen = p ? mask : 128;
af = AF_INET6;
} else {
/* Assume FQDN */
if (lookup_host(arg, (struct in_addr *)paddr) != 0)
errx(EX_NOHOST, "hostname ``%s'' unknown", arg);
masklen = 32;
type = IPFW_TABLE_ADDR;
af = AF_INET;
}
break;
case IPFW_TABLE_INTERFACE:
/* Assume interface name. Copy significant data only */
mask = MIN(strlen(arg), IF_NAMESIZE - 1);
memcpy(paddr, arg, mask);
/* Set mask to exact match */
masklen = 8 * IF_NAMESIZE;
break;
case IPFW_TABLE_NUMBER:
/* Port or any other key */
key = strtol(arg, &p, 10);
if (*p != '\0')
errx(EX_DATAERR, "Invalid number: %s", arg);
pkey = (uint32_t *)paddr;
*pkey = key;
masklen = 32;
break;
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
case IPFW_TABLE_FLOW:
/* Assume [src-ip][,proto][,src-port][,dst-ip][,dst-port] */
tfe = &tentry->k.flow;
af = 0;
/* Handle <ipv4|ipv6> */
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
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 (arg == NULL)
errx(EX_DATAERR, "invalid key: proto missing");
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
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 (arg == NULL)
errx(EX_DATAERR, "invalid key: src port missing");
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
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 (arg == NULL)
errx(EX_DATAERR, "invalid key: dst ip missing");
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
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 (arg == NULL)
errx(EX_DATAERR, "invalid key: dst port missing");
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
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);
}
tentry->subtype = af;
tentry->masklen = masklen;
}
/*
* Tries to guess table key type.
* This procedure is used in legacy table auto-create
* code AND in `ipfw -n` ruleset checking.
*
* Imported from old table_fill_xentry() parse code.
*/
static int
guess_key_type(char *key, uint8_t *ptype)
{
char *p;
struct in6_addr addr;
uint32_t kv;
if (ishexnumber(*key) != 0 || *key == ':') {
/* Remove / if exists */
if ((p = strchr(key, '/')) != NULL)
*p = '\0';
if ((inet_pton(AF_INET, key, &addr) == 1) ||
(inet_pton(AF_INET6, key, &addr) == 1)) {
*ptype = IPFW_TABLE_CIDR;
if (p != NULL)
*p = '/';
return (0);
} else {
/* Port or any other key */
/* Skip non-base 10 entries like 'fa1' */
kv = strtol(key, &p, 10);
if (*p == '\0') {
*ptype = IPFW_TABLE_NUMBER;
return (0);
} else if ((p != key) && (*p == '.')) {
/*
* Warn on IPv4 address strings
* which are "valid" for inet_aton() but not
* in inet_pton().
*
* Typical examples: '10.5' or '10.0.0.05'
*/
return (1);
}
}
}
if (strchr(key, '.') == NULL) {
*ptype = IPFW_TABLE_INTERFACE;
return (0);
}
if (lookup_host(key, (struct in_addr *)&addr) != 0)
return (1);
*ptype = IPFW_TABLE_CIDR;
return (0);
}
static void
tentry_fill_key(ipfw_obj_header *oh, ipfw_obj_tentry *tent, char *key,
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
int add, uint8_t *ptype, uint32_t *pvmask, ipfw_xtable_info *xi)
{
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
uint8_t type, tflags;
uint32_t vmask;
int error;
type = 0;
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
tflags = 0;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
vmask = 0;
if (xi->tablename[0] == '\0')
error = table_get_info(oh, xi);
else
error = 0;
if (error == 0) {
if (co.test_only == 0) {
/* Table found */
type = xi->type;
tflags = xi->tflags;
vmask = xi->vmask;
} else {
/*
* We're running `ipfw -n`
* Compatibility layer: try to guess key type
* before failing.
*/
if (guess_key_type(key, &type) != 0) {
/* Inknown key */
errx(EX_USAGE, "Cannot guess "
"key '%s' type", key);
}
vmask = IPFW_VTYPE_LEGACY;
}
} else {
if (error != ESRCH)
errx(EX_OSERR, "Error requesting table %s info",
oh->ntlv.name);
if (add == 0)
errx(EX_DATAERR, "Table %s does not exist",
oh->ntlv.name);
/*
* Table does not exist
* Compatibility layer: try to guess key type before failing.
