From 810b98f82c5999118c2f74d5a22c7f0836771c7b Mon Sep 17 00:00:00 2001 From: S.Çağlar Onur Date: Tue, 7 Dec 2010 11:05:31 -0500 Subject: [PATCH] linux-2.6-250-ipsets.patch --- include/linux/netfilter_ipv4/ip_set.h | 561 +++++++ include/linux/netfilter_ipv4/ip_set_bitmaps.h | 120 ++ include/linux/netfilter_ipv4/ip_set_compat.h | 92 ++ include/linux/netfilter_ipv4/ip_set_getport.h | 48 + include/linux/netfilter_ipv4/ip_set_hashes.h | 314 ++++ include/linux/netfilter_ipv4/ip_set_iphash.h | 30 + include/linux/netfilter_ipv4/ip_set_ipmap.h | 57 + include/linux/netfilter_ipv4/ip_set_ipporthash.h | 33 + include/linux/netfilter_ipv4/ip_set_ipportiphash.h | 39 + .../linux/netfilter_ipv4/ip_set_ipportnethash.h | 42 + include/linux/netfilter_ipv4/ip_set_iptree.h | 39 + include/linux/netfilter_ipv4/ip_set_iptreemap.h | 40 + include/linux/netfilter_ipv4/ip_set_jhash.h | 157 ++ include/linux/netfilter_ipv4/ip_set_macipmap.h | 39 + include/linux/netfilter_ipv4/ip_set_malloc.h | 153 ++ include/linux/netfilter_ipv4/ip_set_nethash.h | 31 + include/linux/netfilter_ipv4/ip_set_portmap.h | 25 + include/linux/netfilter_ipv4/ip_set_setlist.h | 26 + include/linux/netfilter_ipv4/ipt_set.h | 21 + net/ipv4/netfilter/Kconfig | 117 ++ net/ipv4/netfilter/Makefile | 13 + net/ipv4/netfilter/ip_set.c | 1531 ++++++++++++++++++++ net/ipv4/netfilter/ip_set_iphash.c | 164 +++ net/ipv4/netfilter/ip_set_ipmap.c | 158 ++ net/ipv4/netfilter/ip_set_ipporthash.c | 197 +++ net/ipv4/netfilter/ip_set_ipportiphash.c | 215 +++ net/ipv4/netfilter/ip_set_ipportnethash.c | 298 ++++ net/ipv4/netfilter/ip_set_iptree.c | 464 ++++++ net/ipv4/netfilter/ip_set_iptreemap.c | 699 +++++++++ net/ipv4/netfilter/ip_set_macipmap.c | 179 +++ net/ipv4/netfilter/ip_set_nethash.c | 218 +++ net/ipv4/netfilter/ip_set_portmap.c | 130 ++ net/ipv4/netfilter/ip_set_setlist.c | 324 +++++ net/ipv4/netfilter/ipt_SET.c | 256 ++++ net/ipv4/netfilter/ipt_set.c | 253 ++++ 35 files changed, 7083 insertions(+), 0 deletions(-) diff --git a/include/linux/netfilter_ipv4/ip_set.h b/include/linux/netfilter_ipv4/ip_set.h new file mode 100644 index 0000000..da17319 --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set.h @@ -0,0 +1,561 @@ +#ifndef _IP_SET_H +#define _IP_SET_H + +/* Copyright (C) 2000-2002 Joakim Axelsson + * Patrick Schaaf + * Martin Josefsson + * Copyright (C) 2003-2004 Jozsef Kadlecsik + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#if 0 +#define IP_SET_DEBUG +#endif + +/* + * A sockopt of such quality has hardly ever been seen before on the open + * market! This little beauty, hardly ever used: above 64, so it's + * traditionally used for firewalling, not touched (even once!) by the + * 2.0, 2.2 and 2.4 kernels! + * + * Comes with its own certificate of authenticity, valid anywhere in the + * Free world! + * + * Rusty, 19.4.2000 + */ +#define SO_IP_SET 83 + +/* + * Heavily modify by Joakim Axelsson 08.03.2002 + * - Made it more modulebased + * + * Additional heavy modifications by Jozsef Kadlecsik 22.02.2004 + * - bindings added + * - in order to "deal with" backward compatibility, renamed to ipset + */ + +/* + * Used so that the kernel module and ipset-binary can match their versions + */ +#define IP_SET_PROTOCOL_UNALIGNED 3 +#define IP_SET_PROTOCOL_VERSION 4 + +#define IP_SET_MAXNAMELEN 32 /* set names and set typenames */ + +/* Lets work with our own typedef for representing an IP address. + * We hope to make the code more portable, possibly to IPv6... + * + * The representation works in HOST byte order, because most set types + * will perform arithmetic operations and compare operations. + * + * For now the type is an uint32_t. + * + * Make sure to ONLY use the functions when translating and parsing + * in order to keep the host byte order and make it more portable: + * parse_ip() + * parse_mask() + * parse_ipandmask() + * ip_tostring() + * (Joakim: where are they???) + */ + +typedef uint32_t ip_set_ip_t; + +/* Sets are identified by an id in kernel space. Tweak with ip_set_id_t + * and IP_SET_INVALID_ID if you want to increase the max number of sets. + */ +typedef uint16_t ip_set_id_t; + +#define IP_SET_INVALID_ID 65535 + +/* How deep we follow bindings */ +#define IP_SET_MAX_BINDINGS 6 + +/* + * Option flags for kernel operations (ipt_set_info) + */ +#define IPSET_SRC 0x01 /* Source match/add */ +#define IPSET_DST 0x02 /* Destination match/add */ +#define IPSET_MATCH_INV 0x04 /* Inverse matching */ + +/* + * Set features + */ +#define IPSET_TYPE_IP 0x01 /* IP address type of set */ +#define IPSET_TYPE_PORT 0x02 /* Port type of set */ +#define IPSET_DATA_SINGLE 0x04 /* Single data storage */ +#define IPSET_DATA_DOUBLE 0x08 /* Double data storage */ +#define IPSET_DATA_TRIPLE 0x10 /* Triple data storage */ +#define IPSET_TYPE_IP1 0x20 /* IP address type of set */ +#define IPSET_TYPE_SETNAME 0x40 /* setname type of set */ + +/* Reserved keywords */ +#define IPSET_TOKEN_DEFAULT ":default:" +#define IPSET_TOKEN_ALL ":all:" + +/* SO_IP_SET operation constants, and their request struct types. + * + * Operation ids: + * 0-99: commands with version checking + * 100-199: add/del/test/bind/unbind + * 200-299: list, save, restore + */ + +/* Single shot operations: + * version, create, destroy, flush, rename and swap + * + * Sets are identified by name. + */ + +#define IP_SET_REQ_STD \ + unsigned op; \ + unsigned version; \ + char name[IP_SET_MAXNAMELEN] + +#define IP_SET_OP_CREATE 0x00000001 /* Create a new (empty) set */ +struct ip_set_req_create { + IP_SET_REQ_STD; + char typename[IP_SET_MAXNAMELEN]; +}; + +#define IP_SET_OP_DESTROY 0x00000002 /* Remove a (empty) set */ +struct ip_set_req_std { + IP_SET_REQ_STD; +}; + +#define IP_SET_OP_FLUSH 0x00000003 /* Remove all IPs in a set */ +/* Uses ip_set_req_std */ + +#define IP_SET_OP_RENAME 0x00000004 /* Rename a set */ +/* Uses ip_set_req_create */ + +#define IP_SET_OP_SWAP 0x00000005 /* Swap two sets */ +/* Uses ip_set_req_create */ + +union ip_set_name_index { + char name[IP_SET_MAXNAMELEN]; + ip_set_id_t index; +}; + +#define IP_SET_OP_GET_BYNAME 0x00000006 /* Get set index by name */ +struct ip_set_req_get_set { + unsigned op; + unsigned version; + union ip_set_name_index set; +}; + +#define IP_SET_OP_GET_BYINDEX 0x00000007 /* Get set name by index */ +/* Uses ip_set_req_get_set */ + +#define IP_SET_OP_VERSION 0x00000100 /* Ask kernel version */ +struct ip_set_req_version { + unsigned op; + unsigned version; +}; + +/* Double shots operations: + * add, del, test, bind and unbind. + * + * First we query the kernel to get the index and type of the target set, + * then issue the command. Validity of IP is checked in kernel in order + * to minimalize sockopt operations. + */ + +/* Get minimal set data for add/del/test/bind/unbind IP */ +#define IP_SET_OP_ADT_GET 0x00000010 /* Get set and type */ +struct ip_set_req_adt_get { + unsigned op; + unsigned version; + union ip_set_name_index set; + char typename[IP_SET_MAXNAMELEN]; +}; + +#define IP_SET_REQ_BYINDEX \ + unsigned op; \ + ip_set_id_t index; + +struct ip_set_req_adt { + IP_SET_REQ_BYINDEX; +}; + +#define IP_SET_OP_ADD_IP 0x00000101 /* Add an IP to a set */ +/* Uses ip_set_req_adt, with type specific addage */ + +#define IP_SET_OP_DEL_IP 0x00000102 /* Remove an IP from a set */ +/* Uses ip_set_req_adt, with type specific addage */ + +#define IP_SET_OP_TEST_IP 0x00000103 /* Test an IP in a set */ +/* Uses ip_set_req_adt, with type specific addage */ + +#define IP_SET_OP_BIND_SET 0x00000104 /* Bind an IP to a set */ +/* Uses ip_set_req_bind, with type specific addage */ +struct ip_set_req_bind { + IP_SET_REQ_BYINDEX; + char binding[IP_SET_MAXNAMELEN]; +}; + +#define IP_SET_OP_UNBIND_SET 0x00000105 /* Unbind an IP from a set */ +/* Uses ip_set_req_bind, with type speficic addage + * index = 0 means unbinding for all sets */ + +#define IP_SET_OP_TEST_BIND_SET 0x00000106 /* Test binding an IP to a set */ +/* Uses ip_set_req_bind, with type specific addage */ + +/* Multiple shots operations: list, save, restore. + * + * - check kernel version and query the max number of sets + * - get the basic information on all sets + * and size required for the next step + * - get actual set data: header, data, bindings + */ + +/* Get max_sets and the index of a queried set + */ +#define IP_SET_OP_MAX_SETS 0x00000020 +struct ip_set_req_max_sets { + unsigned op; + unsigned version; + ip_set_id_t max_sets; /* max_sets */ + ip_set_id_t sets; /* real number of sets */ + union ip_set_name_index set; /* index of set if name used */ +}; + +/* Get the id and name of the sets plus size for next step */ +#define IP_SET_OP_LIST_SIZE 0x00000201 +#define IP_SET_OP_SAVE_SIZE 0x00000202 +struct ip_set_req_setnames { + unsigned op; + ip_set_id_t index; /* set to list/save */ + u_int32_t size; /* size to get setdata */ + /* followed by sets number of struct ip_set_name_list */ +}; + +struct ip_set_name_list { + char name[IP_SET_MAXNAMELEN]; + char typename[IP_SET_MAXNAMELEN]; + ip_set_id_t index; + ip_set_id_t id; +}; + +/* The actual list operation */ +#define IP_SET_OP_LIST 0x00000203 +struct ip_set_req_list { + IP_SET_REQ_BYINDEX; + /* sets number of struct ip_set_list in reply */ +}; + +struct ip_set_list { + ip_set_id_t index; + ip_set_id_t binding; + u_int32_t ref; + u_int32_t header_size; /* Set header data of header_size */ + u_int32_t members_size; /* Set members data of members_size */ + u_int32_t bindings_size;/* Set bindings data of bindings_size */ +}; + +struct ip_set_hash_list { + ip_set_ip_t ip; + ip_set_id_t binding; +}; + +/* The save operation */ +#define IP_SET_OP_SAVE 0x00000204 +/* Uses ip_set_req_list, in the reply replaced by + * sets number of struct ip_set_save plus a marker + * ip_set_save followed by ip_set_hash_save structures. + */ +struct ip_set_save { + ip_set_id_t index; + ip_set_id_t binding; + u_int32_t header_size; /* Set header data of header_size */ + u_int32_t members_size; /* Set members data of members_size */ +}; + +/* At restoring, ip == 0 means default binding for the given set: */ +struct ip_set_hash_save { + ip_set_ip_t ip; + ip_set_id_t id; + ip_set_id_t binding; +}; + +/* The restore operation */ +#define IP_SET_OP_RESTORE 0x00000205 +/* Uses ip_set_req_setnames followed by ip_set_restore structures + * plus a marker ip_set_restore, followed by ip_set_hash_save + * structures. + */ +struct ip_set_restore { + char name[IP_SET_MAXNAMELEN]; + char typename[IP_SET_MAXNAMELEN]; + ip_set_id_t index; + u_int32_t header_size; /* Create data of header_size */ + u_int32_t members_size; /* Set members data of members_size */ +}; + +static inline int bitmap_bytes(ip_set_ip_t a, ip_set_ip_t b) +{ + return 4 * ((((b - a + 8) / 8) + 3) / 4); +} + +/* General limit for the elements in a set */ +#define MAX_RANGE 0x0000FFFF + +/* Alignment: 'unsigned long' unsupported */ +#define IPSET_ALIGNTO 4 +#define IPSET_ALIGN(len) (((len) + IPSET_ALIGNTO - 1) & ~(IPSET_ALIGNTO - 1)) +#define IPSET_VALIGN(len, old) ((old) ? (len) : IPSET_ALIGN(len)) + +#ifdef __KERNEL__ +#include +#include + +#define ip_set_printk(format, args...) \ + do { \ + printk("%s: %s: ", __FILE__, __FUNCTION__); \ + printk(format "\n" , ## args); \ + } while (0) + +#if defined(IP_SET_DEBUG) +#define DP(format, args...) \ + do { \ + printk("%s: %s (DBG): ", __FILE__, __FUNCTION__);\ + printk(format "\n" , ## args); \ + } while (0) +#define IP_SET_ASSERT(x) \ + do { \ + if (!(x)) \ + printk("IP_SET_ASSERT: %s:%i(%s)\n", \ + __FILE__, __LINE__, __FUNCTION__); \ + } while (0) +#else +#define DP(format, args...) +#define IP_SET_ASSERT(x) +#endif + +struct ip_set; + +/* + * The ip_set_type definition - one per set type, e.g. "ipmap". + * + * Each individual set has a pointer, set->type, going to one + * of these structures. Function pointers inside the structure implement + * the real behaviour of the sets. + * + * If not mentioned differently, the implementation behind the function + * pointers of a set_type, is expected to return 0 if ok, and a negative + * errno (e.g. -EINVAL) on error. + */ +struct ip_set_type { + struct list_head list; /* next in list of set types */ + + /* test for IP in set (kernel: iptables -m set src|dst) + * return 0 if not in set, 1 if in set. + */ + int (*testip_kernel) (struct ip_set *set, + const struct sk_buff * skb, + const u_int32_t *flags); + + /* test for IP in set (userspace: ipset -T set IP) + * return 0 if not in set, 1 if in set. + */ + int (*testip) (struct ip_set *set, + const void *data, u_int32_t size); + + /* + * Size of the data structure passed by when + * adding/deletin/testing an entry. + */ + u_int32_t reqsize; + + /* Add IP into set (userspace: ipset -A set IP) + * Return -EEXIST if the address is already in the set, + * and -ERANGE if the address lies outside the set bounds. + * If the address was not already in the set, 0 is returned. + */ + int (*addip) (struct ip_set *set, + const void *data, u_int32_t size); + + /* Add IP into set (kernel: iptables ... -j SET set src|dst) + * Return -EEXIST if the address is already in the set, + * and -ERANGE if the address lies outside the set bounds. + * If the address was not already in the set, 0 is returned. + */ + int (*addip_kernel) (struct ip_set *set, + const struct sk_buff * skb, + const u_int32_t *flags); + + /* remove IP from set (userspace: ipset -D set --entry x) + * Return -EEXIST if the address is NOT in the set, + * and -ERANGE if the address lies outside the set bounds. + * If the address really was in the set, 0 is returned. + */ + int (*delip) (struct ip_set *set, + const void *data, u_int32_t size); + + /* remove IP from set (kernel: iptables ... -j SET --entry x) + * Return -EEXIST if the address is NOT in the set, + * and -ERANGE if the address lies outside the set bounds. + * If the address really was in the set, 0 is returned. + */ + int (*delip_kernel) (struct ip_set *set, + const struct sk_buff * skb, + const u_int32_t *flags); + + /* new set creation - allocated type specific items + */ + int (*create) (struct ip_set *set, + const void *data, u_int32_t size); + + /* retry the operation after successfully tweaking the set + */ + int (*retry) (struct ip_set *set); + + /* set destruction - free type specific items + * There is no return value. + * Can be called only when child sets are destroyed. + */ + void (*destroy) (struct ip_set *set); + + /* set flushing - reset all bits in the set, or something similar. + * There is no return value. + */ + void (*flush) (struct ip_set *set); + + /* Listing: size needed for header + */ + u_int32_t header_size; + + /* Listing: Get the header + * + * Fill in the information in "data". + * This function is always run after list_header_size() under a + * writelock on the set. Therefor is the length of "data" always + * correct. + */ + void (*list_header) (const struct ip_set *set, + void *data); + + /* Listing: Get the size for the set members + */ + int (*list_members_size) (const struct ip_set *set, char dont_align); + + /* Listing: Get the set members + * + * Fill in the information in "data". + * This function is always run after list_member_size() under a + * writelock on the set. Therefor is the length of "data" always + * correct. + */ + void (*list_members) (const struct ip_set *set, + void *data, char dont_align); + + char typename[IP_SET_MAXNAMELEN]; + unsigned char features; + int protocol_version; + + /* Set this to THIS_MODULE if you are a module, otherwise NULL */ + struct module *me; +}; + +extern int ip_set_register_set_type(struct ip_set_type *set_type); +extern void ip_set_unregister_set_type(struct ip_set_type *set_type); + +/* A generic ipset */ +struct ip_set { + char name[IP_SET_MAXNAMELEN]; /* the name of the set */ + rwlock_t lock; /* lock for concurrency control */ + ip_set_id_t id; /* set id for swapping */ + atomic_t ref; /* in kernel and in hash references */ + struct ip_set_type *type; /* the set types */ + void *data; /* pooltype specific data */ +}; + +/* register and unregister set references */ +extern ip_set_id_t ip_set_get_byname(const char name[IP_SET_MAXNAMELEN]); +extern ip_set_id_t ip_set_get_byindex(ip_set_id_t index); +extern void ip_set_put_byindex(ip_set_id_t index); +extern ip_set_id_t ip_set_id(ip_set_id_t index); +extern ip_set_id_t __ip_set_get_byname(const char name[IP_SET_MAXNAMELEN], + struct ip_set **set); +extern void __ip_set_put_byindex(ip_set_id_t index); + +/* API for iptables set match, and SET target */ +extern int ip_set_addip_kernel(ip_set_id_t id, + const struct sk_buff *skb, + const u_int32_t *flags); +extern int ip_set_delip_kernel(ip_set_id_t id, + const struct sk_buff *skb, + const u_int32_t *flags); +extern int ip_set_testip_kernel(ip_set_id_t id, + const struct sk_buff *skb, + const u_int32_t *flags); + +/* Macros to generate functions */ + +#define STRUCT(pre, type) CONCAT2(pre, type) +#define CONCAT2(pre, type) struct pre##type + +#define FNAME(pre, mid, post) CONCAT3(pre, mid, post) +#define CONCAT3(pre, mid, post) pre##mid##post + +#define UADT0(type, adt, args...) \ +static int \ +FNAME(type,_u,adt)(struct ip_set *set, const void *data, u_int32_t size)\ +{ \ + const STRUCT(ip_set_req_,type) *req = data; \ + \ + return FNAME(type,_,adt)(set , ## args); \ +} + +#define UADT(type, adt, args...) \ + UADT0(type, adt, req->ip , ## args) + +#define KADT(type, adt, getfn, args...) \ +static int \ +FNAME(type,_k,adt)(struct ip_set *set, \ + const struct sk_buff *skb, \ + const u_int32_t *flags) \ +{ \ + ip_set_ip_t ip = getfn(skb, flags); \ + \ + KADT_CONDITION \ + return FNAME(type,_,adt)(set, ip , ##args); \ +} + +#define REGISTER_MODULE(type) \ +static int __init ip_set_##type##_init(void) \ +{ \ + init_max_page_size(); \ + return ip_set_register_set_type(&ip_set_##type); \ +} \ + \ +static void __exit ip_set_##type##_fini(void) \ +{ \ + /* FIXME: possible race with ip_set_create() */ \ + ip_set_unregister_set_type(&ip_set_##type); \ +} \ + \ +module_init(ip_set_##type##_init); \ +module_exit(ip_set_##type##_fini); + +/* Common functions */ + +static inline ip_set_ip_t +ipaddr(const struct sk_buff *skb, const u_int32_t *flags) +{ + return ntohl(flags[0] & IPSET_SRC ? ip_hdr(skb)->saddr : ip_hdr(skb)->daddr); +} + +#define jhash_ip(map, i, ip) jhash_1word(ip, *(map->initval + i)) + +#define pack_ip_port(map, ip, port) \ + (port + ((ip - ((map)->first_ip)) << 16)) + +#endif /* __KERNEL__ */ + +#define UNUSED __attribute__ ((unused)) + +#endif /*_IP_SET_H*/ diff --git a/include/linux/netfilter_ipv4/ip_set_bitmaps.h b/include/linux/netfilter_ipv4/ip_set_bitmaps.h new file mode 100644 index 0000000..da3493f --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_bitmaps.h @@ -0,0 +1,120 @@ +#ifndef __IP_SET_BITMAPS_H +#define __IP_SET_BITMAPS_H + +/* Macros to generate functions */ + +#ifdef __KERNEL__ +#define BITMAP_CREATE(type) \ +static int \ +type##_create(struct ip_set *set, const void *data, u_int32_t size) \ +{ \ + int newbytes; \ + const struct ip_set_req_##type##_create *req = data; \ + struct ip_set_##type *map; \ + \ + if (req->from > req->to) { \ + DP("bad range"); \ + return -ENOEXEC; \ + } \ + \ + map = kmalloc(sizeof(struct ip_set_##type), GFP_KERNEL); \ + if (!map) { \ + DP("out of memory for %zu bytes", \ + sizeof(struct ip_set_##type)); \ + return -ENOMEM; \ + } \ + map->first_ip = req->from; \ + map->last_ip = req->to; \ + \ + newbytes = __##type##_create(req, map); \ + if (newbytes < 0) { \ + kfree(map); \ + return newbytes; \ + } \ + \ + map->size = newbytes; \ + map->members = ip_set_malloc(newbytes); \ + if (!map->members) { \ + DP("out of memory for %i bytes", newbytes); \ + kfree(map); \ + return -ENOMEM; \ + } \ + memset(map->members, 0, newbytes); \ + \ + set->data = map; \ + return 0; \ +} + +#define BITMAP_DESTROY(type) \ +static void \ +type##_destroy(struct ip_set *set) \ +{ \ + struct ip_set_##type *map = set->data; \ + \ + ip_set_free(map->members, map->size); \ + kfree(map); \ + \ + set->data = NULL; \ +} + +#define BITMAP_FLUSH(type) \ +static void \ +type##_flush(struct ip_set *set) \ +{ \ + struct ip_set_##type *map = set->data; \ + memset(map->members, 0, map->size); \ +} + +#define BITMAP_LIST_HEADER(type) \ +static void \ +type##_list_header(const struct ip_set *set, void *data) \ +{ \ + const struct ip_set_##type *map = set->data; \ + struct ip_set_req_##type##_create *header = data; \ + \ + header->from = map->first_ip; \ + header->to = map->last_ip; \ + __##type##_list_header(map, header); \ +} + +#define BITMAP_LIST_MEMBERS_SIZE(type, dtype, sizeid, testfn) \ +static int \ +type##_list_members_size(const struct ip_set *set, char dont_align) \ +{ \ + const struct ip_set_##type *map = set->data; \ + ip_set_ip_t i, elements = 0; \ + \ + if (dont_align) \ + return map->size; \ + \ + for (i = 0; i < sizeid; i++) \ + if (testfn) \ + elements++; \ + \ + return elements * IPSET_ALIGN(sizeof(dtype)); \ +} + +#define IP_SET_TYPE(type, __features) \ +struct ip_set_type ip_set_##type = { \ + .typename = #type, \ + .features = __features, \ + .protocol_version = IP_SET_PROTOCOL_VERSION, \ + .create = &type##_create, \ + .destroy = &type##_destroy, \ + .flush = &type##_flush, \ + .reqsize = sizeof(struct ip_set_req_##type), \ + .addip = &type##_uadd, \ + .addip_kernel = &type##_kadd, \ + .delip = &type##_udel, \ + .delip_kernel = &type##_kdel, \ + .testip = &type##_utest, \ + .testip_kernel = &type##_ktest, \ + .header_size = sizeof(struct ip_set_req_##type##_create),\ + .list_header = &type##_list_header, \ + .list_members_size = &type##_list_members_size, \ + .list_members = &type##_list_members, \ + .me = THIS_MODULE, \ +}; +#endif /* __KERNEL */ + +#endif /* __IP_SET_BITMAPS_H */ diff --git a/include/linux/netfilter_ipv4/ip_set_compat.h b/include/linux/netfilter_ipv4/ip_set_compat.h new file mode 100644 index 0000000..9f17397 --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_compat.h @@ -0,0 +1,92 @@ +#ifndef _IP_SET_COMPAT_H +#define _IP_SET_COMPAT_H + +#ifdef __KERNEL__ +#include + +/* Arrgh */ +#ifdef MODULE +#define __MOD_INC(foo) __MOD_INC_USE_COUNT(foo) +#define __MOD_DEC(foo) __MOD_DEC_USE_COUNT(foo) +#else +#define __MOD_INC(foo) 1 +#define __MOD_DEC(foo) +#endif + +/* Backward compatibility */ +#ifndef __nocast +#define __nocast +#endif +#ifndef __bitwise__ +#define __bitwise__ +#endif + +/* Compatibility glue code */ +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) +#include +#define DEFINE_RWLOCK(x) rwlock_t x = RW_LOCK_UNLOCKED +#define try_module_get(x) __MOD_INC(x) +#define module_put(x) __MOD_DEC(x) +#define __clear_bit(nr, addr) clear_bit(nr, addr) +#define __set_bit(nr, addr) set_bit(nr, addr) +#define __test_and_set_bit(nr, addr) test_and_set_bit(nr, addr) +#define __test_and_clear_bit(nr, addr) test_and_clear_bit(nr, addr) + +typedef unsigned __bitwise__ gfp_t; + +static inline void *kzalloc(size_t size, gfp_t flags) +{ + void *data = kmalloc(size, flags); + + if (data) + memset(data, 0, size); + + return data; +} +#endif + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,20) +#define __KMEM_CACHE_T__ kmem_cache_t +#else +#define __KMEM_CACHE_T__ struct kmem_cache +#endif + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,22) +#define ip_hdr(skb) ((skb)->nh.iph) +#define skb_mac_header(skb) ((skb)->mac.raw) +#define eth_hdr(skb) ((struct ethhdr *)skb_mac_header(skb)) +#endif + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23) +#include +#define KMEM_CACHE_CREATE(name, size) \ + kmem_cache_create(name, size, 0, 0, NULL, NULL) +#else +#define KMEM_CACHE_CREATE(name, size) \ + kmem_cache_create(name, size, 0, 0, NULL) +#endif + +#ifndef NIPQUAD +#define NIPQUAD(addr) \ + ((unsigned char *)&addr)[0], \ + ((unsigned char *)&addr)[1], \ + ((unsigned char *)&addr)[2], \ + ((unsigned char *)&addr)[3] +#endif + +#ifndef HIPQUAD +#if defined(__LITTLE_ENDIAN) +#define HIPQUAD(addr) \ + ((unsigned char *)&addr)[3], \ + ((unsigned char *)&addr)[2], \ + ((unsigned char *)&addr)[1], \ + ((unsigned char *)&addr)[0] +#elif defined(__BIG_ENDIAN) +#define HIPQUAD NIPQUAD +#else +#error "Please fix asm/byteorder.h" +#endif /* __LITTLE_ENDIAN */ +#endif + +#endif /* __KERNEL__ */ +#endif /* _IP_SET_COMPAT_H */ diff --git a/include/linux/netfilter_ipv4/ip_set_getport.h b/include/linux/netfilter_ipv4/ip_set_getport.h new file mode 100644 index 0000000..18ed729 --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_getport.h @@ -0,0 +1,48 @@ +#ifndef _IP_SET_GETPORT_H +#define _IP_SET_GETPORT_H + +#ifdef __KERNEL__ + +#define INVALID_PORT (MAX_RANGE + 1) + +/* We must handle non-linear skbs */ +static inline ip_set_ip_t +get_port(const struct sk_buff *skb, const u_int32_t *flags) +{ + struct iphdr *iph = ip_hdr(skb); + u_int16_t offset = ntohs(iph->frag_off) & IP_OFFSET; + switch (iph->protocol) { + case IPPROTO_TCP: { + struct tcphdr tcph; + + /* See comments at tcp_match in ip_tables.c */ + if (offset) + return INVALID_PORT; + + if (skb_copy_bits(skb, ip_hdr(skb)->ihl*4, &tcph, sizeof(tcph)) < 0) + /* No choice either */ + return INVALID_PORT; + + return ntohs(flags[0] & IPSET_SRC ? + tcph.source : tcph.dest); + } + case IPPROTO_UDP: { + struct udphdr udph; + + if (offset) + return INVALID_PORT; + + if (skb_copy_bits(skb, ip_hdr(skb)->ihl*4, &udph, sizeof(udph)) < 0) + /* No choice either */ + return INVALID_PORT; + + return ntohs(flags[0] & IPSET_SRC ? + udph.source : udph.dest); + } + default: + return INVALID_PORT; + } +} +#endif /* __KERNEL__ */ + +#endif /*_IP_SET_GETPORT_H*/ diff --git a/include/linux/netfilter_ipv4/ip_set_hashes.h b/include/linux/netfilter_ipv4/ip_set_hashes.h new file mode 100644 index 0000000..8eeced3 --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_hashes.h @@ -0,0 +1,314 @@ +#ifndef __IP_SET_HASHES_H +#define __IP_SET_HASHES_H + +#define initval_t uint32_t + +/* Macros to generate functions */ + +#ifdef __KERNEL__ +#define HASH_RETRY0(type, dtype, cond) \ +static int \ +type##_retry(struct ip_set *set) \ +{ \ + struct ip_set_##type *map = set->data, *tmp; \ + dtype *elem; \ + void *members; \ + u_int32_t i, hashsize = map->hashsize; \ + int res; \ + \ + if (map->resize == 0) \ + return -ERANGE; \ + \ + again: \ + res = 0; \ + \ + /* Calculate new hash size */ \ + hashsize += (hashsize * map->resize)/100; \ + if (hashsize == map->hashsize) \ + hashsize++; \ + \ + ip_set_printk("rehashing of set %s triggered: " \ + "hashsize grows from %lu to %lu", \ + set->name, \ + (long unsigned)map->hashsize, \ + (long unsigned)hashsize); \ + \ + tmp = kmalloc(sizeof(struct ip_set_##type) \ + + map->probes * sizeof(initval_t), GFP_ATOMIC); \ + if (!tmp) { \ + DP("out of memory for %zu bytes", \ + sizeof(struct ip_set_##type) \ + + map->probes * sizeof(initval_t)); \ + return -ENOMEM; \ + } \ + tmp->members = harray_malloc(hashsize, sizeof(dtype), GFP_ATOMIC);\ + if (!tmp->members) { \ + DP("out of memory for %zu bytes", hashsize * sizeof(dtype));\ + kfree(tmp); \ + return -ENOMEM; \ + } \ + tmp->hashsize = hashsize; \ + tmp->elements = 0; \ + tmp->probes = map->probes; \ + tmp->resize = map->resize; \ + memcpy(tmp->initval, map->initval, map->probes * sizeof(initval_t));\ + __##type##_retry(tmp, map); \ + \ + write_lock_bh(&set->lock); \ + map = set->data; /* Play safe */ \ + for (i = 0; i < map->hashsize && res == 0; i++) { \ + elem = HARRAY_ELEM(map->members, dtype *, i); \ + if (cond) \ + res = __##type##_add(tmp, elem); \ + } \ + if (res) { \ + /* Failure, try again */ \ + write_unlock_bh(&set->lock); \ + harray_free(tmp->members); \ + kfree(tmp); \ + goto again; \ + } \ + \ + /* Success at resizing! */ \ + members = map->members; \ + \ + map->hashsize = tmp->hashsize; \ + map->members = tmp->members; \ + write_unlock_bh(&set->lock); \ + \ + harray_free(members); \ + kfree(tmp); \ + \ + return 0; \ +} + +#define HASH_RETRY(type, dtype) \ + HASH_RETRY0(type, dtype, *elem) + +#define HASH_RETRY2(type, dtype) \ + HASH_RETRY0(type, dtype, elem->ip || elem->ip1) + +#define HASH_CREATE(type, dtype) \ +static int \ +type##_create(struct ip_set *set, const void *data, u_int32_t size) \ +{ \ + const struct ip_set_req_##type##_create *req = data; \ + struct ip_set_##type *map; \ + uint16_t i; \ + \ + if (req->hashsize < 1) { \ + ip_set_printk("hashsize too small"); \ + return -ENOEXEC; \ + } \ + \ + if (req->probes < 1) { \ + ip_set_printk("probes too small"); \ + return -ENOEXEC; \ + } \ + \ + map = kmalloc(sizeof(struct ip_set_##type) \ + + req->probes * sizeof(initval_t), GFP_KERNEL); \ + if (!map) { \ + DP("out of memory for %zu bytes", \ + sizeof(struct ip_set_##type) \ + + req->probes * sizeof(initval_t)); \ + return -ENOMEM; \ + } \ + for (i = 0; i < req->probes; i++) \ + get_random_bytes(((initval_t *) map->initval)+i, 4); \ + map->elements = 0; \ + map->hashsize = req->hashsize; \ + map->probes = req->probes; \ + map->resize = req->resize; \ + if (__##type##_create(req, map)) { \ + kfree(map); \ + return -ENOEXEC; \ + } \ + map->members = harray_malloc(map->hashsize, sizeof(dtype), GFP_KERNEL);\ + if (!map->members) { \ + DP("out of memory for %zu bytes", map->hashsize * sizeof(dtype));\ + kfree(map); \ + return -ENOMEM; \ + } \ + \ + set->data = map; \ + return 0; \ +} + +#define HASH_DESTROY(type) \ +static void \ +type##_destroy(struct ip_set *set) \ +{ \ + struct ip_set_##type *map = set->data; \ + \ + harray_free(map->members); \ + kfree(map); \ + \ + set->data = NULL; \ +} + +#define HASH_FLUSH(type, dtype) \ +static void \ +type##_flush(struct ip_set *set) \ +{ \ + struct ip_set_##type *map = set->data; \ + harray_flush(map->members, map->hashsize, sizeof(dtype)); \ + map->elements = 0; \ +} + +#define HASH_FLUSH_CIDR(type, dtype) \ +static void \ +type##_flush(struct ip_set *set) \ +{ \ + struct ip_set_##type *map = set->data; \ + harray_flush(map->members, map->hashsize, sizeof(dtype)); \ + memset(map->cidr, 0, sizeof(map->cidr)); \ + memset(map->nets, 0, sizeof(map->nets)); \ + map->elements = 0; \ +} + +#define HASH_LIST_HEADER(type) \ +static void \ +type##_list_header(const struct ip_set *set, void *data) \ +{ \ + const struct ip_set_##type *map = set->data; \ + struct ip_set_req_##type##_create *header = data; \ + \ + header->hashsize = map->hashsize; \ + header->probes = map->probes; \ + header->resize = map->resize; \ + __##type##_list_header(map, header); \ +} + +#define HASH_LIST_MEMBERS_SIZE(type, dtype) \ +static int \ +type##_list_members_size(const struct ip_set *set, char dont_align) \ +{ \ + const struct ip_set_##type *map = set->data; \ + \ + return (map->elements * IPSET_VALIGN(sizeof(dtype), dont_align));\ +} + +#define HASH_LIST_MEMBERS(type, dtype) \ +static void \ +type##_list_members(const struct ip_set *set, void *data, char dont_align)\ +{ \ + const struct ip_set_##type *map = set->data; \ + dtype *elem, *d; \ + uint32_t i, n = 0; \ + \ + for (i = 0; i < map->hashsize; i++) { \ + elem = HARRAY_ELEM(map->members, dtype *, i); \ + if (*elem) { \ + d = data + n * IPSET_VALIGN(sizeof(dtype), dont_align);\ + *d = *elem; \ + n++; \ + } \ + } \ +} + +#define HASH_LIST_MEMBERS_MEMCPY(type, dtype, nonzero) \ +static void \ +type##_list_members(const struct ip_set *set, void *data, char dont_align)\ +{ \ + const struct ip_set_##type *map = set->data; \ + dtype *elem; \ + uint32_t i, n = 0; \ + \ + for (i = 0; i < map->hashsize; i++) { \ + elem = HARRAY_ELEM(map->members, dtype *, i); \ + if (nonzero) { \ + memcpy(data + n * IPSET_VALIGN(sizeof(dtype), dont_align),\ + elem, sizeof(dtype)); \ + n++; \ + } \ + } \ +} + +#define IP_SET_RTYPE(type, __features) \ +struct ip_set_type ip_set_##type = { \ + .typename = #type, \ + .features = __features, \ + .protocol_version = IP_SET_PROTOCOL_VERSION, \ + .create = &type##_create, \ + .retry = &type##_retry, \ + .destroy = &type##_destroy, \ + .flush = &type##_flush, \ + .reqsize = sizeof(struct ip_set_req_##type), \ + .addip = &type##_uadd, \ + .addip_kernel = &type##_kadd, \ + .delip = &type##_udel, \ + .delip_kernel = &type##_kdel, \ + .testip = &type##_utest, \ + .testip_kernel = &type##_ktest, \ + .header_size = sizeof(struct ip_set_req_##type##_create),\ + .list_header = &type##_list_header, \ + .list_members_size = &type##_list_members_size, \ + .list_members = &type##_list_members, \ + .me = THIS_MODULE, \ +}; + +/* Helper functions */ +static inline void +add_cidr_size(uint8_t *cidr, uint8_t size) +{ + uint8_t next; + int i; + + for (i = 0; i < 30 && cidr[i]; i++) { + if (cidr[i] < size) { + next = cidr[i]; + cidr[i] = size; + size = next; + } + } + if (i < 30) + cidr[i] = size; +} + +static inline void +del_cidr_size(uint8_t *cidr, uint8_t size) +{ + int i; + + for (i = 0; i < 29 && cidr[i]; i++) { + if (cidr[i] == size) + cidr[i] = size = cidr[i+1]; + } + cidr[29] = 0; +} +#else +#include +#endif /* __KERNEL */ + +#ifndef UINT16_MAX +#define UINT16_MAX 65535 +#endif + +static unsigned char shifts[] = {255, 253, 249, 241, 225, 193, 129, 1}; + +static inline ip_set_ip_t +pack_ip_cidr(ip_set_ip_t ip, unsigned char cidr) +{ + ip_set_ip_t addr, *paddr = &addr; + unsigned char n, t, *a; + + addr = htonl(ip & (0xFFFFFFFF << (32 - (cidr)))); +#ifdef __KERNEL__ + DP("ip:%u.%u.%u.%u/%u", NIPQUAD(addr), cidr); +#endif + n = cidr / 8; + t = cidr % 8; + a = &((unsigned char *)paddr)[n]; + *a = *a /(1 << (8 - t)) + shifts[t]; +#ifdef __KERNEL__ + DP("n: %u, t: %u, a: %u", n, t, *a); + DP("ip:%u.%u.%u.%u/%u, %u.%u.%u.%u", + HIPQUAD(ip), cidr, NIPQUAD(addr)); +#endif + + return ntohl(addr); +} + + +#endif /* __IP_SET_HASHES_H */ diff --git a/include/linux/netfilter_ipv4/ip_set_iphash.h b/include/linux/netfilter_ipv4/ip_set_iphash.h new file mode 100644 index 0000000..0a0c7e8 --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_iphash.h @@ -0,0 +1,30 @@ +#ifndef __IP_SET_IPHASH_H +#define __IP_SET_IPHASH_H + +#include +#include + +#define SETTYPE_NAME "iphash" + +struct ip_set_iphash { + ip_set_ip_t *members; /* the iphash proper */ + uint32_t elements; /* number of elements */ + uint32_t hashsize; /* hash size */ + uint16_t probes; /* max number of probes */ + uint16_t resize; /* resize factor in percent */ + ip_set_ip_t netmask; /* netmask */ + initval_t initval[0]; /* initvals for jhash_1word */ +}; + +struct ip_set_req_iphash_create { + uint32_t hashsize; + uint16_t probes; + uint16_t resize; + ip_set_ip_t netmask; +}; + +struct ip_set_req_iphash { + ip_set_ip_t ip; +}; + +#endif /* __IP_SET_IPHASH_H */ diff --git a/include/linux/netfilter_ipv4/ip_set_ipmap.h b/include/linux/netfilter_ipv4/ip_set_ipmap.h new file mode 100644 index 0000000..d16c0ae --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_ipmap.h @@ -0,0 +1,57 @@ +#ifndef __IP_SET_IPMAP_H +#define __IP_SET_IPMAP_H + +#include +#include + +#define SETTYPE_NAME "ipmap" + +struct ip_set_ipmap { + void *members; /* the ipmap proper */ + ip_set_ip_t first_ip; /* host byte order, included in range */ + ip_set_ip_t last_ip; /* host byte order, included in range */ + ip_set_ip_t netmask; /* subnet netmask */ + ip_set_ip_t sizeid; /* size of set in IPs */ + ip_set_ip_t hosts; /* number of hosts in a subnet */ + u_int32_t size; /* size of the ipmap proper */ +}; + +struct ip_set_req_ipmap_create { + ip_set_ip_t from; + ip_set_ip_t to; + ip_set_ip_t netmask; +}; + +struct ip_set_req_ipmap { + ip_set_ip_t ip; +}; + +static inline unsigned int +mask_to_bits(ip_set_ip_t mask) +{ + unsigned int bits = 32; + ip_set_ip_t maskaddr; + + if (mask == 0xFFFFFFFF) + return bits; + + maskaddr = 0xFFFFFFFE; + while (--bits > 0 && maskaddr != mask) + maskaddr <<= 1; + + return bits; +} + +static inline ip_set_ip_t +range_to_mask(ip_set_ip_t from, ip_set_ip_t to, unsigned int *bits) +{ + ip_set_ip_t mask = 0xFFFFFFFE; + + *bits = 32; + while (--(*bits) > 0 && mask && (to & mask) != from) + mask <<= 1; + + return mask; +} + +#endif /* __IP_SET_IPMAP_H */ diff --git a/include/linux/netfilter_ipv4/ip_set_ipporthash.h b/include/linux/netfilter_ipv4/ip_set_ipporthash.h new file mode 100644 index 0000000..a3b781a --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_ipporthash.h @@ -0,0 +1,33 @@ +#ifndef __IP_SET_IPPORTHASH_H +#define __IP_SET_IPPORTHASH_H + +#include +#include + +#define SETTYPE_NAME "ipporthash" + +struct ip_set_ipporthash { + ip_set_ip_t *members; /* the ipporthash proper */ + uint32_t elements; /* number of elements */ + uint32_t hashsize; /* hash size */ + uint16_t probes; /* max number of probes */ + uint16_t resize; /* resize factor in percent */ + ip_set_ip_t first_ip; /* host byte order, included in range */ + ip_set_ip_t last_ip; /* host byte order, included in range */ + initval_t initval[0]; /* initvals for jhash_1word */ +}; + +struct ip_set_req_ipporthash_create { + uint32_t hashsize; + uint16_t probes; + uint16_t resize; + ip_set_ip_t from; + ip_set_ip_t to; +}; + +struct ip_set_req_ipporthash { + ip_set_ip_t ip; + ip_set_ip_t port; +}; + +#endif /* __IP_SET_IPPORTHASH_H */ diff --git a/include/linux/netfilter_ipv4/ip_set_ipportiphash.h b/include/linux/netfilter_ipv4/ip_set_ipportiphash.h new file mode 100644 index 0000000..2202c51 --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_ipportiphash.h @@ -0,0 +1,39 @@ +#ifndef __IP_SET_IPPORTIPHASH_H +#define __IP_SET_IPPORTIPHASH_H + +#include +#include + +#define SETTYPE_NAME "ipportiphash" + +struct ipportip { + ip_set_ip_t ip; + ip_set_ip_t ip1; +}; + +struct ip_set_ipportiphash { + struct ipportip *members; /* the ipportip proper */ + uint32_t elements; /* number of elements */ + uint32_t hashsize; /* hash size */ + uint16_t probes; /* max number of probes */ + uint16_t resize; /* resize factor in percent */ + ip_set_ip_t first_ip; /* host byte order, included in range */ + ip_set_ip_t last_ip; /* host byte order, included in range */ + initval_t initval[0]; /* initvals for jhash_1word */ +}; + +struct ip_set_req_ipportiphash_create { + uint32_t hashsize; + uint16_t probes; + uint16_t resize; + ip_set_ip_t from; + ip_set_ip_t to; +}; + +struct ip_set_req_ipportiphash { + ip_set_ip_t ip; + ip_set_ip_t port; + ip_set_ip_t ip1; +}; + +#endif /* __IP_SET_IPPORTIPHASH_H */ diff --git a/include/linux/netfilter_ipv4/ip_set_ipportnethash.h b/include/linux/netfilter_ipv4/ip_set_ipportnethash.h new file mode 100644 index 0000000..73b2430 --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_ipportnethash.h @@ -0,0 +1,42 @@ +#ifndef __IP_SET_IPPORTNETHASH_H +#define __IP_SET_IPPORTNETHASH_H + +#include +#include + +#define SETTYPE_NAME "ipportnethash" + +struct ipportip { + ip_set_ip_t ip; + ip_set_ip_t ip1; +}; + +struct ip_set_ipportnethash { + struct ipportip *members; /* the ipportip proper */ + uint32_t elements; /* number of elements */ + uint32_t hashsize; /* hash size */ + uint16_t probes; /* max number of probes */ + uint16_t resize; /* resize factor in percent */ + ip_set_ip_t first_ip; /* host byte order, included in range */ + ip_set_ip_t last_ip; /* host byte order, included in range */ + uint8_t cidr[30]; /* CIDR sizes */ + uint16_t nets[30]; /* nr of nets by CIDR sizes */ + initval_t initval[0]; /* initvals for jhash_1word */ +}; + +struct ip_set_req_ipportnethash_create { + uint32_t hashsize; + uint16_t probes; + uint16_t resize; + ip_set_ip_t from; + ip_set_ip_t to; +}; + +struct ip_set_req_ipportnethash { + ip_set_ip_t ip; + ip_set_ip_t port; + ip_set_ip_t ip1; + uint8_t cidr; +}; + +#endif /* __IP_SET_IPPORTNETHASH_H */ diff --git a/include/linux/netfilter_ipv4/ip_set_iptree.h b/include/linux/netfilter_ipv4/ip_set_iptree.h new file mode 100644 index 0000000..36bf5ac --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_iptree.h @@ -0,0 +1,39 @@ +#ifndef __IP_SET_IPTREE_H +#define __IP_SET_IPTREE_H + +#include + +#define SETTYPE_NAME "iptree" + +struct ip_set_iptreed { + unsigned long expires[256]; /* x.x.x.ADDR */ +}; + +struct ip_set_iptreec { + struct ip_set_iptreed *tree[256]; /* x.x.ADDR.