*/
if (guess_key_type(key, &type) != 0) {
/* Inknown key */
errx(EX_USAGE, "Table %s does not exist, cannot guess "
"key '%s' type", oh->ntlv.name, key);
}
vmask = IPFW_VTYPE_LEGACY;
}
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
tentry_fill_key_type(key, tent, type, tflags);
*ptype = type;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
*pvmask = vmask;
}
static void
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
set_legacy_value(uint32_t val, ipfw_table_value *v)
{
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
v->tag = val;
v->pipe = val;
v->divert = val;
v->skipto = val;
v->netgraph = val;
v->fib = val;
v->nat = val;
v->nh4 = val;
v->dscp = (uint8_t)val;
v->limit = val;
}
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
static void
tentry_fill_value(ipfw_obj_header *oh, ipfw_obj_tentry *tent, char *arg,
uint8_t type, uint32_t vmask)
{
struct addrinfo hints, *res;
uint32_t a4, flag, val;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
ipfw_table_value *v;
uint32_t i;
2014-10-04 19:21:30 +02:00
int dval;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
char *comma, *e, *etype, *n, *p;
v = &tent->v.value;
/* Compat layer: keep old behavior for legacy value types */
if (vmask == IPFW_VTYPE_LEGACY) {
/* Try to interpret as number first */
val = strtoul(arg, &p, 0);
if (*p == '\0') {
set_legacy_value(val, v);
return;
}
if (inet_pton(AF_INET, arg, &val) == 1) {
set_legacy_value(ntohl(val), v);
return;
}
/* Try hostname */
if (lookup_host(arg, (struct in_addr *)&val) == 0) {
set_legacy_value(val, v);
return;
}
errx(EX_OSERR, "Unable to parse value %s", arg);
}
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
/*
* Shorthands: handle single value if vmask consists
* of numbers only. e.g.:
* vmask = "fib,skipto" -> treat input "1" as "1,1"
*/
n = arg;
etype = NULL;
for (i = 1; i < (1 << 31); i *= 2) {
if ((flag = (vmask & i)) == 0)
continue;
vmask &= ~flag;
if ((comma = strchr(n, ',')) != NULL)
*comma = '\0';
switch (flag) {
case IPFW_VTYPE_TAG:
v->tag = strtol(n, &e, 10);
if (*e != '\0')
etype = "tag";
break;
case IPFW_VTYPE_PIPE:
v->pipe = strtol(n, &e, 10);
if (*e != '\0')
etype = "pipe";
break;
case IPFW_VTYPE_DIVERT:
v->divert = strtol(n, &e, 10);
if (*e != '\0')
etype = "divert";
break;
case IPFW_VTYPE_SKIPTO:
v->skipto = strtol(n, &e, 10);
if (*e != '\0')
etype = "skipto";
break;
case IPFW_VTYPE_NETGRAPH:
v->netgraph = strtol(n, &e, 10);
if (*e != '\0')
etype = "netgraph";
break;
case IPFW_VTYPE_FIB:
v->fib = strtol(n, &e, 10);
if (*e != '\0')
etype = "fib";
break;
case IPFW_VTYPE_NAT:
v->nat = strtol(n, &e, 10);
if (*e != '\0')
etype = "nat";
break;
case IPFW_VTYPE_LIMIT:
v->limit = strtol(n, &e, 10);
if (*e != '\0')
etype = "limit";
break;
case IPFW_VTYPE_NH4:
if (strchr(n, '.') != NULL &&
inet_pton(AF_INET, n, &a4) == 1) {
v->nh4 = ntohl(a4);
break;
}
if (lookup_host(n, (struct in_addr *)&v->nh4) == 0)
break;
etype = "ipv4";
break;
case IPFW_VTYPE_DSCP:
if (isalpha(*n)) {
2014-10-04 19:21:30 +02:00
if ((dval = match_token(f_ipdscp, n)) != -1) {
v->dscp = dval;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
break;
2014-10-04 19:21:30 +02:00
} else
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
etype = "DSCP code";
} else {
v->dscp = strtol(n, &e, 10);
if (v->dscp > 63 || *e != '\0')
etype = "DSCP value";
}
break;
case IPFW_VTYPE_NH6:
if (strchr(n, ':') != NULL) {
memset(&hints, 0, sizeof(hints));
hints.ai_family = AF_INET6;
hints.ai_flags = AI_NUMERICHOST;
if (getaddrinfo(n, NULL, &hints, &res) == 0) {
v->nh6 = ((struct sockaddr_in6 *)
res->ai_addr)->sin6_addr;
v->zoneid = ((struct sockaddr_in6 *)
res->ai_addr)->sin6_scope_id;
freeaddrinfo(res);
break;
}
}
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
etype = "ipv6";
break;
}
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
if (etype != NULL)
errx(EX_USAGE, "Unable to parse %s as %s", n, etype);
if (comma != NULL)
*comma++ = ',';
if ((n = comma) != NULL)
continue;
/* End of input. */
if (vmask != 0)
errx(EX_USAGE, "Not enough fields inside value");
}
}
/*
* Compare table names.