* */ +}; + +struct ip_set_iptreeb { + struct ip_set_iptreec *tree[256]; /* x.ADDR.*.* */ +}; + +struct ip_set_iptree { + unsigned int timeout; + unsigned int gc_interval; +#ifdef __KERNEL__ + uint32_t elements; /* number of elements */ + struct timer_list gc; + struct ip_set_iptreeb *tree[256]; /* ADDR.*.*.* */ +#endif +}; + +struct ip_set_req_iptree_create { + unsigned int timeout; +}; + +struct ip_set_req_iptree { + ip_set_ip_t ip; + unsigned int timeout; +}; + +#endif /* __IP_SET_IPTREE_H */ diff --git a/include/linux/netfilter_ipv4/ip_set_iptreemap.h b/include/linux/netfilter_ipv4/ip_set_iptreemap.h new file mode 100644 index 0000000..6ea771a --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_iptreemap.h @@ -0,0 +1,40 @@ +#ifndef __IP_SET_IPTREEMAP_H +#define __IP_SET_IPTREEMAP_H + +#include + +#define SETTYPE_NAME "iptreemap" + +#ifdef __KERNEL__ +struct ip_set_iptreemap_d { + unsigned char bitmap[32]; /* x.x.x.y */ +}; + +struct ip_set_iptreemap_c { + struct ip_set_iptreemap_d *tree[256]; /* x.x.y.x */ +}; + +struct ip_set_iptreemap_b { + struct ip_set_iptreemap_c *tree[256]; /* x.y.x.x */ + unsigned char dirty[32]; +}; +#endif + +struct ip_set_iptreemap { + unsigned int gc_interval; +#ifdef __KERNEL__ + struct timer_list gc; + struct ip_set_iptreemap_b *tree[256]; /* y.x.x.x */ +#endif +}; + +struct ip_set_req_iptreemap_create { + unsigned int gc_interval; +}; + +struct ip_set_req_iptreemap { + ip_set_ip_t ip; + ip_set_ip_t end; +}; + +#endif /* __IP_SET_IPTREEMAP_H */ diff --git a/include/linux/netfilter_ipv4/ip_set_jhash.h b/include/linux/netfilter_ipv4/ip_set_jhash.h new file mode 100644 index 0000000..2000b9f --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_jhash.h @@ -0,0 +1,157 @@ +#ifndef _LINUX_JHASH_H +#define _LINUX_JHASH_H + +/* jhash.h: Jenkins hash support. + * + * Copyright (C) 2006. Bob Jenkins (bob_jenkins@burtleburtle.net) + * + * http://burtleburtle.net/bob/hash/ + * + * These are the credits from Bob's sources: + * + * lookup3.c, by Bob Jenkins, May 2006, Public Domain. + * + * These are functions for producing 32-bit hashes for hash table lookup. + * hashword(), hashlittle(), hashlittle2(), hashbig(), mix(), and final() + * are externally useful functions. Routines to test the hash are included + * if SELF_TEST is defined. You can use this free for any purpose. It's in + * the public domain. It has no warranty. + * + * Copyright (C) 2009 Jozsef Kadlecsik (kadlec@blackhole.kfki.hu) + * + * I've modified Bob's hash to be useful in the Linux kernel, and + * any bugs present are my fault. Jozsef + */ + +#define __rot(x,k) (((x)<<(k)) | ((x)>>(32-(k)))) + +/* __jhash_mix - mix 3 32-bit values reversibly. */ +#define __jhash_mix(a,b,c) \ +{ \ + a -= c; a ^= __rot(c, 4); c += b; \ + b -= a; b ^= __rot(a, 6); a += c; \ + c -= b; c ^= __rot(b, 8); b += a; \ + a -= c; a ^= __rot(c,16); c += b; \ + b -= a; b ^= __rot(a,19); a += c; \ + c -= b; c ^= __rot(b, 4); b += a; \ +} + +/* __jhash_final - final mixing of 3 32-bit values (a,b,c) into c */ +#define __jhash_final(a,b,c) \ +{ \ + c ^= b; c -= __rot(b,14); \ + a ^= c; a -= __rot(c,11); \ + b ^= a; b -= __rot(a,25); \ + c ^= b; c -= __rot(b,16); \ + a ^= c; a -= __rot(c,4); \ + b ^= a; b -= __rot(a,14); \ + c ^= b; c -= __rot(b,24); \ +} + +/* The golden ration: an arbitrary value */ +#define JHASH_GOLDEN_RATIO 0xdeadbeef + +/* The most generic version, hashes an arbitrary sequence + * of bytes. No alignment or length assumptions are made about + * the input key. The result depends on endianness. + */ +static inline u32 jhash(const void *key, u32 length, u32 initval) +{ + u32 a,b,c; + const u8 *k = key; + + /* Set up the internal state */ + a = b = c = JHASH_GOLDEN_RATIO + length + initval; + + /* all but the last block: affect some 32 bits of (a,b,c) */ + while (length > 12) { + a += (k[0] + ((u32)k[1]<<8) + ((u32)k[2]<<16) + ((u32)k[3]<<24)); + b += (k[4] + ((u32)k[5]<<8) + ((u32)k[6]<<16) + ((u32)k[7]<<24)); + c += (k[8] + ((u32)k[9]<<8) + ((u32)k[10]<<16) + ((u32)k[11]<<24)); + __jhash_mix(a, b, c); + length -= 12; + k += 12; + } + + /* last block: affect all 32 bits of (c) */ + /* all the case statements fall through */ + switch (length) { + case 12: c += (u32)k[11]<<24; + case 11: c += (u32)k[10]<<16; + case 10: c += (u32)k[9]<<8; + case 9 : c += k[8]; + case 8 : b += (u32)k[7]<<24; + case 7 : b += (u32)k[6]<<16; + case 6 : b += (u32)k[5]<<8; + case 5 : b += k[4]; + case 4 : a += (u32)k[3]<<24; + case 3 : a += (u32)k[2]<<16; + case 2 : a += (u32)k[1]<<8; + case 1 : a += k[0]; + __jhash_final(a, b, c); + case 0 : + break; + } + + return c; +} + +/* A special optimized version that handles 1 or more of u32s. + * The length parameter here is the number of u32s in the key. + */ +static inline u32 jhash2(const u32 *k, u32 length, u32 initval) +{ + u32 a, b, c; + + /* Set up the internal state */ + a = b = c = JHASH_GOLDEN_RATIO + (length<<2) + initval; + + /* handle most of the key */ + while (length > 3) { + a += k[0]; + b += k[1]; + c += k[2]; + __jhash_mix(a, b, c); + length -= 3; + k += 3; + } + + /* handle the last 3 u32's */ + /* all the case statements fall through */ + switch (length) { + case 3: c += k[2]; + case 2: b += k[1]; + case 1: a += k[0]; + __jhash_final(a, b, c); + case 0: /* case 0: nothing left to add */ + break; + } + + return c; +} + +/* A special ultra-optimized versions that knows they are hashing exactly + * 3, 2 or 1 word(s). + */ +static inline u32 jhash_3words(u32 a, u32 b, u32 c, u32 initval) +{ + a += JHASH_GOLDEN_RATIO + initval; + b += JHASH_GOLDEN_RATIO + initval; + c += JHASH_GOLDEN_RATIO + initval; + + __jhash_final(a, b, c); + + return c; +} + +static inline u32 jhash_2words(u32 a, u32 b, u32 initval) +{ + return jhash_3words(0, a, b, initval); +} + +static inline u32 jhash_1word(u32 a, u32 initval) +{ + return jhash_3words(0, 0, a, initval); +} + +#endif /* _LINUX_JHASH_H */ diff --git a/include/linux/netfilter_ipv4/ip_set_macipmap.h b/include/linux/netfilter_ipv4/ip_set_macipmap.h new file mode 100644 index 0000000..0615e9f --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_macipmap.h @@ -0,0 +1,39 @@ +#ifndef __IP_SET_MACIPMAP_H +#define __IP_SET_MACIPMAP_H + +#include +#include + +#define SETTYPE_NAME "macipmap" + +/* general flags */ +#define IPSET_MACIP_MATCHUNSET 1 + +/* per ip flags */ +#define IPSET_MACIP_ISSET 1 + +struct ip_set_macipmap { + void *members; /* the macipmap proper */ + ip_set_ip_t first_ip; /* host byte order, included in range */ + ip_set_ip_t last_ip; /* host byte order, included in range */ + u_int32_t flags; + u_int32_t size; /* size of the ipmap proper */ +}; + +struct ip_set_req_macipmap_create { + ip_set_ip_t from; + ip_set_ip_t to; + u_int32_t flags; +}; + +struct ip_set_req_macipmap { + ip_set_ip_t ip; + unsigned char ethernet[ETH_ALEN]; +}; + +struct ip_set_macip { + unsigned short match; + unsigned char ethernet[ETH_ALEN]; +}; + +#endif /* __IP_SET_MACIPMAP_H */ diff --git a/include/linux/netfilter_ipv4/ip_set_malloc.h b/include/linux/netfilter_ipv4/ip_set_malloc.h new file mode 100644 index 0000000..2a80443 --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_malloc.h @@ -0,0 +1,153 @@ +#ifndef _IP_SET_MALLOC_H +#define _IP_SET_MALLOC_H + +#ifdef __KERNEL__ +#include + +static size_t max_malloc_size = 0, max_page_size = 0; +static size_t default_max_malloc_size = 131072; /* Guaranteed: slab.c */ + +static inline int init_max_page_size(void) +{ +/* Compatibility glues to support 2.4.36 */ +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) +#define __GFP_NOWARN 0 + + /* Guaranteed: slab.c */ + max_malloc_size = max_page_size = default_max_malloc_size; +#else + size_t page_size = 0; + +#define CACHE(x) if (max_page_size == 0 || x < max_page_size) \ + page_size = x; +#include +#undef CACHE + if (page_size) { + if (max_malloc_size == 0) + max_malloc_size = page_size; + + max_page_size = page_size; + + return 1; + } +#endif + return 0; +} + +struct harray { + size_t max_elements; + void *arrays[0]; +}; + +static inline void * +__harray_malloc(size_t hashsize, size_t typesize, gfp_t flags) +{ + struct harray *harray; + size_t max_elements, size, i, j; + + BUG_ON(max_page_size == 0); + + if (typesize > max_page_size) + return NULL; + + max_elements = max_page_size/typesize; + size = hashsize/max_elements; + if (hashsize % max_elements) + size++; + + /* Last pointer signals end of arrays */ + harray = kmalloc(sizeof(struct harray) + (size + 1) * sizeof(void *), + flags); + + if (!harray) + return NULL; + + for (i = 0; i < size - 1; i++) { + harray->arrays[i] = kmalloc(max_elements * typesize, flags); + if (!harray->arrays[i]) + goto undo; + memset(harray->arrays[i], 0, max_elements * typesize); + } + harray->arrays[i] = kmalloc((hashsize - i * max_elements) * typesize, + flags); + if (!harray->arrays[i]) + goto undo; + memset(harray->arrays[i], 0, (hashsize - i * max_elements) * typesize); + + harray->max_elements = max_elements; + harray->arrays[size] = NULL; + + return (void *)harray; + + undo: + for (j = 0; j < i; j++) { + kfree(harray->arrays[j]); + } + kfree(harray); + return NULL; +} + +static inline void * +harray_malloc(size_t hashsize, size_t typesize, gfp_t flags) +{ + void *harray; + + do { + harray = __harray_malloc(hashsize, typesize, flags|__GFP_NOWARN); + } while (harray == NULL && init_max_page_size()); + + return harray; +} + +static inline void harray_free(void *h) +{ + struct harray *harray = (struct harray *) h; + size_t i; + + for (i = 0; harray->arrays[i] != NULL; i++) + kfree(harray->arrays[i]); + kfree(harray); +} + +static inline void harray_flush(void *h, size_t hashsize, size_t typesize) +{ + struct harray *harray = (struct harray *) h; + size_t i; + + for (i = 0; harray->arrays[i+1] != NULL; i++) + memset(harray->arrays[i], 0, harray->max_elements * typesize); + memset(harray->arrays[i], 0, + (hashsize - i * harray->max_elements) * typesize); +} + +#define HARRAY_ELEM(h, type, which) \ +({ \ + struct harray *__h = (struct harray *)(h); \ + ((type)((__h)->arrays[(which)/(__h)->max_elements]) \ + + (which)%(__h)->max_elements); \ +}) + +/* General memory allocation and deallocation */ +static inline void * ip_set_malloc(size_t bytes) +{ + BUG_ON(max_malloc_size == 0); + + if (bytes > default_max_malloc_size) + return vmalloc(bytes); + else + return kmalloc(bytes, GFP_KERNEL | __GFP_NOWARN); +} + +static inline void ip_set_free(void * data, size_t bytes) +{ + BUG_ON(max_malloc_size == 0); + + if (bytes > default_max_malloc_size) + vfree(data); + else + kfree(data); +} + +#endif /* __KERNEL__ */ + +#endif /*_IP_SET_MALLOC_H*/ diff --git a/include/linux/netfilter_ipv4/ip_set_nethash.h b/include/linux/netfilter_ipv4/ip_set_nethash.h new file mode 100644 index 0000000..cf0b794 --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_nethash.h @@ -0,0 +1,31 @@ +#ifndef __IP_SET_NETHASH_H +#define __IP_SET_NETHASH_H + +#include +#include + +#define SETTYPE_NAME "nethash" + +struct ip_set_nethash { + ip_set_ip_t *members; /* the nethash proper */ + uint32_t elements; /* number of elements */ + uint32_t hashsize; /* hash size */ + uint16_t probes; /* max number of probes */ + uint16_t resize; /* resize factor in percent */ + uint8_t cidr[30]; /* CIDR sizes */ + uint16_t nets[30]; /* nr of nets by CIDR sizes */ + initval_t initval[0]; /* initvals for jhash_1word */ +}; + +struct ip_set_req_nethash_create { + uint32_t hashsize; + uint16_t probes; + uint16_t resize; +}; + +struct ip_set_req_nethash { + ip_set_ip_t ip; + uint8_t cidr; +}; + +#endif /* __IP_SET_NETHASH_H */ diff --git a/include/linux/netfilter_ipv4/ip_set_portmap.h b/include/linux/netfilter_ipv4/ip_set_portmap.h new file mode 100644 index 0000000..37f411e --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_portmap.h @@ -0,0 +1,25 @@ +#ifndef __IP_SET_PORTMAP_H +#define __IP_SET_PORTMAP_H + +#include +#include + +#define SETTYPE_NAME "portmap" + +struct ip_set_portmap { + void *members; /* the portmap proper */ + ip_set_ip_t first_ip; /* host byte order, included in range */ + ip_set_ip_t last_ip; /* host byte order, included in range */ + u_int32_t size; /* size of the ipmap proper */ +}; + +struct ip_set_req_portmap_create { + ip_set_ip_t from; + ip_set_ip_t to; +}; + +struct ip_set_req_portmap { + ip_set_ip_t ip; +}; + +#endif /* __IP_SET_PORTMAP_H */ diff --git a/include/linux/netfilter_ipv4/ip_set_setlist.h b/include/linux/netfilter_ipv4/ip_set_setlist.h new file mode 100644 index 0000000..7cc6ed0 --- /dev/null +++ b/include/linux/netfilter_ipv4/ip_set_setlist.h @@ -0,0 +1,26 @@ +#ifndef __IP_SET_SETLIST_H +#define __IP_SET_SETLIST_H + +#include + +#define SETTYPE_NAME "setlist" + +#define IP_SET_SETLIST_ADD_AFTER 0 +#define IP_SET_SETLIST_ADD_BEFORE 1 + +struct ip_set_setlist { + uint8_t size; + ip_set_id_t index[0]; +}; + +struct ip_set_req_setlist_create { + uint8_t size; +}; + +struct ip_set_req_setlist { + char name[IP_SET_MAXNAMELEN]; + char ref[IP_SET_MAXNAMELEN]; + uint8_t before; +}; + +#endif /* __IP_SET_SETLIST_H */ diff --git a/include/linux/netfilter_ipv4/ipt_set.h b/include/linux/netfilter_ipv4/ipt_set.h new file mode 100644 index 0000000..2a18b93 --- /dev/null +++ b/include/linux/netfilter_ipv4/ipt_set.h @@ -0,0 +1,21 @@ +#ifndef _IPT_SET_H +#define _IPT_SET_H + +#include + +struct ipt_set_info { + ip_set_id_t index; + u_int32_t flags[IP_SET_MAX_BINDINGS + 1]; +}; + +/* match info */ +struct ipt_set_info_match { + struct ipt_set_info match_set; +}; + +struct ipt_set_info_target { + struct ipt_set_info add_set; + struct ipt_set_info del_set; +}; + +#endif /*_IPT_SET_H*/ diff --git a/net/ipv4/netfilter/Kconfig b/net/ipv4/netfilter/Kconfig index 1833bdb..4c315cf 100644 --- a/net/ipv4/netfilter/Kconfig +++ b/net/ipv4/netfilter/Kconfig @@ -388,5 +388,122 @@ config IP_NF_ARP_MANGLE endif # IP_NF_ARPTABLES +config IP_NF_SET + tristate "IP set support" + depends on INET && NETFILTER + help + This option adds IP set support to the kernel. + In order to define and use sets, you need the userspace utility + ipset(8). + + To compile it as a module, choose M here. If unsure, say N. + +config IP_NF_SET_MAX + int "Maximum number of IP sets" + default 256 + range 2 65534 + depends on IP_NF_SET + help + You can define here default value of the maximum number + of IP sets for the kernel. + + The value can be overriden by the 'max_sets' module + parameter of the 'ip_set' module. + +config IP_NF_SET_HASHSIZE + int "Hash size for bindings of IP sets" + default 1024 + depends on IP_NF_SET + help + You can define here default value of the hash size for + bindings of IP sets. + + The value can be overriden by the 'hash_size' module + parameter of the 'ip_set' module. + +config IP_NF_SET_IPMAP + tristate "ipmap set support" + depends on IP_NF_SET + help + This option adds the ipmap set type support. + + To compile it as a module, choose M here. If unsure, say N. + +config IP_NF_SET_MACIPMAP + tristate "macipmap set support" + depends on IP_NF_SET + help + This option adds the macipmap set type support. + + To compile it as a module, choose M here. If unsure, say N. + +config IP_NF_SET_PORTMAP + tristate "portmap set support" + depends on IP_NF_SET + help + This option adds the portmap set type support. + + To compile it as a module, choose M here. If unsure, say N. + +config IP_NF_SET_IPHASH + tristate "iphash set support" + depends on IP_NF_SET + help + This option adds the iphash set type support. + + To compile it as a module, choose M here. If unsure, say N. + +config IP_NF_SET_NETHASH + tristate "nethash set support" + depends on IP_NF_SET + help + This option adds the nethash set type support. + + To compile it as a module, choose M here. If unsure, say N. + +config IP_NF_SET_IPPORTHASH + tristate "ipporthash set support" + depends on IP_NF_SET + help + This option adds the ipporthash set type support. + + To compile it as a module, choose M here. If unsure, say N. + +config IP_NF_SET_IPTREE + tristate "iptree set support" + depends on IP_NF_SET + help + This option adds the iptree set type support. + + To compile it as a module, choose M here. If unsure, say N. + +config IP_NF_SET_IPTREEMAP + tristate "iptreemap set support" + depends on IP_NF_SET + help + This option adds the iptreemap set type support. + + To compile it as a module, choose M here. If unsure, say N. + +config IP_NF_MATCH_SET + tristate "set match support" + depends on IP_NF_SET + help + Set matching matches against given IP sets. + You need the ipset utility to create and set up the sets. + + To compile it as a module, choose M here. If unsure, say N. + +config IP_NF_TARGET_SET + tristate "SET target support" + depends on IP_NF_SET + help + The SET target makes possible to add/delete entries + in IP sets. + You need the ipset utility to create and set up the sets. + + To compile it as a module, choose M here. If unsure, say N. + + endmenu diff --git a/net/ipv4/netfilter/Makefile b/net/ipv4/netfilter/Makefile index 4811159..fd4913a 100644 --- a/net/ipv4/netfilter/Makefile +++ b/net/ipv4/netfilter/Makefile @@ -51,6 +51,7 @@ obj-$(CONFIG_IP_NF_SECURITY) += iptable_security.o obj-$(CONFIG_IP_NF_MATCH_ADDRTYPE) += ipt_addrtype.o obj-$(CONFIG_IP_NF_MATCH_AH) += ipt_ah.o obj-$(CONFIG_IP_NF_MATCH_ECN) += ipt_ecn.o +obj-$(CONFIG_IP_NF_MATCH_SET) += ipt_set.o # targets obj-$(CONFIG_IP_NF_TARGET_CLUSTERIP) += ipt_CLUSTERIP.o @@ -60,8 +61,20 @@ obj-$(CONFIG_IP_NF_TARGET_MASQUERADE) += ipt_MASQUERADE.o obj-$(CONFIG_IP_NF_TARGET_NETMAP) += ipt_NETMAP.o obj-$(CONFIG_IP_NF_TARGET_REDIRECT) += ipt_REDIRECT.o obj-$(CONFIG_IP_NF_TARGET_REJECT) += ipt_REJECT.o +obj-$(CONFIG_IP_NF_TARGET_SET) += ipt_SET.o obj-$(CONFIG_IP_NF_TARGET_ULOG) += ipt_ULOG.o +# sets +obj-$(CONFIG_IP_NF_SET) += ip_set.o +obj-$(CONFIG_IP_NF_SET_IPMAP) += ip_set_ipmap.o +obj-$(CONFIG_IP_NF_SET_PORTMAP) += ip_set_portmap.o +obj-$(CONFIG_IP_NF_SET_MACIPMAP) += ip_set_macipmap.o +obj-$(CONFIG_IP_NF_SET_IPHASH) += ip_set_iphash.o +obj-$(CONFIG_IP_NF_SET_NETHASH) += ip_set_nethash.o +obj-$(CONFIG_IP_NF_SET_IPPORTHASH) += ip_set_ipporthash.o +obj-$(CONFIG_IP_NF_SET_IPTREE) += ip_set_iptree.o +obj-$(CONFIG_IP_NF_SET_IPTREEMAP) += ip_set_iptreemap.o + # generic ARP tables obj-$(CONFIG_IP_NF_ARPTABLES) += arp_tables.o obj-$(CONFIG_IP_NF_ARP_MANGLE) += arpt_mangle.o diff --git a/net/ipv4/netfilter/ip_set.c b/net/ipv4/netfilter/ip_set.c new file mode 100644 index 0000000..7ecdc83 --- /dev/null +++ b/net/ipv4/netfilter/ip_set.c @@ -0,0 +1,1531 @@ +/* Copyright (C) 2000-2002 Joakim Axelsson + * Patrick Schaaf + * Copyright (C) 2003-2004 Jozsef Kadlecsik + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +/* Kernel module for IP set management */ + +#include +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) +#include +#endif +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,27) +#include +#else +#include +#endif +#include + +#define ASSERT_READ_LOCK(x) +#define ASSERT_WRITE_LOCK(x) +#include +#include + +static struct list_head set_type_list; /* all registered sets */ +static struct ip_set **ip_set_list; /* all individual sets */ +static DEFINE_RWLOCK(ip_set_lock); /* protects the lists and the hash */ +static struct semaphore ip_set_app_mutex; /* serializes user access */ +static ip_set_id_t ip_set_max = CONFIG_IP_NF_SET_MAX; +static int protocol_version = IP_SET_PROTOCOL_VERSION; + +#define STREQ(a,b) (strncmp(a,b,IP_SET_MAXNAMELEN) == 0) +#define DONT_ALIGN (protocol_version == IP_SET_PROTOCOL_UNALIGNED) +#define ALIGNED(len) IPSET_VALIGN(len, DONT_ALIGN) + +/* + * Sets are identified either by the index in ip_set_list or by id. + * The id never changes. The index may change by swapping and used + * by external references (set/SET netfilter modules, etc.) + * + * Userspace requests are serialized by ip_set_mutex and sets can + * be deleted only from userspace. Therefore ip_set_list locking + * must obey the following rules: + * + * - kernel requests: read and write locking mandatory + * - user requests: read locking optional, write locking mandatory + */ + +static inline void +__ip_set_get(ip_set_id_t index) +{ + atomic_inc(&ip_set_list[index]->ref); +} + +static inline void +__ip_set_put(ip_set_id_t index) +{ + atomic_dec(&ip_set_list[index]->ref); +} + +/* Add, del and test set entries from kernel */ + +int +ip_set_testip_kernel(ip_set_id_t index, + const struct sk_buff *skb, + const u_int32_t *flags) +{ + struct ip_set *set; + int res; + + read_lock_bh(&ip_set_lock); + set = ip_set_list[index]; + IP_SET_ASSERT(set); + DP("set %s, index %u", set->name, index); + + read_lock_bh(&set->lock); + res = set->type->testip_kernel(set, skb, flags); + read_unlock_bh(&set->lock); + + read_unlock_bh(&ip_set_lock); + + return (res < 0 ? 0 : res); +} + +int +ip_set_addip_kernel(ip_set_id_t index, + const struct sk_buff *skb, + const u_int32_t *flags) +{ + struct ip_set *set; + int res; + + retry: + read_lock_bh(&ip_set_lock); + set = ip_set_list[index]; + IP_SET_ASSERT(set); + DP("set %s, index %u", set->name, index); + + write_lock_bh(&set->lock); + res = set->type->addip_kernel(set, skb, flags); + write_unlock_bh(&set->lock); + + read_unlock_bh(&ip_set_lock); + /* Retry function called without holding any lock */ + if (res == -EAGAIN + && set->type->retry + && (res = set->type->retry(set)) == 0) + goto retry; + + return res; +} + +int +ip_set_delip_kernel(ip_set_id_t index, + const struct sk_buff *skb, + const u_int32_t *flags) +{ + struct ip_set *set; + int res; + + read_lock_bh(&ip_set_lock); + set = ip_set_list[index]; + IP_SET_ASSERT(set); + DP("set %s, index %u", set->name, index); + + write_lock_bh(&set->lock); + res = set->type->delip_kernel(set, skb, flags); + write_unlock_bh(&set->lock); + + read_unlock_bh(&ip_set_lock); + + return res; +} + +/* Register and deregister settype */ + +static inline struct ip_set_type * +find_set_type(const char *name) +{ + struct ip_set_type *set_type; + + list_for_each_entry(set_type, &set_type_list, list) + if (STREQ(set_type->typename, name)) + return set_type; + return NULL; +} + +int +ip_set_register_set_type(struct ip_set_type *set_type) +{ + int ret = 0; + + if (set_type->protocol_version != IP_SET_PROTOCOL_VERSION) { + ip_set_printk("'%s' uses wrong protocol version %u (want %u)", + set_type->typename, + set_type->protocol_version, + IP_SET_PROTOCOL_VERSION); + return -EINVAL; + } + + write_lock_bh(&ip_set_lock); + if (find_set_type(set_type->typename)) { + /* Duplicate! */ + ip_set_printk("'%s' already registered!", + set_type->typename); + ret = -EINVAL; + goto unlock; + } + if (!try_module_get(THIS_MODULE)) { + ret = -EFAULT; + goto unlock; + } + list_add(&set_type->list, &set_type_list); + DP("'%s' registered.", set_type->typename); + unlock: + write_unlock_bh(&ip_set_lock); + return ret; +} + +void +ip_set_unregister_set_type(struct ip_set_type *set_type) +{ + write_lock_bh(&ip_set_lock); + if (!find_set_type(set_type->typename)) { + ip_set_printk("'%s' not registered?", + set_type->typename); + goto unlock; + } + list_del(&set_type->list); + module_put(THIS_MODULE); + DP("'%s' unregistered.", set_type->typename); + unlock: + write_unlock_bh(&ip_set_lock); + +} + +ip_set_id_t +__ip_set_get_byname(const char *name, struct ip_set **set) +{ + ip_set_id_t i, index = IP_SET_INVALID_ID; + + for (i = 0; i < ip_set_max; i++) { + if (ip_set_list[i] != NULL + && STREQ(ip_set_list[i]->name, name)) { + __ip_set_get(i); + index = i; + *set = ip_set_list[i]; + break; + } + } + return index; +} + +void +__ip_set_put_byindex(ip_set_id_t index) +{ + if (ip_set_list[index]) + __ip_set_put(index); +} + +/* + * Userspace routines + */ + +/* + * Find set by name, reference it once. The reference makes sure the + * thing pointed to, does not go away under our feet. Drop the reference + * later, using ip_set_put(). + */ +ip_set_id_t +ip_set_get_byname(const char *name) +{ + ip_set_id_t i, index = IP_SET_INVALID_ID; + + down(&ip_set_app_mutex); + for (i = 0; i < ip_set_max; i++) { + if (ip_set_list[i] != NULL + && STREQ(ip_set_list[i]->name, name)) { + __ip_set_get(i); + index = i; + break; + } + } + up(&ip_set_app_mutex); + return index; +} + +/* + * Find set by index, reference it once. The reference makes sure the + * thing pointed to, does not go away under our feet. Drop the reference + * later, using ip_set_put(). + */ +ip_set_id_t +ip_set_get_byindex(ip_set_id_t index) +{ + down(&ip_set_app_mutex); + + if (index >= ip_set_max) + return IP_SET_INVALID_ID; + + if (ip_set_list[index]) + __ip_set_get(index); + else + index = IP_SET_INVALID_ID; + + up(&ip_set_app_mutex); + return index; +} + +/* + * Find the set id belonging to the index. + * We are protected by the mutex, so we do not need to use + * ip_set_lock. There is no need to reference the sets either. + */ +ip_set_id_t +ip_set_id(ip_set_id_t index) +{ + if (index >= ip_set_max || !ip_set_list[index]) + return IP_SET_INVALID_ID; + + return ip_set_list[index]->id; +} + +/* + * If the given set pointer points to a valid set, decrement + * reference count by 1. The caller shall not assume the index + * to be valid, after calling this function. + */ +void +ip_set_put_byindex(ip_set_id_t index) +{ + down(&ip_set_app_mutex); + if (ip_set_list[index]) + __ip_set_put(index); + up(&ip_set_app_mutex); +} + +/* Find a set by name or index */ +static ip_set_id_t +ip_set_find_byname(const char *name) +{ + ip_set_id_t i, index = IP_SET_INVALID_ID; + + for (i = 0; i < ip_set_max; i++) { + if (ip_set_list[i] != NULL + && STREQ(ip_set_list[i]->name, name)) { + index = i; + break; + } + } + return index; +} + +static ip_set_id_t +ip_set_find_byindex(ip_set_id_t index) +{ + if (index >= ip_set_max || ip_set_list[index] == NULL) + index = IP_SET_INVALID_ID; + + return index; +} + +/* + * Add, del and test + */ + +static int +ip_set_addip(struct ip_set *set, const void *data, u_int32_t size) +{ + int res; + + IP_SET_ASSERT(set); + do { + write_lock_bh(&set->lock); + res = set->type->addip(set, data, size); + write_unlock_bh(&set->lock); + } while (res == -EAGAIN + && set->type->retry + && (res = set->type->retry(set)) == 0); + + return res; +} + +static int +ip_set_delip(struct ip_set *set, const void *data, u_int32_t size) +{ + int res; + + IP_SET_ASSERT(set); + + write_lock_bh(&set->lock); + res = set->type->delip(set, data, size); + write_unlock_bh(&set->lock); + + return res; +} + +static int +ip_set_testip(struct ip_set *set, const void *data, u_int32_t size) +{ + int res; + + IP_SET_ASSERT(set); + + read_lock_bh(&set->lock); + res = set->type->testip(set, data, size); + read_unlock_bh(&set->lock); + + return (res > 0 ? -EEXIST : res); +} + +static struct ip_set_type * +find_set_type_rlock(const char *typename) +{ + struct ip_set_type *type; + + read_lock_bh(&ip_set_lock); + type = find_set_type(typename); + if (type == NULL) + read_unlock_bh(&ip_set_lock); + + return type; +} + +static int +find_free_id(const char *name, + ip_set_id_t *index, + ip_set_id_t *id) +{ + ip_set_id_t i; + + *id = IP_SET_INVALID_ID; + for (i = 0; i < ip_set_max; i++) { + if (ip_set_list[i] == NULL) { + if (*id == IP_SET_INVALID_ID) + *id = *index = i; + } else if (STREQ(name, ip_set_list[i]->name)) + /* Name clash */ + return -EEXIST; + } + if (*id == IP_SET_INVALID_ID) + /* No free slot remained */ + return -ERANGE; + /* Check that index is usable as id (swapping) */ + check: + for (i = 0; i < ip_set_max; i++) { + if (ip_set_list[i] != NULL + && ip_set_list[i]->id == *id) { + *id = i; + goto check; + } + } + return 0; +} + +/* + * Create a set + */ +static int +ip_set_create(const char *name, + const char *typename, + ip_set_id_t restore, + const void *data, + u_int32_t size) +{ + struct ip_set *set; + ip_set_id_t index = 0, id; + int res = 0; + + DP("setname: %s, typename: %s, id: %u", name, typename, restore); + + /* + * First, and without any locks, allocate and initialize + * a normal base set structure. + */ + set = kmalloc(sizeof(struct ip_set), GFP_KERNEL); + if (!set) + return -ENOMEM; + rwlock_init(&set->lock); + strncpy(set->name, name, IP_SET_MAXNAMELEN); + atomic_set(&set->ref, 0); + + /* + * Next, take the &ip_set_lock, check that we know the type, + * and take a reference on the type, to make sure it + * stays available while constructing our new set. + * + * After referencing the type, we drop the &ip_set_lock, + * and let the new set construction run without locks. + */ + set->type = find_set_type_rlock(typename); + if (set->type == NULL) { + /* Try loading the module */ + char modulename[IP_SET_MAXNAMELEN + strlen("ip_set_") + 1]; + strcpy(modulename, "ip_set_"); + strcat(modulename, typename); + DP("try to load %s", modulename); + request_module(modulename); + set->type = find_set_type_rlock(typename); + } + if (set->type == NULL) { + ip_set_printk("no set type '%s', set '%s' not created", + typename, name); + res = -ENOENT; + goto out; + } + if (!try_module_get(set->type->me)) { + read_unlock_bh(&ip_set_lock); + res = -EFAULT; + goto out; + } + read_unlock_bh(&ip_set_lock); + + /* Check request size */ + if (size != set->type->header_size) { + ip_set_printk("data length wrong (want %lu, have %lu)", + (long unsigned)set->type->header_size, + (long unsigned)size); + goto put_out; + } + + /* + * Without holding any locks, create private part. + */ + res = set->type->create(set, data, size); + if (res != 0) + goto put_out; + + /* BTW, res==0 here. */ + + /* + * Here, we have a valid, constructed set. &ip_set_lock again, + * find free id/index and check that it is not already in + * ip_set_list. + */ + write_lock_bh(&ip_set_lock); + if ((res = find_free_id(set->name, &index, &id)) != 0) { + DP("no free id!"); + goto cleanup; + } + + /* Make sure restore gets the same index */ + if (restore != IP_SET_INVALID_ID && index != restore) { + DP("Can't restore, sets are screwed up"); + res = -ERANGE; + goto cleanup; + } + + /* + * Finally! Add our shiny new set to the list, and be done. + */ + DP("create: '%s' created with index %u, id %u!", set->name, index, id); + set->id = id; + ip_set_list[index] = set; + write_unlock_bh(&ip_set_lock); + return res; + + cleanup: + write_unlock_bh(&ip_set_lock); + set->type->destroy(set); + put_out: + module_put(set->type->me); + out: + kfree(set); + return res; +} + +/* + * Destroy a given existing set + */ +static void +ip_set_destroy_set(ip_set_id_t index) +{ + struct ip_set *set = ip_set_list[index]; + + IP_SET_ASSERT(set); + DP("set: %s", set->name); + write_lock_bh(&ip_set_lock); + ip_set_list[index] = NULL; + write_unlock_bh(&ip_set_lock); + + /* Must call it without holding any lock */ + set->type->destroy(set); + module_put(set->type->me); + kfree(set); +} + +/* + * Destroy a set - or all sets + * Sets must not be referenced/used. + */ +static int +ip_set_destroy(ip_set_id_t index) +{ + ip_set_id_t i; + + /* ref modification always protected by the mutex */ + if (index != IP_SET_INVALID_ID) { + if (atomic_read(&ip_set_list[index]->ref)) + return -EBUSY; + ip_set_destroy_set(index); + } else { + for (i = 0; i < ip_set_max; i++) { + if (ip_set_list[i] != NULL + && (atomic_read(&ip_set_list[i]->ref))) + return -EBUSY; + } + + for (i = 0; i < ip_set_max; i++) { + if (ip_set_list[i] != NULL) + ip_set_destroy_set(i); + } + } + return 0; +} + +static void +ip_set_flush_set(struct ip_set *set) +{ + DP("set: %s %u", set->name, set->id); + + write_lock_bh(&set->lock); + set->type->flush(set); + write_unlock_bh(&set->lock); +} + +/* + * Flush data in a set - or in all sets + */ +static int +ip_set_flush(ip_set_id_t index) +{ + if (index != IP_SET_INVALID_ID) { + IP_SET_ASSERT(ip_set_list[index]); + ip_set_flush_set(ip_set_list[index]); + } else { + ip_set_id_t i; + + for (i = 0; i < ip_set_max; i++) + if (ip_set_list[i] != NULL) + ip_set_flush_set(ip_set_list[i]); + } + + return 0; +} + +/* Rename a set */ +static int +ip_set_rename(ip_set_id_t index, const char *name) +{ + struct ip_set *set = ip_set_list[index]; + ip_set_id_t i; + int res = 0; + + DP("set: %s to %s", set->name, name); + write_lock_bh(&ip_set_lock); + for (i = 0; i < ip_set_max; i++) { + if (ip_set_list[i] != NULL + && STREQ(ip_set_list[i]->name, name)) { + res = -EEXIST; + goto unlock; + } + } + strncpy(set->name, name, IP_SET_MAXNAMELEN); + unlock: + write_unlock_bh(&ip_set_lock); + return res; +} + +/* + * Swap two sets so that name/index points to the other. + * References are also swapped. + */ +static int +ip_set_swap(ip_set_id_t from_index, ip_set_id_t to_index) +{ + struct ip_set *from = ip_set_list[from_index]; + struct ip_set *to = ip_set_list[to_index]; + char from_name[IP_SET_MAXNAMELEN]; + u_int32_t from_ref; + + DP("set: %s to %s", from->name, to->name); + /* Features must not change. + * Not an artifical restriction anymore, as we must prevent + * possible loops created by swapping in setlist type of sets. */ + if (from->type->features != to->type->features) + return -ENOEXEC; + + /* No magic here: ref munging protected by the mutex */ + write_lock_bh(&ip_set_lock); + strncpy(from_name, from->name, IP_SET_MAXNAMELEN); + from_ref = atomic_read(&from->ref); + + strncpy(from->name, to->name, IP_SET_MAXNAMELEN); + atomic_set(&from->ref, atomic_read(&to->ref)); + strncpy(to->name, from_name, IP_SET_MAXNAMELEN); + atomic_set(&to->ref, from_ref); + + ip_set_list[from_index] = to; + ip_set_list[to_index] = from; + + write_unlock_bh(&ip_set_lock); + return 0; +} + +/* + * List set data + */ + +static int +ip_set_list_set(ip_set_id_t index, void *data, int *used, int len) +{ + struct ip_set *set = ip_set_list[index]; + struct ip_set_list *set_list; + + /* Pointer to our header */ + set_list = data + *used; + + DP("set: %s, used: %d len %u %p %p", set->name, *used, len, data, data + *used); + + /* Get and ensure header size */ + if (*used + ALIGNED(sizeof(struct ip_set_list)) > len) + goto not_enough_mem; + *used += ALIGNED(sizeof(struct ip_set_list)); + + read_lock_bh(&set->lock); + /* Get and ensure set specific header size */ + set_list->header_size = ALIGNED(set->type->header_size); + if (*used + set_list->header_size > len) + goto unlock_set; + + /* Fill in the header */ + set_list->index = index; + set_list->binding = IP_SET_INVALID_ID; + set_list->ref = atomic_read(&set->ref); + + /* Fill in set spefific header data */ + set->type->list_header(set, data + *used); + *used += set_list->header_size; + + /* Get and ensure set specific members size */ + set_list->members_size = set->type->list_members_size(set, DONT_ALIGN); + if (*used + set_list->members_size > len) + goto unlock_set; + + /* Fill in set spefific members data */ + set->type->list_members(set, data + *used, DONT_ALIGN); + *used += set_list->members_size; + read_unlock_bh(&set->lock); + + /* Bindings */ + set_list->bindings_size = 0; + + return 0; + + unlock_set: + read_unlock_bh(&set->lock); + not_enough_mem: + DP("not enough mem, try again"); + return -EAGAIN; +} + +/* + * Save sets + */ +static inline int +ip_set_save_marker(void *data, int *used, int len) +{ + struct ip_set_save *set_save; + + DP("used %u, len %u", *used, len); + /* Get and ensure header size */ + if (*used + ALIGNED(sizeof(struct ip_set_save)) > len) + return -ENOMEM; + + /* Marker: just for backward compatibility */ + set_save = data + *used; + set_save->index = IP_SET_INVALID_ID; + set_save->header_size = 0; + set_save->members_size = 0; + *used += ALIGNED(sizeof(struct ip_set_save)); + + return 0; +} + +static int +ip_set_save_set(ip_set_id_t index, void *data, int *used, int len) +{ + struct ip_set *set; + struct ip_set_save *set_save; + + /* Pointer to our header */ + set_save = data + *used; + + /* Get and ensure header size */ + if (*used + ALIGNED(sizeof(struct ip_set_save)) > len) + goto not_enough_mem; + *used += ALIGNED(sizeof(struct ip_set_save)); + + set = ip_set_list[index]; + DP("set: %s, used: %d(%d) %p %p", set->name, *used, len, + data, data + *used); + + read_lock_bh(&set->lock); + /* Get and ensure set specific header size */ + set_save->header_size = ALIGNED(set->type->header_size); + if (*used + set_save->header_size > len) + goto unlock_set; + + /* Fill in the header */ + set_save->index = index; + set_save->binding = IP_SET_INVALID_ID; + + /* Fill in set spefific header data */ + set->type->list_header(set, data + *used); + *used += set_save->header_size; + + DP("set header filled: %s, used: %d(%lu) %p %p", set->name, *used, + (unsigned long)set_save->header_size, data, data + *used); + /* Get and ensure set specific members size */ + set_save->members_size = set->type->list_members_size(set, DONT_ALIGN); + if (*used + set_save->members_size > len) + goto unlock_set; + + /* Fill in set spefific members data */ + set->type->list_members(set, data + *used, DONT_ALIGN); + *used += set_save->members_size; + read_unlock_bh(&set->lock); + DP("set members filled: %s, used: %d(%lu) %p %p", set->name, *used, + (unsigned long)set_save->members_size, data, data + *used); + return 0; + + unlock_set: + read_unlock_bh(&set->lock); + not_enough_mem: + DP("not enough mem, try again"); + return -EAGAIN; +} + +/* + * Restore sets + */ +static int +ip_set_restore(void *data, int len) +{ + int res = 0; + int line = 0, used = 0, members_size; + struct ip_set *set; + struct ip_set_restore *set_restore; + ip_set_id_t index; + + /* Loop to restore sets */ + while (1) { + line++; + + DP("%d %zu %d", used, ALIGNED(sizeof(struct ip_set_restore)), len); + /* Get and ensure header size */ + if (used + ALIGNED(sizeof(struct ip_set_restore)) > len) + return line; + set_restore = data + used; + used += ALIGNED(sizeof(struct ip_set_restore)); + + /* Ensure data size */ + if (used + + set_restore->header_size + + set_restore->members_size > len) + return line; + + /* Check marker */ + if (set_restore->index == IP_SET_INVALID_ID) { + line--; + goto finish; + } + + /* Try to create the set */ + DP("restore %s %s", set_restore->name, set_restore->typename); + res = ip_set_create(set_restore->name, + set_restore->typename, + set_restore->index, + data + used, + set_restore->header_size); + + if (res != 0) + return line; + used += ALIGNED(set_restore->header_size); + + index = ip_set_find_byindex(set_restore->index); + DP("index %u, restore_index %u", index, set_restore->index); + if (index != set_restore->index) + return line; + /* Try to restore members data */ + set = ip_set_list[index]; + members_size = 0; + DP("members_size %lu reqsize %lu", + (unsigned long)set_restore->members_size, + (unsigned long)set->type->reqsize); + while (members_size + ALIGNED(set->type->reqsize) <= + set_restore->members_size) { + line++; + DP("members: %d, line %d", members_size, line); + res = ip_set_addip(set, + data + used + members_size, + set->type->reqsize); + if (!