* Honor number comparison.
*/
static int
tablename_cmp(const void *a, const void *b)
{
ipfw_xtable_info *ia, *ib;
ia = (ipfw_xtable_info *)a;
ib = (ipfw_xtable_info *)b;
return (stringnum_cmp(ia->tablename, ib->tablename));
}
/*
* Retrieves table list from kernel,
* optionally sorts it and calls requested function for each table.
* Returns 0 on success.
*/
static int
tables_foreach(table_cb_t *f, void *arg, int sort)
{
ipfw_obj_lheader *olh;
ipfw_xtable_info *info;
size_t sz;
int i, error;
/* Start with reasonable default */
sz = sizeof(*olh) + 16 * sizeof(ipfw_xtable_info);
for (;;) {
if ((olh = calloc(1, sz)) == NULL)
return (ENOMEM);
olh->size = sz;
if (do_get3(IP_FW_TABLES_XLIST, &olh->opheader, &sz) != 0) {
sz = olh->size;
free(olh);
if (errno != ENOMEM)
return (errno);
continue;
}
if (sort != 0)
qsort(olh + 1, olh->count, olh->objsize, tablename_cmp);
info = (ipfw_xtable_info *)(olh + 1);
for (i = 0; i < olh->count; i++) {
error = f(info, arg); /* Ignore errors for now */
info = (ipfw_xtable_info *)((caddr_t)info + olh->objsize);
}
free(olh);
break;
}
return (0);
}
/*
* Retrieves all entries for given table @i in
* eXtended format. Allocate buffer large enough
* to store result. Called needs to free it later.
*
* Returns 0 on success.
*/
static int
table_do_get_list(ipfw_xtable_info *i, ipfw_obj_header **poh)
{
ipfw_obj_header *oh;
size_t sz;
int c;
sz = 0;
oh = NULL;
for (c = 0; c < 8; c++) {
if (sz < i->size)
sz = i->size + 44;
if (oh != NULL)
free(oh);
if ((oh = calloc(1, sz)) == NULL)
continue;
table_fill_objheader(oh, i);
oh->opheader.version = 1; /* Current version */
if (do_get3(IP_FW_TABLE_XLIST, &oh->opheader, &sz) == 0) {
*poh = oh;
return (0);
}
if (errno != ENOMEM)
break;
}
free(oh);
return (errno);
}
/*
* Shows all entries from @oh in human-readable format
*/
static void
table_show_list(ipfw_obj_header *oh, int need_header)
{
ipfw_obj_tentry *tent;
uint32_t count;
ipfw_xtable_info *i;
i = (ipfw_xtable_info *)(oh + 1);
tent = (ipfw_obj_tentry *)(i + 1);
if (need_header)
printf("--- table(%s), set(%u) ---\n", i->tablename, i->set);
count = i->count;
while (count > 0) {
table_show_entry(i, tent);
tent = (ipfw_obj_tentry *)((caddr_t)tent + tent->head.length);
count--;
}
}
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
static void
table_show_value(char *buf, size_t bufsize, ipfw_table_value *v,
uint32_t vmask, int print_ip)
{
char abuf[INET6_ADDRSTRLEN + IF_NAMESIZE + 2];
struct sockaddr_in6 sa6;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
uint32_t flag, i, l;
size_t sz;
struct in_addr a4;
sz = bufsize;
/*
* Some shorthands for printing values:
* legacy assumes all values are equal, so keep the first one.