(res == 0 || res == -EEXIST)) + return line; + members_size += ALIGNED(set->type->reqsize); + } + + DP("members_size %lu %d", + (unsigned long)set_restore->members_size, members_size); + if (members_size != set_restore->members_size) + return line++; + used += set_restore->members_size; + } + + finish: + if (used != len) + return line; + + return 0; +} + +static int +ip_set_sockfn_set(struct sock *sk, int optval, void *user, unsigned int len) +{ + void *data; + int res = 0; /* Assume OK */ + size_t offset; + unsigned *op; + struct ip_set_req_adt *req_adt; + ip_set_id_t index = IP_SET_INVALID_ID; + int (*adtfn)(struct ip_set *set, + const void *data, u_int32_t size); + struct fn_table { + int (*fn)(struct ip_set *set, + const void *data, u_int32_t size); + } adtfn_table[] = + { { ip_set_addip }, { ip_set_delip }, { ip_set_testip}, + }; + + DP("optval=%d, user=%p, len=%d", optval, user, len); + if (!capable(CAP_NET_ADMIN)) + return -EPERM; + if (optval != SO_IP_SET) + return -EBADF; + if (len <= sizeof(unsigned)) { + ip_set_printk("short userdata (want >%zu, got %u)", + sizeof(unsigned), len); + return -EINVAL; + } + data = vmalloc(len); + if (!data) { + DP("out of mem for %u bytes", len); + return -ENOMEM; + } + if (copy_from_user(data, user, len) != 0) { + res = -EFAULT; + goto cleanup; + } + if (down_interruptible(&ip_set_app_mutex)) { + res = -EINTR; + goto cleanup; + } + + op = (unsigned *)data; + DP("op=%x", *op); + + if (*op < IP_SET_OP_VERSION) { + /* Check the version at the beginning of operations */ + struct ip_set_req_version *req_version = data; + if (!(req_version->version == IP_SET_PROTOCOL_UNALIGNED + || req_version->version == IP_SET_PROTOCOL_VERSION)) { + res = -EPROTO; + goto done; + } + protocol_version = req_version->version; + } + + switch (*op) { + case IP_SET_OP_CREATE:{ + struct ip_set_req_create *req_create = data; + offset = ALIGNED(sizeof(struct ip_set_req_create)); + + if (len < offset) { + ip_set_printk("short CREATE data (want >=%zu, got %u)", + offset, len); + res = -EINVAL; + goto done; + } + req_create->name[IP_SET_MAXNAMELEN - 1] = '\0'; + req_create->typename[IP_SET_MAXNAMELEN - 1] = '\0'; + res = ip_set_create(req_create->name, + req_create->typename, + IP_SET_INVALID_ID, + data + offset, + len - offset); + goto done; + } + case IP_SET_OP_DESTROY:{ + struct ip_set_req_std *req_destroy = data; + + if (len != sizeof(struct ip_set_req_std)) { + ip_set_printk("invalid DESTROY data (want %zu, got %u)", + sizeof(struct ip_set_req_std), len); + res = -EINVAL; + goto done; + } + if (STREQ(req_destroy->name, IPSET_TOKEN_ALL)) { + /* Destroy all sets */ + index = IP_SET_INVALID_ID; + } else { + req_destroy->name[IP_SET_MAXNAMELEN - 1] = '\0'; + index = ip_set_find_byname(req_destroy->name); + + if (index == IP_SET_INVALID_ID) { + res = -ENOENT; + goto done; + } + } + + res = ip_set_destroy(index); + goto done; + } + case IP_SET_OP_FLUSH:{ + struct ip_set_req_std *req_flush = data; + + if (len != sizeof(struct ip_set_req_std)) { + ip_set_printk("invalid FLUSH data (want %zu, got %u)", + sizeof(struct ip_set_req_std), len); + res = -EINVAL; + goto done; + } + if (STREQ(req_flush->name, IPSET_TOKEN_ALL)) { + /* Flush all sets */ + index = IP_SET_INVALID_ID; + } else { + req_flush->name[IP_SET_MAXNAMELEN - 1] = '\0'; + index = ip_set_find_byname(req_flush->name); + + if (index == IP_SET_INVALID_ID) { + res = -ENOENT; + goto done; + } + } + res = ip_set_flush(index); + goto done; + } + case IP_SET_OP_RENAME:{ + struct ip_set_req_create *req_rename = data; + + if (len != sizeof(struct ip_set_req_create)) { + ip_set_printk("invalid RENAME data (want %zu, got %u)", + sizeof(struct ip_set_req_create), len); + res = -EINVAL; + goto done; + } + + req_rename->name[IP_SET_MAXNAMELEN - 1] = '\0'; + req_rename->typename[IP_SET_MAXNAMELEN - 1] = '\0'; + + index = ip_set_find_byname(req_rename->name); + if (index == IP_SET_INVALID_ID) { + res = -ENOENT; + goto done; + } + res = ip_set_rename(index, req_rename->typename); + goto done; + } + case IP_SET_OP_SWAP:{ + struct ip_set_req_create *req_swap = data; + ip_set_id_t to_index; + + if (len != sizeof(struct ip_set_req_create)) { + ip_set_printk("invalid SWAP data (want %zu, got %u)", + sizeof(struct ip_set_req_create), len); + res = -EINVAL; + goto done; + } + + req_swap->name[IP_SET_MAXNAMELEN - 1] = '\0'; + req_swap->typename[IP_SET_MAXNAMELEN - 1] = '\0'; + + index = ip_set_find_byname(req_swap->name); + if (index == IP_SET_INVALID_ID) { + res = -ENOENT; + goto done; + } + to_index = ip_set_find_byname(req_swap->typename); + if (to_index == IP_SET_INVALID_ID) { + res = -ENOENT; + goto done; + } + res = ip_set_swap(index, to_index); + goto done; + } + default: + break; /* Set identified by id */ + } + + /* There we may have add/del/test/bind/unbind/test_bind operations */ + if (*op < IP_SET_OP_ADD_IP || *op > IP_SET_OP_TEST_IP) { + res = -EBADMSG; + goto done; + } + adtfn = adtfn_table[*op - IP_SET_OP_ADD_IP].fn; + + if (len < ALIGNED(sizeof(struct ip_set_req_adt))) { + ip_set_printk("short data in adt request (want >=%zu, got %u)", + ALIGNED(sizeof(struct ip_set_req_adt)), len); + res = -EINVAL; + goto done; + } + req_adt = data; + + index = ip_set_find_byindex(req_adt->index); + if (index == IP_SET_INVALID_ID) { + res = -ENOENT; + goto done; + } + do { + struct ip_set *set = ip_set_list[index]; + size_t offset = ALIGNED(sizeof(struct ip_set_req_adt)); + + IP_SET_ASSERT(set); + + if (len - offset != set->type->reqsize) { + ip_set_printk("data length wrong (want %lu, have %zu)", + (long unsigned)set->type->reqsize, + len - offset); + res = -EINVAL; + goto done; + } + res = adtfn(set, data + offset, len - offset); + } while (0); + + done: + up(&ip_set_app_mutex); + cleanup: + vfree(data); + if (res > 0) + res = 0; + DP("final result %d", res); + return res; +} + +static int +ip_set_sockfn_get(struct sock *sk, int optval, void *user, int *len) +{ + int res = 0; + unsigned *op; + ip_set_id_t index = IP_SET_INVALID_ID; + void *data; + int copylen = *len; + + DP("optval=%d, user=%p, len=%d", optval, user, *len); + if (!capable(CAP_NET_ADMIN)) + return -EPERM; + if (optval != SO_IP_SET) + return -EBADF; + if (*len < sizeof(unsigned)) { + ip_set_printk("short userdata (want >=%zu, got %d)", + sizeof(unsigned), *len); + return -EINVAL; + } + data = vmalloc(*len); + if (!data) { + DP("out of mem for %d bytes", *len); + return -ENOMEM; + } + if (copy_from_user(data, user, *len) != 0) { + res = -EFAULT; + goto cleanup; + } + if (down_interruptible(&ip_set_app_mutex)) { + res = -EINTR; + goto cleanup; + } + + op = (unsigned *) data; + DP("op=%x", *op); + + if (*op < IP_SET_OP_VERSION) { + /* Check the version at the beginning of operations */ + struct ip_set_req_version *req_version = data; + if (!(req_version->version == IP_SET_PROTOCOL_UNALIGNED + || req_version->version == IP_SET_PROTOCOL_VERSION)) { + res = -EPROTO; + goto done; + } + protocol_version = req_version->version; + } + + switch (*op) { + case IP_SET_OP_VERSION: { + struct ip_set_req_version *req_version = data; + + if (*len != sizeof(struct ip_set_req_version)) { + ip_set_printk("invalid VERSION (want %zu, got %d)", + sizeof(struct ip_set_req_version), + *len); + res = -EINVAL; + goto done; + } + + req_version->version = IP_SET_PROTOCOL_VERSION; + res = copy_to_user(user, req_version, + sizeof(struct ip_set_req_version)); + goto done; + } + case IP_SET_OP_GET_BYNAME: { + struct ip_set_req_get_set *req_get = data; + + if (*len != sizeof(struct ip_set_req_get_set)) { + ip_set_printk("invalid GET_BYNAME (want %zu, got %d)", + sizeof(struct ip_set_req_get_set), *len); + res = -EINVAL; + goto done; + } + req_get->set.name[IP_SET_MAXNAMELEN - 1] = '\0'; + index = ip_set_find_byname(req_get->set.name); + req_get->set.index = index; + goto copy; + } + case IP_SET_OP_GET_BYINDEX: { + struct ip_set_req_get_set *req_get = data; + + if (*len != sizeof(struct ip_set_req_get_set)) { + ip_set_printk("invalid GET_BYINDEX (want %zu, got %d)", + sizeof(struct ip_set_req_get_set), *len); + res = -EINVAL; + goto done; + } + req_get->set.name[IP_SET_MAXNAMELEN - 1] = '\0'; + index = ip_set_find_byindex(req_get->set.index); + strncpy(req_get->set.name, + index == IP_SET_INVALID_ID ? "" + : ip_set_list[index]->name, IP_SET_MAXNAMELEN); + goto copy; + } + case IP_SET_OP_ADT_GET: { + struct ip_set_req_adt_get *req_get = data; + + if (*len != sizeof(struct ip_set_req_adt_get)) { + ip_set_printk("invalid ADT_GET (want %zu, got %d)", + sizeof(struct ip_set_req_adt_get), *len); + res = -EINVAL; + goto done; + } + req_get->set.name[IP_SET_MAXNAMELEN - 1] = '\0'; + index = ip_set_find_byname(req_get->set.name); + if (index != IP_SET_INVALID_ID) { + req_get->set.index = index; + strncpy(req_get->typename, + ip_set_list[index]->type->typename, + IP_SET_MAXNAMELEN - 1); + } else { + res = -ENOENT; + goto done; + } + goto copy; + } + case IP_SET_OP_MAX_SETS: { + struct ip_set_req_max_sets *req_max_sets = data; + ip_set_id_t i; + + if (*len != sizeof(struct ip_set_req_max_sets)) { + ip_set_printk("invalid MAX_SETS (want %zu, got %d)", + sizeof(struct ip_set_req_max_sets), *len); + res = -EINVAL; + goto done; + } + + if (STREQ(req_max_sets->set.name, IPSET_TOKEN_ALL)) { + req_max_sets->set.index = IP_SET_INVALID_ID; + } else { + req_max_sets->set.name[IP_SET_MAXNAMELEN - 1] = '\0'; + req_max_sets->set.index = + ip_set_find_byname(req_max_sets->set.name); + if (req_max_sets->set.index == IP_SET_INVALID_ID) { + res = -ENOENT; + goto done; + } + } + req_max_sets->max_sets = ip_set_max; + req_max_sets->sets = 0; + for (i = 0; i < ip_set_max; i++) { + if (ip_set_list[i] != NULL) + req_max_sets->sets++; + } + goto copy; + } + case IP_SET_OP_LIST_SIZE: + case IP_SET_OP_SAVE_SIZE: { + struct ip_set_req_setnames *req_setnames = data; + struct ip_set_name_list *name_list; + struct ip_set *set; + ip_set_id_t i; + int used; + + if (*len < ALIGNED(sizeof(struct ip_set_req_setnames))) { + ip_set_printk("short LIST_SIZE (want >=%zu, got %d)", + ALIGNED(sizeof(struct ip_set_req_setnames)), + *len); + res = -EINVAL; + goto done; + } + + req_setnames->size = 0; + used = ALIGNED(sizeof(struct ip_set_req_setnames)); + for (i = 0; i < ip_set_max; i++) { + if (ip_set_list[i] == NULL) + continue; + name_list = data + used; + used += ALIGNED(sizeof(struct ip_set_name_list)); + if (used > copylen) { + res = -EAGAIN; + goto done; + } + set = ip_set_list[i]; + /* Fill in index, name, etc. */ + name_list->index = i; + name_list->id = set->id; + strncpy(name_list->name, + set->name, + IP_SET_MAXNAMELEN - 1); + strncpy(name_list->typename, + set->type->typename, + IP_SET_MAXNAMELEN - 1); + DP("filled %s of type %s, index %u\n", + name_list->name, name_list->typename, + name_list->index); + if (!(req_setnames->index == IP_SET_INVALID_ID + || req_setnames->index == i)) + continue; + /* Update size */ + req_setnames->size += + (*op == IP_SET_OP_LIST_SIZE ? + ALIGNED(sizeof(struct ip_set_list)) : + ALIGNED(sizeof(struct ip_set_save))) + + ALIGNED(set->type->header_size) + + set->type->list_members_size(set, DONT_ALIGN); + } + if (copylen != used) { + res = -EAGAIN; + goto done; + } + goto copy; + } + case IP_SET_OP_LIST: { + struct ip_set_req_list *req_list = data; + ip_set_id_t i; + int used; + + if (*len < sizeof(struct ip_set_req_list)) { + ip_set_printk("short LIST (want >=%zu, got %d)", + sizeof(struct ip_set_req_list), *len); + res = -EINVAL; + goto done; + } + index = req_list->index; + if (index != IP_SET_INVALID_ID + && ip_set_find_byindex(index) != index) { + res = -ENOENT; + goto done; + } + used = 0; + if (index == IP_SET_INVALID_ID) { + /* List all sets */ + for (i = 0; i < ip_set_max && res == 0; i++) { + if (ip_set_list[i] != NULL) + res = ip_set_list_set(i, data, &used, *len); + } + } else { + /* List an individual set */ + res = ip_set_list_set(index, data, &used, *len); + } + if (res != 0) + goto done; + else if (copylen != used) { + res = -EAGAIN; + goto done; + } + goto copy; + } + case IP_SET_OP_SAVE: { + struct ip_set_req_list *req_save = data; + ip_set_id_t i; + int used; + + if (*len < sizeof(struct ip_set_req_list)) { + ip_set_printk("short SAVE (want >=%zu, got %d)", + sizeof(struct ip_set_req_list), *len); + res = -EINVAL; + goto done; + } + index = req_save->index; + if (index != IP_SET_INVALID_ID + && ip_set_find_byindex(index) != index) { + res = -ENOENT; + goto done; + } + +#define SETLIST(set) (strcmp(set->type->typename, "setlist") == 0) + + used = 0; + if (index == IP_SET_INVALID_ID) { + /* Save all sets: ugly setlist type dependency */ + int setlist = 0; + setlists: + for (i = 0; i < ip_set_max && res == 0; i++) { + if (ip_set_list[i] != NULL + && !(setlist ^ SETLIST(ip_set_list[i]))) + res = ip_set_save_set(i, data, &used, *len); + } + if (!setlist) { + setlist = 1; + goto setlists; + } + } else { + /* Save an individual set */ + res = ip_set_save_set(index, data, &used, *len); + } + if (res == 0) + res = ip_set_save_marker(data, &used, *len); + + if (res != 0) + goto done; + else if (copylen != used) { + res = -EAGAIN; + goto done; + } + goto copy; + } + case IP_SET_OP_RESTORE: { + struct ip_set_req_setnames *req_restore = data; + size_t offset = ALIGNED(sizeof(struct ip_set_req_setnames)); + int line; + + if (*len < offset || *len != req_restore->size) { + ip_set_printk("invalid RESTORE (want =%lu, got %d)", + (long unsigned)req_restore->size, *len); + res = -EINVAL; + goto done; + } + line = ip_set_restore(data + offset, req_restore->size - offset); + DP("ip_set_restore: %d", line); + if (line != 0) { + res = -EAGAIN; + req_restore->size = line; + copylen = sizeof(struct ip_set_req_setnames); + goto copy; + } + goto done; + } + default: + res = -EBADMSG; + goto done; + } /* end of switch(op) */ + + copy: + DP("set %s, copylen %d", index != IP_SET_INVALID_ID + && ip_set_list[index] + ? ip_set_list[index]->name + : ":all:", copylen); + res = copy_to_user(user, data, copylen); + + done: + up(&ip_set_app_mutex); + cleanup: + vfree(data); + if (res > 0) + res = 0; + DP("final result %d", res); + return res; +} + +static struct nf_sockopt_ops so_set = { + .pf = PF_INET, + .set_optmin = SO_IP_SET, + .set_optmax = SO_IP_SET + 1, + .set = &ip_set_sockfn_set, + .get_optmin = SO_IP_SET, + .get_optmax = SO_IP_SET + 1, + .get = &ip_set_sockfn_get, +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23) + .use = 0, +#else + .owner = THIS_MODULE, +#endif +}; + +static int max_sets; + +module_param(max_sets, int, 0600); +MODULE_PARM_DESC(max_sets, "maximal number of sets"); +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("module implementing core IP set support"); + +static int __init +ip_set_init(void) +{ + int res; + + /* For the -rt branch, DECLARE_MUTEX/init_MUTEX avoided */ + sema_init(&ip_set_app_mutex, 1); + + if (max_sets) + ip_set_max = max_sets; + if (ip_set_max >= IP_SET_INVALID_ID) + ip_set_max = IP_SET_INVALID_ID - 1; + + ip_set_list = vmalloc(sizeof(struct ip_set *) * ip_set_max); + if (!ip_set_list) { + printk(KERN_ERR "Unable to create ip_set_list\n"); + return -ENOMEM; + } + memset(ip_set_list, 0, sizeof(struct ip_set *) * ip_set_max); + + INIT_LIST_HEAD(&set_type_list); + + res = nf_register_sockopt(&so_set); + if (res != 0) { + ip_set_printk("SO_SET registry failed: %d", res); + vfree(ip_set_list); + return res; + } + + printk("ip_set version %u loaded\n", IP_SET_PROTOCOL_VERSION); + return 0; +} + +static void __exit +ip_set_fini(void) +{ + /* There can't be any existing set or binding */ + nf_unregister_sockopt(&so_set); + vfree(ip_set_list); + DP("these are the famous last words"); +} + +EXPORT_SYMBOL(ip_set_register_set_type); +EXPORT_SYMBOL(ip_set_unregister_set_type); + +EXPORT_SYMBOL(ip_set_get_byname); +EXPORT_SYMBOL(ip_set_get_byindex); +EXPORT_SYMBOL(ip_set_put_byindex); +EXPORT_SYMBOL(ip_set_id); +EXPORT_SYMBOL(__ip_set_get_byname); +EXPORT_SYMBOL(__ip_set_put_byindex); + +EXPORT_SYMBOL(ip_set_addip_kernel); +EXPORT_SYMBOL(ip_set_delip_kernel); +EXPORT_SYMBOL(ip_set_testip_kernel); + +module_init(ip_set_init); +module_exit(ip_set_fini); diff --git a/net/ipv4/netfilter/ip_set_iphash.c b/net/ipv4/netfilter/ip_set_iphash.c new file mode 100644 index 0000000..1accbe3 --- /dev/null +++ b/net/ipv4/netfilter/ip_set_iphash.c @@ -0,0 +1,164 @@ +/* Copyright (C) 2003-2008 Jozsef Kadlecsik + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +/* Kernel module implementing an ip hash set */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +#include + +static int limit = MAX_RANGE; + +static inline __u32 +iphash_id(struct ip_set *set, ip_set_ip_t ip) +{ + struct ip_set_iphash *map = set->data; + __u32 id; + u_int16_t i; + ip_set_ip_t *elem; + + + ip &= map->netmask; + DP("set: %s, ip:%u.%u.%u.%u", set->name, HIPQUAD(ip)); + for (i = 0; i < map->probes; i++) { + id = jhash_ip(map, i, ip) % map->hashsize; + DP("hash key: %u", id); + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); + if (*elem == ip) + return id; + /* No shortcut - there can be deleted entries. */ + } + return UINT_MAX; +} + +static inline int +iphash_test(struct ip_set *set, ip_set_ip_t ip) +{ + return (ip && iphash_id(set, ip) != UINT_MAX); +} + +#define KADT_CONDITION + +UADT(iphash, test) +KADT(iphash, test, ipaddr) + +static inline int +__iphash_add(struct ip_set_iphash *map, ip_set_ip_t *ip) +{ + __u32 probe; + u_int16_t i; + ip_set_ip_t *elem, *slot = NULL; + + for (i = 0; i < map->probes; i++) { + probe = jhash_ip(map, i, *ip) % map->hashsize; + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, probe); + if (*elem == *ip) + return -EEXIST; + if (!(slot || *elem)) + slot = elem; + /* There can be deleted entries, must check all slots */ + } + if (slot) { + *slot = *ip; + map->elements++; + return 0; + } + /* Trigger rehashing */ + return -EAGAIN; +} + +static inline int +iphash_add(struct ip_set *set, ip_set_ip_t ip) +{ + struct ip_set_iphash *map = set->data; + + if (!ip || map->elements >= limit) + return -ERANGE; + + ip &= map->netmask; + return __iphash_add(map, &ip); +} + +UADT(iphash, add) +KADT(iphash, add, ipaddr) + +static inline void +__iphash_retry(struct ip_set_iphash *tmp, struct ip_set_iphash *map) +{ + tmp->netmask = map->netmask; +} + +HASH_RETRY(iphash, ip_set_ip_t) + +static inline int +iphash_del(struct ip_set *set, ip_set_ip_t ip) +{ + struct ip_set_iphash *map = set->data; + ip_set_ip_t id, *elem; + + if (!ip) + return -ERANGE; + + id = iphash_id(set, ip); + if (id == UINT_MAX) + return -EEXIST; + + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); + *elem = 0; + map->elements--; + + return 0; +} + +UADT(iphash, del) +KADT(iphash, del, ipaddr) + +static inline int +__iphash_create(const struct ip_set_req_iphash_create *req, + struct ip_set_iphash *map) +{ + map->netmask = req->netmask; + + return 0; +} + +HASH_CREATE(iphash, ip_set_ip_t) +HASH_DESTROY(iphash) + +HASH_FLUSH(iphash, ip_set_ip_t) + +static inline void +__iphash_list_header(const struct ip_set_iphash *map, + struct ip_set_req_iphash_create *header) +{ + header->netmask = map->netmask; +} + +HASH_LIST_HEADER(iphash) +HASH_LIST_MEMBERS_SIZE(iphash, ip_set_ip_t) +HASH_LIST_MEMBERS(iphash, ip_set_ip_t) + +IP_SET_RTYPE(iphash, IPSET_TYPE_IP | IPSET_DATA_SINGLE) + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("iphash type of IP sets"); +module_param(limit, int, 0600); +MODULE_PARM_DESC(limit, "maximal number of elements stored in the sets"); + +REGISTER_MODULE(iphash) diff --git a/net/ipv4/netfilter/ip_set_ipmap.c b/net/ipv4/netfilter/ip_set_ipmap.c new file mode 100644 index 0000000..be3c538 --- /dev/null +++ b/net/ipv4/netfilter/ip_set_ipmap.c @@ -0,0 +1,158 @@ +/* Copyright (C) 2000-2002 Joakim Axelsson + * Patrick Schaaf + * Copyright (C) 2003-2008 Jozsef Kadlecsik + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +/* Kernel module implementing an IP set type: the single bitmap type */ + +#include +#include +#include +#include +#include +#include +#include + +#include + +static inline ip_set_ip_t +ip_to_id(const struct ip_set_ipmap *map, ip_set_ip_t ip) +{ + return ((ip & map->netmask) - map->first_ip)/map->hosts; +} + +static inline int +ipmap_test(const struct ip_set *set, ip_set_ip_t ip) +{ + const struct ip_set_ipmap *map = set->data; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + DP("set: %s, ip:%u.%u.%u.%u", set->name, HIPQUAD(ip)); + return !!test_bit(ip_to_id(map, ip), map->members); +} + +#define KADT_CONDITION + +UADT(ipmap, test) +KADT(ipmap, test, ipaddr) + +static inline int +ipmap_add(struct ip_set *set, ip_set_ip_t ip) +{ + struct ip_set_ipmap *map = set->data; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + DP("set: %s, ip:%u.%u.%u.%u", set->name, HIPQUAD(ip)); + if (test_and_set_bit(ip_to_id(map, ip), map->members)) + return -EEXIST; + + return 0; +} + +UADT(ipmap, add) +KADT(ipmap, add, ipaddr) + +static inline int +ipmap_del(struct ip_set *set, ip_set_ip_t ip) +{ + struct ip_set_ipmap *map = set->data; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + DP("set: %s, ip:%u.%u.%u.%u", set->name, HIPQUAD(ip)); + if (!test_and_clear_bit(ip_to_id(map, ip), map->members)) + return -EEXIST; + + return 0; +} + +UADT(ipmap, del) +KADT(ipmap, del, ipaddr) + +static inline int +__ipmap_create(const struct ip_set_req_ipmap_create *req, + struct ip_set_ipmap *map) +{ + map->netmask = req->netmask; + + if (req->netmask == 0xFFFFFFFF) { + map->hosts = 1; + map->sizeid = map->last_ip - map->first_ip + 1; + } else { + unsigned int mask_bits, netmask_bits; + ip_set_ip_t mask; + + map->first_ip &= map->netmask; /* Should we better bark? */ + + mask = range_to_mask(map->first_ip, map->last_ip, &mask_bits); + netmask_bits = mask_to_bits(map->netmask); + + if ((!mask && (map->first_ip || map->last_ip != 0xFFFFFFFF)) + || netmask_bits <= mask_bits) + return -ENOEXEC; + + DP("mask_bits %u, netmask_bits %u", + mask_bits, netmask_bits); + map->hosts = 2 << (32 - netmask_bits - 1); + map->sizeid = 2 << (netmask_bits - mask_bits - 1); + } + if (map->sizeid > MAX_RANGE + 1) { + ip_set_printk("range too big, %d elements (max %d)", + map->sizeid, MAX_RANGE+1); + return -ENOEXEC; + } + DP("hosts %u, sizeid %u", map->hosts, map->sizeid); + return bitmap_bytes(0, map->sizeid - 1); +} + +BITMAP_CREATE(ipmap) +BITMAP_DESTROY(ipmap) +BITMAP_FLUSH(ipmap) + +static inline void +__ipmap_list_header(const struct ip_set_ipmap *map, + struct ip_set_req_ipmap_create *header) +{ + header->netmask = map->netmask; +} + +BITMAP_LIST_HEADER(ipmap) +BITMAP_LIST_MEMBERS_SIZE(ipmap, ip_set_ip_t, map->sizeid, + test_bit(i, map->members)) + +static void +ipmap_list_members(const struct ip_set *set, void *data, char dont_align) +{ + const struct ip_set_ipmap *map = set->data; + uint32_t i, n = 0; + ip_set_ip_t *d; + + if (dont_align) { + memcpy(data, map->members, map->size); + return; + } + + for (i = 0; i < map->sizeid; i++) + if (test_bit(i, map->members)) { + d = data + n * IPSET_ALIGN(sizeof(ip_set_ip_t)); + *d = map->first_ip + i * map->hosts; + n++; + } +} + +IP_SET_TYPE(ipmap, IPSET_TYPE_IP | IPSET_DATA_SINGLE) + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("ipmap type of IP sets"); + +REGISTER_MODULE(ipmap) diff --git a/net/ipv4/netfilter/ip_set_ipporthash.c b/net/ipv4/netfilter/ip_set_ipporthash.c new file mode 100644 index 0000000..4a79c8b --- /dev/null +++ b/net/ipv4/netfilter/ip_set_ipporthash.c @@ -0,0 +1,197 @@ +/* Copyright (C) 2003-2008 Jozsef Kadlecsik + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +/* Kernel module implementing an ip+port hash set */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +#include +#include + +static int limit = MAX_RANGE; + +static inline __u32 +ipporthash_id(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t port) +{ + struct ip_set_ipporthash *map = set->data; + __u32 id; + u_int16_t i; + ip_set_ip_t *elem; + + ip = pack_ip_port(map, ip, port); + + if (!ip) + return UINT_MAX; + + for (i = 0; i < map->probes; i++) { + id = jhash_ip(map, i, ip) % map->hashsize; + DP("hash key: %u", id); + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); + if (*elem == ip) + return id; + /* No shortcut - there can be deleted entries. */ + } + return UINT_MAX; +} + +static inline int +ipporthash_test(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t port) +{ + struct ip_set_ipporthash *map = set->data; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + return (ipporthash_id(set, ip, port) != UINT_MAX); +} + +#define KADT_CONDITION \ + ip_set_ip_t port; \ + \ + if (flags[1] == 0) \ + return 0; \ + \ + port = get_port(skb, ++flags); \ + \ + if (port == INVALID_PORT) \ + return 0; + +UADT(ipporthash, test, req->port) +KADT(ipporthash, test, ipaddr, port) + +static inline int +__ipporthash_add(struct ip_set_ipporthash *map, ip_set_ip_t *ip) +{ + __u32 probe; + u_int16_t i; + ip_set_ip_t *elem, *slot = NULL; + + for (i = 0; i < map->probes; i++) { + probe = jhash_ip(map, i, *ip) % map->hashsize; + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, probe); + if (*elem == *ip) + return -EEXIST; + if (!(slot || *elem)) + slot = elem; + /* There can be deleted entries, must check all slots */ + } + if (slot) { + *slot = *ip; + map->elements++; + return 0; + } + /* Trigger rehashing */ + return -EAGAIN; +} + +static inline int +ipporthash_add(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t port) +{ + struct ip_set_ipporthash *map = set->data; + if (map->elements > limit) + return -ERANGE; + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + ip = pack_ip_port(map, ip, port); + + if (!ip) + return -ERANGE; + + return __ipporthash_add(map, &ip); +} + +UADT(ipporthash, add, req->port) +KADT(ipporthash, add, ipaddr, port) + +static inline void +__ipporthash_retry(struct ip_set_ipporthash *tmp, + struct ip_set_ipporthash *map) +{ + tmp->first_ip = map->first_ip; + tmp->last_ip = map->last_ip; +} + +HASH_RETRY(ipporthash, ip_set_ip_t) + +static inline int +ipporthash_del(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t port) +{ + struct ip_set_ipporthash *map = set->data; + ip_set_ip_t id; + ip_set_ip_t *elem; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + id = ipporthash_id(set, ip, port); + + if (id == UINT_MAX) + return -EEXIST; + + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); + *elem = 0; + map->elements--; + + return 0; +} + +UADT(ipporthash, del, req->port) +KADT(ipporthash, del, ipaddr, port) + +static inline int +__ipporthash_create(const struct ip_set_req_ipporthash_create *req, + struct ip_set_ipporthash *map) +{ + if (req->to - req->from > MAX_RANGE) { + ip_set_printk("range too big, %d elements (max %d)", + req->to - req->from + 1, MAX_RANGE+1); + return -ENOEXEC; + } + map->first_ip = req->from; + map->last_ip = req->to; + return 0; +} + +HASH_CREATE(ipporthash, ip_set_ip_t) +HASH_DESTROY(ipporthash) +HASH_FLUSH(ipporthash, ip_set_ip_t) + +static inline void +__ipporthash_list_header(const struct ip_set_ipporthash *map, + struct ip_set_req_ipporthash_create *header) +{ + header->from = map->first_ip; + header->to = map->last_ip; +} + +HASH_LIST_HEADER(ipporthash) +HASH_LIST_MEMBERS_SIZE(ipporthash, ip_set_ip_t) +HASH_LIST_MEMBERS(ipporthash, ip_set_ip_t) + +IP_SET_RTYPE(ipporthash, IPSET_TYPE_IP | IPSET_TYPE_PORT | IPSET_DATA_DOUBLE) + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("ipporthash type of IP sets"); +module_param(limit, int, 0600); +MODULE_PARM_DESC(limit, "maximal number of elements stored in the sets"); + +REGISTER_MODULE(ipporthash) diff --git a/net/ipv4/netfilter/ip_set_ipportiphash.c b/net/ipv4/netfilter/ip_set_ipportiphash.c new file mode 100644 index 0000000..c80087f --- /dev/null +++ b/net/ipv4/netfilter/ip_set_ipportiphash.c @@ -0,0 +1,215 @@ +/* Copyright (C) 2008 Jozsef Kadlecsik + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +/* Kernel module implementing an ip+port+ip hash set */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +#include +#include + +static int limit = MAX_RANGE; + +#define jhash_ip2(map, i, ipport, ip1) \ + jhash_2words(ipport, ip1, *(map->initval + i)) + +static inline __u32 +ipportiphash_id(struct ip_set *set, + ip_set_ip_t ip, ip_set_ip_t port, ip_set_ip_t ip1) +{ + struct ip_set_ipportiphash *map = set->data; + __u32 id; + u_int16_t i; + struct ipportip *elem; + + ip = pack_ip_port(map, ip, port); + if (!(ip || ip1)) + return UINT_MAX; + + for (i = 0; i < map->probes; i++) { + id = jhash_ip2(map, i, ip, ip1) % map->hashsize; + DP("hash key: %u", id); + elem = HARRAY_ELEM(map->members, struct ipportip *, id); + if (elem->ip == ip && elem->ip1 == ip1) + return id; + /* No shortcut - there can be deleted entries. */ + } + return UINT_MAX; +} + +static inline int +ipportiphash_test(struct ip_set *set, + ip_set_ip_t ip, ip_set_ip_t port, ip_set_ip_t ip1) +{ + struct ip_set_ipportiphash *map = set->data; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + return (ipportiphash_id(set, ip, port, ip1) != UINT_MAX); +} + +#define KADT_CONDITION \ + ip_set_ip_t port, ip1; \ + \ + if (flags[2] == 0) \ + return 0; \ + \ + port = get_port(skb, ++flags); \ + ip1 = ipaddr(skb, ++flags); \ + \ + if (port == INVALID_PORT) \ + return 0; + +UADT(ipportiphash, test, req->port, req->ip1) +KADT(ipportiphash, test, ipaddr, port, ip1) + +static inline int +__ipportip_add(struct ip_set_ipportiphash *map, + ip_set_ip_t ip, ip_set_ip_t ip1) +{ + __u32 probe; + u_int16_t i; + struct ipportip *elem, *slot = NULL; + + for (i = 0; i < map->probes; i++) { + probe = jhash_ip2(map, i, ip, ip1) % map->hashsize; + elem = HARRAY_ELEM(map->members, struct ipportip *, probe); + if (elem->ip == ip && elem->ip1 == ip1) + return -EEXIST; + if (!(slot || elem->ip || elem->ip1)) + slot = elem; + /* There can be deleted entries, must check all slots */ + } + if (slot) { + slot->ip = ip; + slot->ip1 = ip1; + map->elements++; + return 0; + } + /* Trigger rehashing */ + return -EAGAIN; +} + +static inline int +__ipportiphash_add(struct ip_set_ipportiphash *map, + struct ipportip *elem) +{ + return __ipportip_add(map, elem->ip, elem->ip1); +} + +static inline int +ipportiphash_add(struct ip_set *set, + ip_set_ip_t ip, ip_set_ip_t port, ip_set_ip_t ip1) +{ + struct ip_set_ipportiphash *map = set->data; + + if (map->elements > limit) + return -ERANGE; + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + ip = pack_ip_port(map, ip, port); + if (!(ip || ip1)) + return -ERANGE; + + return __ipportip_add(map, ip, ip1); +} + +UADT(ipportiphash, add, req->port, req->ip1) +KADT(ipportiphash, add, ipaddr, port, ip1) + +static inline void +__ipportiphash_retry(struct ip_set_ipportiphash *tmp, + struct ip_set_ipportiphash *map) +{ + tmp->first_ip = map->first_ip; + tmp->last_ip = map->last_ip; +} + +HASH_RETRY2(ipportiphash, struct ipportip) + +static inline int +ipportiphash_del(struct ip_set *set, + ip_set_ip_t ip, ip_set_ip_t port, ip_set_ip_t ip1) +{ + struct ip_set_ipportiphash *map = set->data; + ip_set_ip_t id; + struct ipportip *elem; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + id = ipportiphash_id(set, ip, port, ip1); + + if (id == UINT_MAX) + return -EEXIST; + + elem = HARRAY_ELEM(map->members, struct ipportip *, id); + elem->ip = elem->ip1 = 0; + map->elements--; + + return 0; +} + +UADT(ipportiphash, del, req->port, req->ip1) +KADT(ipportiphash, del, ipaddr, port, ip1) + +static inline int +__ipportiphash_create(const struct ip_set_req_ipportiphash_create *req, + struct ip_set_ipportiphash *map) +{ + if (req->to - req->from > MAX_RANGE) { + ip_set_printk("range too big, %d elements (max %d)", + req->to - req->from + 1, MAX_RANGE+1); + return -ENOEXEC; + } + map->first_ip = req->from; + map->last_ip = req->to; + return 0; +} + +HASH_CREATE(ipportiphash, struct ipportip) +HASH_DESTROY(ipportiphash) +HASH_FLUSH(ipportiphash, struct ipportip) + +static inline void +__ipportiphash_list_header(const struct ip_set_ipportiphash *map, + struct ip_set_req_ipportiphash_create *header) +{ + header->from = map->first_ip; + header->to = map->last_ip; +} + +HASH_LIST_HEADER(ipportiphash) +HASH_LIST_MEMBERS_SIZE(ipportiphash, struct ipportip) +HASH_LIST_MEMBERS_MEMCPY(ipportiphash, struct ipportip, + (elem->ip || elem->ip1)) + +IP_SET_RTYPE(ipportiphash, IPSET_TYPE_IP | IPSET_TYPE_PORT + | IPSET_TYPE_IP1 | IPSET_DATA_TRIPLE) + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("ipportiphash type of IP sets"); +module_param(limit, int, 0600); +MODULE_PARM_DESC(limit, "maximal number of elements stored in the sets"); + +REGISTER_MODULE(ipportiphash) diff --git a/net/ipv4/netfilter/ip_set_ipportnethash.c b/net/ipv4/netfilter/ip_set_ipportnethash.c new file mode 100644 index 0000000..2680cd9 --- /dev/null +++ b/net/ipv4/netfilter/ip_set_ipportnethash.c @@ -0,0 +1,298 @@ +/* Copyright (C) 2008 Jozsef Kadlecsik + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +/* Kernel module implementing an ip+port+net hash set */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +#include +#include + +static int limit = MAX_RANGE; + +#define jhash_ip2(map, i, ipport, ip1) \ + jhash_2words(ipport, ip1, *(map->initval + i)) + +static inline __u32 +ipportnethash_id_cidr(struct ip_set *set, + ip_set_ip_t ip, ip_set_ip_t port, + ip_set_ip_t ip1, uint8_t cidr) +{ + struct ip_set_ipportnethash *map = set->data; + __u32 id; + u_int16_t i; + struct ipportip *elem; + + ip = pack_ip_port(map, ip, port); + ip1 = pack_ip_cidr(ip1, cidr); + if (!(ip || ip1)) + return UINT_MAX; + + for (i = 0; i < map->probes; i++) { + id = jhash_ip2(map, i, ip, ip1) % map->hashsize; + DP("hash key: %u", id); + elem = HARRAY_ELEM(map->members, struct ipportip *, id); + if (elem->ip == ip && elem->ip1 == ip1) + return id; + /* No shortcut - there can be deleted entries. */ + } + return UINT_MAX; +} + +static inline __u32 +ipportnethash_id(struct ip_set *set, + ip_set_ip_t ip, ip_set_ip_t port, ip_set_ip_t ip1) +{ + struct ip_set_ipportnethash *map = set->data; + __u32 id = UINT_MAX; + int i; + + for (i = 0; i < 30 && map->cidr[i]; i++) { + id = ipportnethash_id_cidr(set, ip, port, ip1, map->cidr[i]); + if (id != UINT_MAX) + break; + } + return id; +} + +static inline int +ipportnethash_test_cidr(struct ip_set *set, + ip_set_ip_t ip, ip_set_ip_t port, + ip_set_ip_t ip1, uint8_t cidr) +{ + struct ip_set_ipportnethash *map = set->data; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + return (ipportnethash_id_cidr(set, ip, port, ip1, cidr) != UINT_MAX); +} + +static inline int +ipportnethash_test(struct ip_set *set, + ip_set_ip_t ip, ip_set_ip_t port, ip_set_ip_t ip1) +{ + struct ip_set_ipportnethash *map = set->data; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + return (ipportnethash_id(set, ip, port, ip1) != UINT_MAX); +} + +static int +ipportnethash_utest(struct ip_set *set, const void *data, u_int32_t size) +{ + const struct ip_set_req_ipportnethash *req = data; + + if (req->cidr <= 0 || req->cidr > 32) + return -EINVAL; + return (req->cidr == 32 + ? ipportnethash_test(set, req->ip, req->port, req->ip1) + : ipportnethash_test_cidr(set, req->ip, req->port, + req->ip1, req->cidr)); +} + +#define KADT_CONDITION \ + ip_set_ip_t port, ip1; \ + \ + if (flags[2] == 0) \ + return 0; \ + \ + port = get_port(skb, ++flags); \ + ip1 = ipaddr(skb, ++flags); \ + \ + if (port == INVALID_PORT) \ + return 0; + +KADT(ipportnethash, test, ipaddr, port, ip1) + +static inline int +__ipportnet_add(struct ip_set_ipportnethash *map, + ip_set_ip_t ip, ip_set_ip_t ip1) +{ + __u32 probe; + u_int16_t i; + struct ipportip *elem, *slot = NULL; + + for (i = 0; i < map->probes; i++) { + probe = jhash_ip2(map, i, ip, ip1) % map->hashsize; + elem = HARRAY_ELEM(map->members, struct ipportip *, probe); + if (elem->ip == ip && elem->ip1 == ip1) + return -EEXIST; + if (!