*/
if (vmask == IPFW_VTYPE_LEGACY) {
if (print_ip != 0) {
flag = htonl(v->tag);
inet_ntop(AF_INET, &flag, buf, sz);
} else
snprintf(buf, sz, "%u", v->tag);
return;
}
for (i = 1; i < (1 << 31); i *= 2) {
if ((flag = (vmask & i)) == 0)
continue;
l = 0;
switch (flag) {
case IPFW_VTYPE_TAG:
l = snprintf(buf, sz, "%u,", v->tag);
break;
case IPFW_VTYPE_PIPE:
l = snprintf(buf, sz, "%u,", v->pipe);
break;
case IPFW_VTYPE_DIVERT:
l = snprintf(buf, sz, "%d,", v->divert);
break;
case IPFW_VTYPE_SKIPTO:
l = snprintf(buf, sz, "%d,", v->skipto);
break;
case IPFW_VTYPE_NETGRAPH:
l = snprintf(buf, sz, "%u,", v->netgraph);
break;
case IPFW_VTYPE_FIB:
l = snprintf(buf, sz, "%u,", v->fib);
break;
case IPFW_VTYPE_NAT:
l = snprintf(buf, sz, "%u,", v->nat);
break;
case IPFW_VTYPE_LIMIT:
l = snprintf(buf, sz, "%u,", v->limit);
break;
case IPFW_VTYPE_NH4:
a4.s_addr = htonl(v->nh4);
inet_ntop(AF_INET, &a4, abuf, sizeof(abuf));
l = snprintf(buf, sz, "%s,", abuf);
break;
case IPFW_VTYPE_DSCP:
l = snprintf(buf, sz, "%d,", v->dscp);
break;
case IPFW_VTYPE_NH6:
sa6.sin6_family = AF_INET6;
sa6.sin6_len = sizeof(sa6);
sa6.sin6_addr = v->nh6;
sa6.sin6_port = 0;
sa6.sin6_scope_id = v->zoneid;
if (getnameinfo((const struct sockaddr *)&sa6,
sa6.sin6_len, abuf, sizeof(abuf), NULL, 0,
NI_NUMERICHOST) == 0)
l = snprintf(buf, sz, "%s,", abuf);
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
break;
}
buf += l;
sz -= l;
}
if (sz != bufsize)
*(buf - 1) = '\0';
}
static void
table_show_entry(ipfw_xtable_info *i, ipfw_obj_tentry *tent)
{
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
char *comma, tbuf[128], pval[128];
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
void *paddr;
struct tflow_entry *tfe;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
table_show_value(pval, sizeof(pval), &tent->v.value, i->vmask,
co.do_value_as_ip);
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
switch (i->type) {
case IPFW_TABLE_ADDR:
/* IPv4 or IPv6 prefixes */
inet_ntop(tent->subtype, &tent->k, tbuf, sizeof(tbuf));
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
printf("%s/%u %s\n", tbuf, tent->masklen, pval);
break;
case IPFW_TABLE_INTERFACE:
/* Interface names */
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
printf("%s %s\n", tent->k.iface, pval);
break;
case IPFW_TABLE_NUMBER:
/* numbers */
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
printf("%u %s\n", tent->k.key, pval);
break;
* Add new "flow" table type to support N=1..5-tuple lookups * 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
2014-07-31 22:08:19 +02:00
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);
}
}
static int
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
table_do_get_stdlist(uint16_t opcode, ipfw_obj_lheader **polh)
{
ipfw_obj_lheader req, *olh;
size_t sz;
memset(&req, 0, sizeof(req));
sz = sizeof(req);
if (do_get3(opcode, &req.opheader, &sz) != 0)
if (errno != ENOMEM)
return (errno);
sz = req.size;
if ((olh = calloc(1, sz)) == NULL)
return (ENOMEM);
olh->size = sz;
if (do_get3(opcode, &olh->opheader, &sz) != 0) {
free(olh);
return (errno);
}
*polh = olh;
return (0);
}
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
static int
table_do_get_algolist(ipfw_obj_lheader **polh)
{
return (table_do_get_stdlist(IP_FW_TABLES_ALIST, polh));
}
static int
table_do_get_vlist(ipfw_obj_lheader **polh)
{
return (table_do_get_stdlist(IP_FW_TABLE_VLIST, polh));
}
void
ipfw_list_ta(int ac, char *av[])
{
ipfw_obj_lheader *olh;
ipfw_ta_info *info;
int error, i;
const char *atype;
error = table_do_get_algolist(&olh);
if (error != 0)
err(EX_OSERR, "Unable to request algorithm list");
info = (ipfw_ta_info *)(olh + 1);
for (i = 0; i < olh->count; i++) {
if ((atype = match_value(tabletypes, info->type)) == NULL)
atype = "unknown";
2014-07-30 17:01:32 +02:00