(slot || elem->ip || elem->ip1)) + slot = elem; + /* There can be deleted entries, must check all slots */ + } + if (slot) { + slot->ip = ip; + slot->ip1 = ip1; + map->elements++; + return 0; + } + /* Trigger rehashing */ + return -EAGAIN; +} + +static inline int +__ipportnethash_add(struct ip_set_ipportnethash *map, + struct ipportip *elem) +{ + return __ipportnet_add(map, elem->ip, elem->ip1); +} + +static inline int +ipportnethash_add(struct ip_set *set, + ip_set_ip_t ip, ip_set_ip_t port, + ip_set_ip_t ip1, uint8_t cidr) +{ + struct ip_set_ipportnethash *map = set->data; + struct ipportip; + int ret; + + if (map->elements > limit) + return -ERANGE; + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + if (cidr <= 0 || cidr >= 32) + return -EINVAL; + if (map->nets[cidr-1] == UINT16_MAX) + return -ERANGE; + + ip = pack_ip_port(map, ip, port); + ip1 = pack_ip_cidr(ip1, cidr); + if (!(ip || ip1)) + return -ERANGE; + + ret =__ipportnet_add(map, ip, ip1); + if (ret == 0) { + if (!map->nets[cidr-1]++) + add_cidr_size(map->cidr, cidr); + } + return ret; +} + +#undef KADT_CONDITION +#define KADT_CONDITION \ + struct ip_set_ipportnethash *map = set->data; \ + uint8_t cidr = map->cidr[0] ? map->cidr[0] : 31; \ + ip_set_ip_t port, ip1; \ + \ + if (flags[2] == 0) \ + return 0; \ + \ + port = get_port(skb, flags++); \ + ip1 = ipaddr(skb, flags++); \ + \ + if (port == INVALID_PORT) \ + return 0; + +UADT(ipportnethash, add, req->port, req->ip1, req->cidr) +KADT(ipportnethash, add, ipaddr, port, ip1, cidr) + +static inline void +__ipportnethash_retry(struct ip_set_ipportnethash *tmp, + struct ip_set_ipportnethash *map) +{ + tmp->first_ip = map->first_ip; + tmp->last_ip = map->last_ip; + memcpy(tmp->cidr, map->cidr, sizeof(tmp->cidr)); + memcpy(tmp->nets, map->nets, sizeof(tmp->nets)); +} + +HASH_RETRY2(ipportnethash, struct ipportip) + +static inline int +ipportnethash_del(struct ip_set *set, + ip_set_ip_t ip, ip_set_ip_t port, + ip_set_ip_t ip1, uint8_t cidr) +{ + struct ip_set_ipportnethash *map = set->data; + ip_set_ip_t id; + struct ipportip *elem; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + if (!ip) + return -ERANGE; + if (cidr <= 0 || cidr >= 32) + return -EINVAL; + + id = ipportnethash_id_cidr(set, ip, port, ip1, cidr); + + if (id == UINT_MAX) + return -EEXIST; + + elem = HARRAY_ELEM(map->members, struct ipportip *, id); + elem->ip = elem->ip1 = 0; + map->elements--; + if (!map->nets[cidr-1]--) + del_cidr_size(map->cidr, cidr); + + return 0; +} + +UADT(ipportnethash, del, req->port, req->ip1, req->cidr) +KADT(ipportnethash, del, ipaddr, port, ip1, cidr) + +static inline int +__ipportnethash_create(const struct ip_set_req_ipportnethash_create *req, + struct ip_set_ipportnethash *map) +{ + if (req->to - req->from > MAX_RANGE) { + ip_set_printk("range too big, %d elements (max %d)", + req->to - req->from + 1, MAX_RANGE+1); + return -ENOEXEC; + } + map->first_ip = req->from; + map->last_ip = req->to; + memset(map->cidr, 0, sizeof(map->cidr)); + memset(map->nets, 0, sizeof(map->nets)); + return 0; +} + +HASH_CREATE(ipportnethash, struct ipportip) +HASH_DESTROY(ipportnethash) +HASH_FLUSH_CIDR(ipportnethash, struct ipportip); + +static inline void +__ipportnethash_list_header(const struct ip_set_ipportnethash *map, + struct ip_set_req_ipportnethash_create *header) +{ + header->from = map->first_ip; + header->to = map->last_ip; +} + +HASH_LIST_HEADER(ipportnethash) + +HASH_LIST_MEMBERS_SIZE(ipportnethash, struct ipportip) +HASH_LIST_MEMBERS_MEMCPY(ipportnethash, struct ipportip, + (elem->ip || elem->ip1)) + +IP_SET_RTYPE(ipportnethash, IPSET_TYPE_IP | IPSET_TYPE_PORT + | IPSET_TYPE_IP1 | IPSET_DATA_TRIPLE) + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("ipportnethash type of IP sets"); +module_param(limit, int, 0600); +MODULE_PARM_DESC(limit, "maximal number of elements stored in the sets"); + +REGISTER_MODULE(ipportnethash) diff --git a/net/ipv4/netfilter/ip_set_iptree.c b/net/ipv4/netfilter/ip_set_iptree.c new file mode 100644 index 0000000..77eb180 --- /dev/null +++ b/net/ipv4/netfilter/ip_set_iptree.c @@ -0,0 +1,464 @@ +/* Copyright (C) 2005-2008 Jozsef Kadlecsik + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +/* Kernel module implementing an IP set type: the iptree type */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include + +static int limit = MAX_RANGE; + +/* Garbage collection interval in seconds: */ +#define IPTREE_GC_TIME 5*60 +/* Sleep so many milliseconds before trying again + * to delete the gc timer at destroying/flushing a set */ +#define IPTREE_DESTROY_SLEEP 100 + +static __KMEM_CACHE_T__ *branch_cachep; +static __KMEM_CACHE_T__ *leaf_cachep; + + +#if defined(__LITTLE_ENDIAN) +#define ABCD(a,b,c,d,addrp) do { \ + a = ((unsigned char *)addrp)[3]; \ + b = ((unsigned char *)addrp)[2]; \ + c = ((unsigned char *)addrp)[1]; \ + d = ((unsigned char *)addrp)[0]; \ +} while (0) +#elif defined(__BIG_ENDIAN) +#define ABCD(a,b,c,d,addrp) do { \ + a = ((unsigned char *)addrp)[0]; \ + b = ((unsigned char *)addrp)[1]; \ + c = ((unsigned char *)addrp)[2]; \ + d = ((unsigned char *)addrp)[3]; \ +} while (0) +#else +#error "Please fix asm/byteorder.h" +#endif /* __LITTLE_ENDIAN */ + +#define TESTIP_WALK(map, elem, branch) do { \ + if ((map)->tree[elem]) { \ + branch = (map)->tree[elem]; \ + } else \ + return 0; \ +} while (0) + +static inline int +iptree_test(struct ip_set *set, ip_set_ip_t ip) +{ + struct ip_set_iptree *map = set->data; + struct ip_set_iptreeb *btree; + struct ip_set_iptreec *ctree; + struct ip_set_iptreed *dtree; + unsigned char a,b,c,d; + + if (!ip) + return -ERANGE; + + ABCD(a, b, c, d, &ip); + DP("%u %u %u %u timeout %u", a, b, c, d, map->timeout); + TESTIP_WALK(map, a, btree); + TESTIP_WALK(btree, b, ctree); + TESTIP_WALK(ctree, c, dtree); + DP("%lu %lu", dtree->expires[d], jiffies); + return dtree->expires[d] + && (!map->timeout + || time_after(dtree->expires[d], jiffies)); +} + +#define KADT_CONDITION + +UADT(iptree, test) +KADT(iptree, test, ipaddr) + +#define ADDIP_WALK(map, elem, branch, type, cachep) do { \ + if ((map)->tree[elem]) { \ + DP("found %u", elem); \ + branch = (map)->tree[elem]; \ + } else { \ + branch = (type *) \ + kmem_cache_alloc(cachep, GFP_ATOMIC); \ + if (branch == NULL) \ + return -ENOMEM; \ + memset(branch, 0, sizeof(*branch)); \ + (map)->tree[elem] = branch; \ + DP("alloc %u", elem); \ + } \ +} while (0) + +static inline int +iptree_add(struct ip_set *set, ip_set_ip_t ip, unsigned int timeout) +{ + struct ip_set_iptree *map = set->data; + struct ip_set_iptreeb *btree; + struct ip_set_iptreec *ctree; + struct ip_set_iptreed *dtree; + unsigned char a,b,c,d; + int ret = 0; + + if (!ip || map->elements >= limit) + /* We could call the garbage collector + * but it's probably overkill */ + return -ERANGE; + + ABCD(a, b, c, d, &ip); + DP("%u %u %u %u timeout %u", a, b, c, d, timeout); + ADDIP_WALK(map, a, btree, struct ip_set_iptreeb, branch_cachep); + ADDIP_WALK(btree, b, ctree, struct ip_set_iptreec, branch_cachep); + ADDIP_WALK(ctree, c, dtree, struct ip_set_iptreed, leaf_cachep); + if (dtree->expires[d] + && (!map->timeout || time_after(dtree->expires[d], jiffies))) + ret = -EEXIST; + if (map->timeout && timeout == 0) + timeout = map->timeout; + dtree->expires[d] = map->timeout ? (timeout * HZ + jiffies) : 1; + /* Lottery: I won! */ + if (dtree->expires[d] == 0) + dtree->expires[d] = 1; + DP("%u %lu", d, dtree->expires[d]); + if (ret == 0) + map->elements++; + return ret; +} + +UADT(iptree, add, req->timeout) +KADT(iptree, add, ipaddr, 0) + +#define DELIP_WALK(map, elem, branch) do { \ + if ((map)->tree[elem]) { \ + branch = (map)->tree[elem]; \ + } else \ + return -EEXIST; \ +} while (0) + +static inline int +iptree_del(struct ip_set *set, ip_set_ip_t ip) +{ + struct ip_set_iptree *map = set->data; + struct ip_set_iptreeb *btree; + struct ip_set_iptreec *ctree; + struct ip_set_iptreed *dtree; + unsigned char a,b,c,d; + + if (!ip) + return -ERANGE; + + ABCD(a, b, c, d, &ip); + DELIP_WALK(map, a, btree); + DELIP_WALK(btree, b, ctree); + DELIP_WALK(ctree, c, dtree); + + if (dtree->expires[d]) { + dtree->expires[d] = 0; + map->elements--; + return 0; + } + return -EEXIST; +} + +UADT(iptree, del) +KADT(iptree, del, ipaddr) + +#define LOOP_WALK_BEGIN(map, i, branch) \ + for (i = 0; i < 256; i++) { \ + if (!(map)->tree[i]) \ + continue; \ + branch = (map)->tree[i] + +#define LOOP_WALK_END } + +static void +ip_tree_gc(unsigned long ul_set) +{ + struct ip_set *set = (struct ip_set *) ul_set; + struct ip_set_iptree *map = set->data; + struct ip_set_iptreeb *btree; + struct ip_set_iptreec *ctree; + struct ip_set_iptreed *dtree; + unsigned int a,b,c,d; + unsigned char i,j,k; + + i = j = k = 0; + DP("gc: %s", set->name); + write_lock_bh(&set->lock); + LOOP_WALK_BEGIN(map, a, btree); + LOOP_WALK_BEGIN(btree, b, ctree); + LOOP_WALK_BEGIN(ctree, c, dtree); + for (d = 0; d < 256; d++) { + if (dtree->expires[d]) { + DP("gc: %u %u %u %u: expires %lu jiffies %lu", + a, b, c, d, + dtree->expires[d], jiffies); + if (map->timeout + && time_before(dtree->expires[d], jiffies)) { + dtree->expires[d] = 0; + map->elements--; + } else + k = 1; + } + } + if (k == 0) { + DP("gc: %s: leaf %u %u %u empty", + set->name, a, b, c); + kmem_cache_free(leaf_cachep, dtree); + ctree->tree[c] = NULL; + } else { + DP("gc: %s: leaf %u %u %u not empty", + set->name, a, b, c); + j = 1; + k = 0; + } + LOOP_WALK_END; + if (j == 0) { + DP("gc: %s: branch %u %u empty", + set->name, a, b); + kmem_cache_free(branch_cachep, ctree); + btree->tree[b] = NULL; + } else { + DP("gc: %s: branch %u %u not empty", + set->name, a, b); + i = 1; + j = k = 0; + } + LOOP_WALK_END; + if (i == 0) { + DP("gc: %s: branch %u empty", + set->name, a); + kmem_cache_free(branch_cachep, btree); + map->tree[a] = NULL; + } else { + DP("gc: %s: branch %u not empty", + set->name, a); + i = j = k = 0; + } + LOOP_WALK_END; + write_unlock_bh(&set->lock); + + map->gc.expires = jiffies + map->gc_interval * HZ; + add_timer(&map->gc); +} + +static inline void +init_gc_timer(struct ip_set *set) +{ + struct ip_set_iptree *map = set->data; + + /* Even if there is no timeout for the entries, + * we still have to call gc because delete + * do not clean up empty branches */ + map->gc_interval = IPTREE_GC_TIME; + init_timer(&map->gc); + map->gc.data = (unsigned long) set; + map->gc.function = ip_tree_gc; + map->gc.expires = jiffies + map->gc_interval * HZ; + add_timer(&map->gc); +} + +static int +iptree_create(struct ip_set *set, const void *data, u_int32_t size) +{ + const struct ip_set_req_iptree_create *req = data; + struct ip_set_iptree *map; + + if (size != sizeof(struct ip_set_req_iptree_create)) { + ip_set_printk("data length wrong (want %zu, have %lu)", + sizeof(struct ip_set_req_iptree_create), + (unsigned long)size); + return -EINVAL; + } + + map = kmalloc(sizeof(struct ip_set_iptree), GFP_KERNEL); + if (!map) { + DP("out of memory for %zu bytes", + sizeof(struct ip_set_iptree)); + return -ENOMEM; + } + memset(map, 0, sizeof(*map)); + map->timeout = req->timeout; + map->elements = 0; + set->data = map; + + init_gc_timer(set); + + return 0; +} + +static inline void +__flush(struct ip_set_iptree *map) +{ + struct ip_set_iptreeb *btree; + struct ip_set_iptreec *ctree; + struct ip_set_iptreed *dtree; + unsigned int a,b,c; + + LOOP_WALK_BEGIN(map, a, btree); + LOOP_WALK_BEGIN(btree, b, ctree); + LOOP_WALK_BEGIN(ctree, c, dtree); + kmem_cache_free(leaf_cachep, dtree); + LOOP_WALK_END; + kmem_cache_free(branch_cachep, ctree); + LOOP_WALK_END; + kmem_cache_free(branch_cachep, btree); + LOOP_WALK_END; + map->elements = 0; +} + +static void +iptree_destroy(struct ip_set *set) +{ + struct ip_set_iptree *map = set->data; + + /* gc might be running */ + while (!del_timer(&map->gc)) + msleep(IPTREE_DESTROY_SLEEP); + __flush(map); + kfree(map); + set->data = NULL; +} + +static void +iptree_flush(struct ip_set *set) +{ + struct ip_set_iptree *map = set->data; + unsigned int timeout = map->timeout; + + /* gc might be running */ + while (!del_timer(&map->gc)) + msleep(IPTREE_DESTROY_SLEEP); + __flush(map); + memset(map, 0, sizeof(*map)); + map->timeout = timeout; + + init_gc_timer(set); +} + +static void +iptree_list_header(const struct ip_set *set, void *data) +{ + const struct ip_set_iptree *map = set->data; + struct ip_set_req_iptree_create *header = data; + + header->timeout = map->timeout; +} + +static int +iptree_list_members_size(const struct ip_set *set, char dont_align) +{ + const struct ip_set_iptree *map = set->data; + struct ip_set_iptreeb *btree; + struct ip_set_iptreec *ctree; + struct ip_set_iptreed *dtree; + unsigned int a,b,c,d; + unsigned int count = 0; + + LOOP_WALK_BEGIN(map, a, btree); + LOOP_WALK_BEGIN(btree, b, ctree); + LOOP_WALK_BEGIN(ctree, c, dtree); + for (d = 0; d < 256; d++) { + if (dtree->expires[d] + && (!map->timeout || time_after(dtree->expires[d], jiffies))) + count++; + } + LOOP_WALK_END; + LOOP_WALK_END; + LOOP_WALK_END; + + DP("members %u", count); + return (count * IPSET_VALIGN(sizeof(struct ip_set_req_iptree), dont_align)); +} + +static void +iptree_list_members(const struct ip_set *set, void *data, char dont_align) +{ + const struct ip_set_iptree *map = set->data; + struct ip_set_iptreeb *btree; + struct ip_set_iptreec *ctree; + struct ip_set_iptreed *dtree; + unsigned int a,b,c,d; + size_t offset = 0, datasize; + struct ip_set_req_iptree *entry; + + datasize = IPSET_VALIGN(sizeof(struct ip_set_req_iptree), dont_align); + LOOP_WALK_BEGIN(map, a, btree); + LOOP_WALK_BEGIN(btree, b, ctree); + LOOP_WALK_BEGIN(ctree, c, dtree); + for (d = 0; d < 256; d++) { + if (dtree->expires[d] + && (!map->timeout || time_after(dtree->expires[d], jiffies))) { + entry = data + offset; + entry->ip = ((a << 24) | (b << 16) | (c << 8) | d); + entry->timeout = !map->timeout ? 0 + : (dtree->expires[d] - jiffies)/HZ; + offset += datasize; + } + } + LOOP_WALK_END; + LOOP_WALK_END; + LOOP_WALK_END; +} + +IP_SET_TYPE(iptree, IPSET_TYPE_IP | IPSET_DATA_SINGLE) + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("iptree type of IP sets"); +module_param(limit, int, 0600); +MODULE_PARM_DESC(limit, "maximal number of elements stored in the sets"); + +static int __init ip_set_iptree_init(void) +{ + int ret; + + branch_cachep = KMEM_CACHE_CREATE("ip_set_iptreeb", + sizeof(struct ip_set_iptreeb)); + if (!branch_cachep) { + printk(KERN_ERR "Unable to create ip_set_iptreeb slab cache\n"); + ret = -ENOMEM; + goto out; + } + leaf_cachep = KMEM_CACHE_CREATE("ip_set_iptreed", + sizeof(struct ip_set_iptreed)); + if (!leaf_cachep) { + printk(KERN_ERR "Unable to create ip_set_iptreed slab cache\n"); + ret = -ENOMEM; + goto free_branch; + } + ret = ip_set_register_set_type(&ip_set_iptree); + if (ret == 0) + goto out; + + kmem_cache_destroy(leaf_cachep); + free_branch: + kmem_cache_destroy(branch_cachep); + out: + return ret; +} + +static void __exit ip_set_iptree_fini(void) +{ + /* FIXME: possible race with ip_set_create() */ + ip_set_unregister_set_type(&ip_set_iptree); + kmem_cache_destroy(leaf_cachep); + kmem_cache_destroy(branch_cachep); +} + +module_init(ip_set_iptree_init); +module_exit(ip_set_iptree_fini); diff --git a/net/ipv4/netfilter/ip_set_iptreemap.c b/net/ipv4/netfilter/ip_set_iptreemap.c new file mode 100644 index 0000000..76d0142 --- /dev/null +++ b/net/ipv4/netfilter/ip_set_iptreemap.c @@ -0,0 +1,699 @@ +/* Copyright (C) 2007 Sven Wegener + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License version 2 as published by + * the Free Software Foundation. + */ + +/* This modules implements the iptreemap ipset type. It uses bitmaps to + * represent every single IPv4 address as a bit. The bitmaps are managed in a + * tree structure, where the first three octets of an address are used as an + * index to find the bitmap and the last octet is used as the bit number. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include +#include + +#define IPTREEMAP_DEFAULT_GC_TIME (5 * 60) +#define IPTREEMAP_DESTROY_SLEEP (100) + +static __KMEM_CACHE_T__ *cachep_b; +static __KMEM_CACHE_T__ *cachep_c; +static __KMEM_CACHE_T__ *cachep_d; + +static struct ip_set_iptreemap_d *fullbitmap_d; +static struct ip_set_iptreemap_c *fullbitmap_c; +static struct ip_set_iptreemap_b *fullbitmap_b; + +#if defined(__LITTLE_ENDIAN) +#define ABCD(a, b, c, d, addr) \ + do { \ + a = ((unsigned char *)addr)[3]; \ + b = ((unsigned char *)addr)[2]; \ + c = ((unsigned char *)addr)[1]; \ + d = ((unsigned char *)addr)[0]; \ + } while (0) +#elif defined(__BIG_ENDIAN) +#define ABCD(a,b,c,d,addrp) do { \ + a = ((unsigned char *)addrp)[0]; \ + b = ((unsigned char *)addrp)[1]; \ + c = ((unsigned char *)addrp)[2]; \ + d = ((unsigned char *)addrp)[3]; \ +} while (0) +#else +#error "Please fix asm/byteorder.h" +#endif /* __LITTLE_ENDIAN */ + +#define TESTIP_WALK(map, elem, branch, full) \ + do { \ + branch = (map)->tree[elem]; \ + if (!branch) \ + return 0; \ + else if (branch == full) \ + return 1; \ + } while (0) + +#define ADDIP_WALK(map, elem, branch, type, cachep, full) \ + do { \ + branch = (map)->tree[elem]; \ + if (!branch) { \ + branch = (type *) kmem_cache_alloc(cachep, GFP_ATOMIC); \ + if (!branch) \ + return -ENOMEM; \ + memset(branch, 0, sizeof(*branch)); \ + (map)->tree[elem] = branch; \ + } else if (branch == full) { \ + return -EEXIST; \ + } \ + } while (0) + +#define ADDIP_RANGE_LOOP(map, a, a1, a2, hint, branch, full, cachep, free) \ + for (a = a1; a <= a2; a++) { \ + branch = (map)->tree[a]; \ + if (branch != full) { \ + if ((a > a1 && a < a2) || (hint)) { \ + if (branch) \ + free(branch); \ + (map)->tree[a] = full; \ + continue; \ + } else if (!branch) { \ + branch = kmem_cache_alloc(cachep, GFP_ATOMIC); \ + if (!branch) \ + return -ENOMEM; \ + memset(branch, 0, sizeof(*branch)); \ + (map)->tree[a] = branch; \ + } + +#define ADDIP_RANGE_LOOP_END() \ + } \ + } + +#define DELIP_WALK(map, elem, branch, cachep, full) \ + do { \ + branch = (map)->tree[elem]; \ + if (!branch) { \ + return -EEXIST; \ + } else if (branch == full) { \ + branch = kmem_cache_alloc(cachep, GFP_ATOMIC); \ + if (!branch) \ + return -ENOMEM; \ + memcpy(branch, full, sizeof(*full)); \ + (map)->tree[elem] = branch; \ + } \ + } while (0) + +#define DELIP_RANGE_LOOP(map, a, a1, a2, hint, branch, full, cachep, free) \ + for (a = a1; a <= a2; a++) { \ + branch = (map)->tree[a]; \ + if (branch) { \ + if ((a > a1 && a < a2) || (hint)) { \ + if (branch != full) \ + free(branch); \ + (map)->tree[a] = NULL; \ + continue; \ + } else if (branch == full) { \ + branch = kmem_cache_alloc(cachep, GFP_ATOMIC); \ + if (!branch) \ + return -ENOMEM; \ + memcpy(branch, full, sizeof(*branch)); \ + (map)->tree[a] = branch; \ + } + +#define DELIP_RANGE_LOOP_END() \ + } \ + } + +#define LOOP_WALK_BEGIN(map, i, branch) \ + for (i = 0; i < 256; i++) { \ + branch = (map)->tree[i]; \ + if (likely(!branch)) \ + continue; + +#define LOOP_WALK_END() \ + } + +#define LOOP_WALK_BEGIN_GC(map, i, branch, full, cachep, count) \ + count = -256; \ + for (i = 0; i < 256; i++) { \ + branch = (map)->tree[i]; \ + if (likely(!branch)) \ + continue; \ + count++; \ + if (branch == full) { \ + count++; \ + continue; \ + } + +#define LOOP_WALK_END_GC(map, i, branch, full, cachep, count) \ + if (-256 == count) { \ + kmem_cache_free(cachep, branch); \ + (map)->tree[i] = NULL; \ + } else if (256 == count) { \ + kmem_cache_free(cachep, branch); \ + (map)->tree[i] = full; \ + } \ + } + +#define LOOP_WALK_BEGIN_COUNT(map, i, branch, inrange, count) \ + for (i = 0; i < 256; i++) { \ + if (!