printf("--- %s ---\n", info->algoname);
printf(" type: %s\n refcount: %u\n", atype, info->refcnt);
info = (ipfw_ta_info *)((caddr_t)info + olh->objsize);
}
free(olh);
}
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
/* Copy of current kernel table_value structure */
struct _table_value {
uint32_t tag; /* O_TAG/O_TAGGED */
uint32_t pipe; /* O_PIPE/O_QUEUE */
uint16_t divert; /* O_DIVERT/O_TEE */
uint16_t skipto; /* skipto, CALLRET */
uint32_t netgraph; /* O_NETGRAPH/O_NGTEE */
uint32_t fib; /* O_SETFIB */
uint32_t nat; /* O_NAT */
uint32_t nh4;
uint8_t dscp;
uint8_t spare0;
uint16_t spare1;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
/* -- 32 bytes -- */
struct in6_addr nh6;
uint32_t limit; /* O_LIMIT */
uint32_t zoneid;
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
uint64_t refcnt; /* Number of references */
};
int
compare_values(const void *_a, const void *_b)
{
struct _table_value *a, *b;
a = (struct _table_value *)_a;
b = (struct _table_value *)_b;
if (a->spare1 < b->spare1)
return (-1);
else if (a->spare1 > b->spare1)
return (1);
return (0);
}
void
ipfw_list_values(int ac, char *av[])
{
ipfw_obj_lheader *olh;
struct _table_value *v;
int error, i;
uint32_t vmask;
char buf[128];
error = table_do_get_vlist(&olh);
if (error != 0)
err(EX_OSERR, "Unable to request value list");
vmask = 0x7FFFFFFF; /* Similar to IPFW_VTYPE_LEGACY */
table_print_valheader(buf, sizeof(buf), vmask);
printf("HEADER: %s\n", buf);
v = (struct _table_value *)(olh + 1);
qsort(v, olh->count, olh->objsize, compare_values);
for (i = 0; i < olh->count; i++) {
table_show_value(buf, sizeof(buf), (ipfw_table_value *)v,
vmask, 0);
printf("[%u] refs=%lu %s\n", v->spare1, (u_long)v->refcnt, buf);
Add support for multi-field values inside ipfw tables. This is the last major change in given branch. Kernel changes: * Use 64-bytes structures to hold multi-value variables. * Use shared array to hold values from all tables (assume each table algo is capable of holding 32-byte variables). * Add some placeholders to support per-table value arrays in future. * Use simple eventhandler-style API to ease the process of adding new table items. Currently table addition may required multiple UH drops/ acquires which is quite tricky due to atomic table modificatio/swap support, shared array resize, etc. Deal with it by calling special notifier capable of rolling back state before actually performing swap/resize operations. Original operation then restarts itself after acquiring UH lock. * Bump all objhash users default values to at least 64 * Fix custom hashing inside objhash. Userland changes: * Add support for dumping shared value array via "vlist" internal cmd. * Some small print/fill_flags dixes to support u32 values. * valtype is now bitmask of <skipto|pipe|fib|nat|dscp|tag|divert|netgraph|limit|ipv4|ipv6>. New values can hold distinct values for each of this types. * Provide special "legacy" type which assumes all values are the same. * More helpers/docs following.. Some examples: 3:41 [1] zfscurr0# ipfw table mimimi create valtype skipto,limit,ipv4,ipv6 3:41 [1] zfscurr0# ipfw table mimimi info +++ table(mimimi), set(0) +++ kindex: 2, type: addr references: 0, valtype: skipto,limit,ipv4,ipv6 algorithm: addr:radix items: 0, size: 296 3:42 [1] zfscurr0# ipfw table mimimi add 10.0.0.5 3000,10,10.0.0.1,2a02:978:2::1 added: 10.0.0.5/32 3000,10,10.0.0.1,2a02:978:2::1 3:42 [1] zfscurr0# ipfw table mimimi list +++ table(mimimi), set(0) +++ 10.0.0.5/32 3000,0,10.0.0.1,2a02:978:2::1
2014-09-01 01:51:09 +02:00
v = (struct _table_value *)((caddr_t)v + olh->objsize);
}
free(olh);
}
int
table_check_name(const char *tablename)
{
if (ipfw_check_object_name(tablename) != 0)
return (EINVAL);
/* Restrict some 'special' names */
if (strcmp(tablename, "all") == 0)
return (EINVAL);
return (0);
}