(map)->tree[i]) { \ + if (inrange) { \ + count++; \ + inrange = 0; \ + } \ + continue; \ + } \ + branch = (map)->tree[i]; + +#define LOOP_WALK_END_COUNT() \ + } + +#define GETVALUE1(a, a1, b1, r) \ + (a == a1 ? b1 : r) + +#define GETVALUE2(a, b, a1, b1, c1, r) \ + (a == a1 && b == b1 ? c1 : r) + +#define GETVALUE3(a, b, c, a1, b1, c1, d1, r) \ + (a == a1 && b == b1 && c == c1 ? d1 : r) + +#define CHECK1(a, a1, a2, b1, b2, c1, c2, d1, d2) \ + ( \ + GETVALUE1(a, a1, b1, 0) == 0 \ + && GETVALUE1(a, a2, b2, 255) == 255 \ + && c1 == 0 \ + && c2 == 255 \ + && d1 == 0 \ + && d2 == 255 \ + ) + +#define CHECK2(a, b, a1, a2, b1, b2, c1, c2, d1, d2) \ + ( \ + GETVALUE2(a, b, a1, b1, c1, 0) == 0 \ + && GETVALUE2(a, b, a2, b2, c2, 255) == 255 \ + && d1 == 0 \ + && d2 == 255 \ + ) + +#define CHECK3(a, b, c, a1, a2, b1, b2, c1, c2, d1, d2) \ + ( \ + GETVALUE3(a, b, c, a1, b1, c1, d1, 0) == 0 \ + && GETVALUE3(a, b, c, a2, b2, c2, d2, 255) == 255 \ + ) + + +static inline void +free_d(struct ip_set_iptreemap_d *map) +{ + kmem_cache_free(cachep_d, map); +} + +static inline void +free_c(struct ip_set_iptreemap_c *map) +{ + struct ip_set_iptreemap_d *dtree; + unsigned int i; + + LOOP_WALK_BEGIN(map, i, dtree) { + if (dtree != fullbitmap_d) + free_d(dtree); + } LOOP_WALK_END(); + + kmem_cache_free(cachep_c, map); +} + +static inline void +free_b(struct ip_set_iptreemap_b *map) +{ + struct ip_set_iptreemap_c *ctree; + unsigned int i; + + LOOP_WALK_BEGIN(map, i, ctree) { + if (ctree != fullbitmap_c) + free_c(ctree); + } LOOP_WALK_END(); + + kmem_cache_free(cachep_b, map); +} + +static inline int +iptreemap_test(struct ip_set *set, ip_set_ip_t ip) +{ + struct ip_set_iptreemap *map = set->data; + struct ip_set_iptreemap_b *btree; + struct ip_set_iptreemap_c *ctree; + struct ip_set_iptreemap_d *dtree; + unsigned char a, b, c, d; + + ABCD(a, b, c, d, &ip); + + TESTIP_WALK(map, a, btree, fullbitmap_b); + TESTIP_WALK(btree, b, ctree, fullbitmap_c); + TESTIP_WALK(ctree, c, dtree, fullbitmap_d); + + return !!test_bit(d, (void *) dtree->bitmap); +} + +#define KADT_CONDITION + +UADT(iptreemap, test) +KADT(iptreemap, test, ipaddr) + +static inline int +__addip_single(struct ip_set *set, ip_set_ip_t ip) +{ + struct ip_set_iptreemap *map = (struct ip_set_iptreemap *) set->data; + struct ip_set_iptreemap_b *btree; + struct ip_set_iptreemap_c *ctree; + struct ip_set_iptreemap_d *dtree; + unsigned char a, b, c, d; + + ABCD(a, b, c, d, &ip); + + ADDIP_WALK(map, a, btree, struct ip_set_iptreemap_b, cachep_b, fullbitmap_b); + ADDIP_WALK(btree, b, ctree, struct ip_set_iptreemap_c, cachep_c, fullbitmap_c); + ADDIP_WALK(ctree, c, dtree, struct ip_set_iptreemap_d, cachep_d, fullbitmap_d); + + if (__test_and_set_bit(d, (void *) dtree->bitmap)) + return -EEXIST; + + __set_bit(b, (void *) btree->dirty); + + return 0; +} + +static inline int +iptreemap_add(struct ip_set *set, ip_set_ip_t start, ip_set_ip_t end) +{ + struct ip_set_iptreemap *map = set->data; + struct ip_set_iptreemap_b *btree; + struct ip_set_iptreemap_c *ctree; + struct ip_set_iptreemap_d *dtree; + unsigned int a, b, c, d; + unsigned char a1, b1, c1, d1; + unsigned char a2, b2, c2, d2; + + if (start == end) + return __addip_single(set, start); + + ABCD(a1, b1, c1, d1, &start); + ABCD(a2, b2, c2, d2, &end); + + /* This is sooo ugly... */ + ADDIP_RANGE_LOOP(map, a, a1, a2, CHECK1(a, a1, a2, b1, b2, c1, c2, d1, d2), btree, fullbitmap_b, cachep_b, free_b) { + ADDIP_RANGE_LOOP(btree, b, GETVALUE1(a, a1, b1, 0), GETVALUE1(a, a2, b2, 255), CHECK2(a, b, a1, a2, b1, b2, c1, c2, d1, d2), ctree, fullbitmap_c, cachep_c, free_c) { + ADDIP_RANGE_LOOP(ctree, c, GETVALUE2(a, b, a1, b1, c1, 0), GETVALUE2(a, b, a2, b2, c2, 255), CHECK3(a, b, c, a1, a2, b1, b2, c1, c2, d1, d2), dtree, fullbitmap_d, cachep_d, free_d) { + for (d = GETVALUE3(a, b, c, a1, b1, c1, d1, 0); d <= GETVALUE3(a, b, c, a2, b2, c2, d2, 255); d++) + __set_bit(d, (void *) dtree->bitmap); + __set_bit(b, (void *) btree->dirty); + } ADDIP_RANGE_LOOP_END(); + } ADDIP_RANGE_LOOP_END(); + } ADDIP_RANGE_LOOP_END(); + + return 0; +} + +UADT0(iptreemap, add, min(req->ip, req->end), max(req->ip, req->end)) +KADT(iptreemap, add, ipaddr, ip) + +static inline int +__delip_single(struct ip_set *set, ip_set_ip_t ip) +{ + struct ip_set_iptreemap *map = set->data; + struct ip_set_iptreemap_b *btree; + struct ip_set_iptreemap_c *ctree; + struct ip_set_iptreemap_d *dtree; + unsigned char a,b,c,d; + + ABCD(a, b, c, d, &ip); + + DELIP_WALK(map, a, btree, cachep_b, fullbitmap_b); + DELIP_WALK(btree, b, ctree, cachep_c, fullbitmap_c); + DELIP_WALK(ctree, c, dtree, cachep_d, fullbitmap_d); + + if (!__test_and_clear_bit(d, (void *) dtree->bitmap)) + return -EEXIST; + + __set_bit(b, (void *) btree->dirty); + + return 0; +} + +static inline int +iptreemap_del(struct ip_set *set, ip_set_ip_t start, ip_set_ip_t end) +{ + struct ip_set_iptreemap *map = set->data; + struct ip_set_iptreemap_b *btree; + struct ip_set_iptreemap_c *ctree; + struct ip_set_iptreemap_d *dtree; + unsigned int a, b, c, d; + unsigned char a1, b1, c1, d1; + unsigned char a2, b2, c2, d2; + + if (start == end) + return __delip_single(set, start); + + ABCD(a1, b1, c1, d1, &start); + ABCD(a2, b2, c2, d2, &end); + + /* This is sooo ugly... */ + DELIP_RANGE_LOOP(map, a, a1, a2, CHECK1(a, a1, a2, b1, b2, c1, c2, d1, d2), btree, fullbitmap_b, cachep_b, free_b) { + DELIP_RANGE_LOOP(btree, b, GETVALUE1(a, a1, b1, 0), GETVALUE1(a, a2, b2, 255), CHECK2(a, b, a1, a2, b1, b2, c1, c2, d1, d2), ctree, fullbitmap_c, cachep_c, free_c) { + DELIP_RANGE_LOOP(ctree, c, GETVALUE2(a, b, a1, b1, c1, 0), GETVALUE2(a, b, a2, b2, c2, 255), CHECK3(a, b, c, a1, a2, b1, b2, c1, c2, d1, d2), dtree, fullbitmap_d, cachep_d, free_d) { + for (d = GETVALUE3(a, b, c, a1, b1, c1, d1, 0); d <= GETVALUE3(a, b, c, a2, b2, c2, d2, 255); d++) + __clear_bit(d, (void *) dtree->bitmap); + __set_bit(b, (void *) btree->dirty); + } DELIP_RANGE_LOOP_END(); + } DELIP_RANGE_LOOP_END(); + } DELIP_RANGE_LOOP_END(); + + return 0; +} + +UADT0(iptreemap, del, min(req->ip, req->end), max(req->ip, req->end)) +KADT(iptreemap, del, ipaddr, ip) + +/* Check the status of the bitmap + * -1 == all bits cleared + * 1 == all bits set + * 0 == anything else + */ +static inline int +bitmap_status(struct ip_set_iptreemap_d *dtree) +{ + unsigned char first = dtree->bitmap[0]; + int a; + + for (a = 1; a < 32; a++) + if (dtree->bitmap[a] != first) + return 0; + + return (first == 0 ? -1 : (first == 255 ? 1 : 0)); +} + +static void +gc(unsigned long addr) +{ + struct ip_set *set = (struct ip_set *) addr; + struct ip_set_iptreemap *map = set->data; + struct ip_set_iptreemap_b *btree; + struct ip_set_iptreemap_c *ctree; + struct ip_set_iptreemap_d *dtree; + unsigned int a, b, c; + int i, j, k; + + write_lock_bh(&set->lock); + + LOOP_WALK_BEGIN_GC(map, a, btree, fullbitmap_b, cachep_b, i) { + LOOP_WALK_BEGIN_GC(btree, b, ctree, fullbitmap_c, cachep_c, j) { + if (!__test_and_clear_bit(b, (void *) btree->dirty)) + continue; + LOOP_WALK_BEGIN_GC(ctree, c, dtree, fullbitmap_d, cachep_d, k) { + switch (bitmap_status(dtree)) { + case -1: + kmem_cache_free(cachep_d, dtree); + ctree->tree[c] = NULL; + k--; + break; + case 1: + kmem_cache_free(cachep_d, dtree); + ctree->tree[c] = fullbitmap_d; + k++; + break; + } + } LOOP_WALK_END(); + } LOOP_WALK_END_GC(btree, b, ctree, fullbitmap_c, cachep_c, k); + } LOOP_WALK_END_GC(map, a, btree, fullbitmap_b, cachep_b, j); + + write_unlock_bh(&set->lock); + + map->gc.expires = jiffies + map->gc_interval * HZ; + add_timer(&map->gc); +} + +static inline void +init_gc_timer(struct ip_set *set) +{ + struct ip_set_iptreemap *map = set->data; + + init_timer(&map->gc); + map->gc.data = (unsigned long) set; + map->gc.function = gc; + map->gc.expires = jiffies + map->gc_interval * HZ; + add_timer(&map->gc); +} + +static int +iptreemap_create(struct ip_set *set, const void *data, u_int32_t size) +{ + const struct ip_set_req_iptreemap_create *req = data; + struct ip_set_iptreemap *map; + + map = kzalloc(sizeof(*map), GFP_KERNEL); + if (!map) + return -ENOMEM; + + map->gc_interval = req->gc_interval ? req->gc_interval : IPTREEMAP_DEFAULT_GC_TIME; + set->data = map; + + init_gc_timer(set); + + return 0; +} + +static inline void +__flush(struct ip_set_iptreemap *map) +{ + struct ip_set_iptreemap_b *btree; + unsigned int a; + + LOOP_WALK_BEGIN(map, a, btree); + if (btree != fullbitmap_b) + free_b(btree); + LOOP_WALK_END(); +} + +static void +iptreemap_destroy(struct ip_set *set) +{ + struct ip_set_iptreemap *map = set->data; + + while (!del_timer(&map->gc)) + msleep(IPTREEMAP_DESTROY_SLEEP); + + __flush(map); + kfree(map); + + set->data = NULL; +} + +static void +iptreemap_flush(struct ip_set *set) +{ + struct ip_set_iptreemap *map = set->data; + unsigned int gc_interval = map->gc_interval; + + while (!del_timer(&map->gc)) + msleep(IPTREEMAP_DESTROY_SLEEP); + + __flush(map); + + memset(map, 0, sizeof(*map)); + map->gc_interval = gc_interval; + + init_gc_timer(set); +} + +static void +iptreemap_list_header(const struct ip_set *set, void *data) +{ + struct ip_set_iptreemap *map = set->data; + struct ip_set_req_iptreemap_create *header = data; + + header->gc_interval = map->gc_interval; +} + +static int +iptreemap_list_members_size(const struct ip_set *set, char dont_align) +{ + struct ip_set_iptreemap *map = set->data; + struct ip_set_iptreemap_b *btree; + struct ip_set_iptreemap_c *ctree; + struct ip_set_iptreemap_d *dtree; + unsigned int a, b, c, d, inrange = 0, count = 0; + + LOOP_WALK_BEGIN_COUNT(map, a, btree, inrange, count) { + LOOP_WALK_BEGIN_COUNT(btree, b, ctree, inrange, count) { + LOOP_WALK_BEGIN_COUNT(ctree, c, dtree, inrange, count) { + for (d = 0; d < 256; d++) { + if (test_bit(d, (void *) dtree->bitmap)) { + inrange = 1; + } else if (inrange) { + count++; + inrange = 0; + } + } + } LOOP_WALK_END_COUNT(); + } LOOP_WALK_END_COUNT(); + } LOOP_WALK_END_COUNT(); + + if (inrange) + count++; + + return (count * IPSET_VALIGN(sizeof(struct ip_set_req_iptreemap), dont_align)); +} + +static inline void +add_member(void *data, size_t offset, ip_set_ip_t start, ip_set_ip_t end) +{ + struct ip_set_req_iptreemap *entry = data + offset; + + entry->ip = start; + entry->end = end; +} + +static void +iptreemap_list_members(const struct ip_set *set, void *data, char dont_align) +{ + struct ip_set_iptreemap *map = set->data; + struct ip_set_iptreemap_b *btree; + struct ip_set_iptreemap_c *ctree; + struct ip_set_iptreemap_d *dtree; + unsigned int a, b, c, d, inrange = 0; + size_t offset = 0, datasize; + ip_set_ip_t start = 0, end = 0, ip; + + datasize = IPSET_VALIGN(sizeof(struct ip_set_req_iptreemap), dont_align); + LOOP_WALK_BEGIN(map, a, btree) { + LOOP_WALK_BEGIN(btree, b, ctree) { + LOOP_WALK_BEGIN(ctree, c, dtree) { + for (d = 0; d < 256; d++) { + if (test_bit(d, (void *) dtree->bitmap)) { + ip = ((a << 24) | (b << 16) | (c << 8) | d); + if (!inrange) { + inrange = 1; + start = ip; + } else if (end < ip - 1) { + add_member(data, offset, start, end); + offset += datasize; + start = ip; + } + end = ip; + } else if (inrange) { + add_member(data, offset, start, end); + offset += datasize; + inrange = 0; + } + } + } LOOP_WALK_END(); + } LOOP_WALK_END(); + } LOOP_WALK_END(); + + if (inrange) + add_member(data, offset, start, end); +} + +IP_SET_TYPE(iptreemap, IPSET_TYPE_IP | IPSET_DATA_SINGLE) + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Sven Wegener "); +MODULE_DESCRIPTION("iptreemap type of IP sets"); + +static int __init ip_set_iptreemap_init(void) +{ + int ret = -ENOMEM; + int a; + + cachep_b = KMEM_CACHE_CREATE("ip_set_iptreemap_b", + sizeof(struct ip_set_iptreemap_b)); + if (!cachep_b) { + ip_set_printk("Unable to create ip_set_iptreemap_b slab cache"); + goto out; + } + + cachep_c = KMEM_CACHE_CREATE("ip_set_iptreemap_c", + sizeof(struct ip_set_iptreemap_c)); + if (!cachep_c) { + ip_set_printk("Unable to create ip_set_iptreemap_c slab cache"); + goto outb; + } + + cachep_d = KMEM_CACHE_CREATE("ip_set_iptreemap_d", + sizeof(struct ip_set_iptreemap_d)); + if (!cachep_d) { + ip_set_printk("Unable to create ip_set_iptreemap_d slab cache"); + goto outc; + } + + fullbitmap_d = kmem_cache_alloc(cachep_d, GFP_KERNEL); + if (!fullbitmap_d) + goto outd; + + fullbitmap_c = kmem_cache_alloc(cachep_c, GFP_KERNEL); + if (!fullbitmap_c) + goto outbitmapd; + + fullbitmap_b = kmem_cache_alloc(cachep_b, GFP_KERNEL); + if (!fullbitmap_b) + goto outbitmapc; + + ret = ip_set_register_set_type(&ip_set_iptreemap); + if (0 > ret) + goto outbitmapb; + + /* Now init our global bitmaps */ + memset(fullbitmap_d->bitmap, 0xff, sizeof(fullbitmap_d->bitmap)); + + for (a = 0; a < 256; a++) + fullbitmap_c->tree[a] = fullbitmap_d; + + for (a = 0; a < 256; a++) + fullbitmap_b->tree[a] = fullbitmap_c; + memset(fullbitmap_b->dirty, 0, sizeof(fullbitmap_b->dirty)); + + return 0; + +outbitmapb: + kmem_cache_free(cachep_b, fullbitmap_b); +outbitmapc: + kmem_cache_free(cachep_c, fullbitmap_c); +outbitmapd: + kmem_cache_free(cachep_d, fullbitmap_d); +outd: + kmem_cache_destroy(cachep_d); +outc: + kmem_cache_destroy(cachep_c); +outb: + kmem_cache_destroy(cachep_b); +out: + + return ret; +} + +static void __exit ip_set_iptreemap_fini(void) +{ + ip_set_unregister_set_type(&ip_set_iptreemap); + kmem_cache_free(cachep_d, fullbitmap_d); + kmem_cache_free(cachep_c, fullbitmap_c); + kmem_cache_free(cachep_b, fullbitmap_b); + kmem_cache_destroy(cachep_d); + kmem_cache_destroy(cachep_c); + kmem_cache_destroy(cachep_b); +} + +module_init(ip_set_iptreemap_init); +module_exit(ip_set_iptreemap_fini); diff --git a/net/ipv4/netfilter/ip_set_macipmap.c b/net/ipv4/netfilter/ip_set_macipmap.c new file mode 100644 index 0000000..89e907b --- /dev/null +++ b/net/ipv4/netfilter/ip_set_macipmap.c @@ -0,0 +1,179 @@ +/* Copyright (C) 2000-2002 Joakim Axelsson + * Patrick Schaaf + * Martin Josefsson + * Copyright (C) 2003-2008 Jozsef Kadlecsik + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +/* Kernel module implementing an IP set type: the macipmap type */ + +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +static int +macipmap_utest(struct ip_set *set, const void *data, u_int32_t size) +{ + const struct ip_set_macipmap *map = set->data; + const struct ip_set_macip *table = map->members; + const struct ip_set_req_macipmap *req = data; + + if (req->ip < map->first_ip || req->ip > map->last_ip) + return -ERANGE; + + DP("set: %s, ip:%u.%u.%u.%u", set->name, HIPQUAD(req->ip)); + if (table[req->ip - map->first_ip].match) { + return (memcmp(req->ethernet, + &table[req->ip - map->first_ip].ethernet, + ETH_ALEN) == 0); + } else { + return (map->flags & IPSET_MACIP_MATCHUNSET ? 1 : 0); + } +} + +static int +macipmap_ktest(struct ip_set *set, + const struct sk_buff *skb, + const u_int32_t *flags) +{ + const struct ip_set_macipmap *map = set->data; + const struct ip_set_macip *table = map->members; + ip_set_ip_t ip; + + ip = ipaddr(skb, flags); + + if (ip < map->first_ip || ip > map->last_ip) + return 0; + + DP("set: %s, ip:%u.%u.%u.%u", set->name, HIPQUAD(ip)); + if (table[ip - map->first_ip].match) { + /* Is mac pointer valid? + * If so, compare... */ + return (skb_mac_header(skb) >= skb->head + && (skb_mac_header(skb) + ETH_HLEN) <= skb->data + && (memcmp(eth_hdr(skb)->h_source, + &table[ip - map->first_ip].ethernet, + ETH_ALEN) == 0)); + } else { + return (map->flags & IPSET_MACIP_MATCHUNSET ? 1 : 0); + } +} + +/* returns 0 on success */ +static inline int +macipmap_add(struct ip_set *set, + ip_set_ip_t ip, const unsigned char *ethernet) +{ + struct ip_set_macipmap *map = set->data; + struct ip_set_macip *table = map->members; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + if (table[ip - map->first_ip].match) + return -EEXIST; + + DP("set: %s, ip: %u.%u.%u.%u", set->name, HIPQUAD(ip)); + memcpy(&table[ip - map->first_ip].ethernet, ethernet, ETH_ALEN); + table[ip - map->first_ip].match = IPSET_MACIP_ISSET; + return 0; +} + +#define KADT_CONDITION \ + if (!(skb_mac_header(skb) >= skb->head \ + && (skb_mac_header(skb) + ETH_HLEN) <= skb->data))\ + return -EINVAL; + +UADT(macipmap, add, req->ethernet) +KADT(macipmap, add, ipaddr, eth_hdr(skb)->h_source) + +static inline int +macipmap_del(struct ip_set *set, ip_set_ip_t ip) +{ + struct ip_set_macipmap *map = set->data; + struct ip_set_macip *table = map->members; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + if (!table[ip - map->first_ip].match) + return -EEXIST; + + table[ip - map->first_ip].match = 0; + DP("set: %s, ip: %u.%u.%u.%u", set->name, HIPQUAD(ip)); + return 0; +} + +#undef KADT_CONDITION +#define KADT_CONDITION + +UADT(macipmap, del) +KADT(macipmap, del, ipaddr) + +static inline int +__macipmap_create(const struct ip_set_req_macipmap_create *req, + struct ip_set_macipmap *map) +{ + if (req->to - req->from > MAX_RANGE) { + ip_set_printk("range too big, %d elements (max %d)", + req->to - req->from + 1, MAX_RANGE+1); + return -ENOEXEC; + } + map->flags = req->flags; + return (req->to - req->from + 1) * sizeof(struct ip_set_macip); +} + +BITMAP_CREATE(macipmap) +BITMAP_DESTROY(macipmap) +BITMAP_FLUSH(macipmap) + +static inline void +__macipmap_list_header(const struct ip_set_macipmap *map, + struct ip_set_req_macipmap_create *header) +{ + header->flags = map->flags; +} + +BITMAP_LIST_HEADER(macipmap) +BITMAP_LIST_MEMBERS_SIZE(macipmap, struct ip_set_req_macipmap, + (map->last_ip - map->first_ip + 1), + ((const struct ip_set_macip *)map->members)[i].match) + + +static void +macipmap_list_members(const struct ip_set *set, void *data, char dont_align) +{ + const struct ip_set_macipmap *map = set->data; + const struct ip_set_macip *table = map->members; + uint32_t i, n = 0; + struct ip_set_req_macipmap *d; + + if (dont_align) { + memcpy(data, map->members, map->size); + return; + } + + for (i = 0; i < map->last_ip - map->first_ip + 1; i++) + if (table[i].match) { + d = data + n * IPSET_ALIGN(sizeof(struct ip_set_req_macipmap)); + d->ip = map->first_ip + i; + memcpy(d->ethernet, &table[i].ethernet, ETH_ALEN); + n++; + } +} + +IP_SET_TYPE(macipmap, IPSET_TYPE_IP | IPSET_DATA_SINGLE) + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("macipmap type of IP sets"); + +REGISTER_MODULE(macipmap) diff --git a/net/ipv4/netfilter/ip_set_nethash.c b/net/ipv4/netfilter/ip_set_nethash.c new file mode 100644 index 0000000..e3b09e0 --- /dev/null +++ b/net/ipv4/netfilter/ip_set_nethash.c @@ -0,0 +1,218 @@ +/* Copyright (C) 2003-2008 Jozsef Kadlecsik + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +/* Kernel module implementing a cidr nethash set */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +#include + +static int limit = MAX_RANGE; + +static inline __u32 +nethash_id_cidr(const struct ip_set_nethash *map, + ip_set_ip_t ip, + uint8_t cidr) +{ + __u32 id; + u_int16_t i; + ip_set_ip_t *elem; + + ip = pack_ip_cidr(ip, cidr); + if (!ip) + return MAX_RANGE; + + for (i = 0; i < map->probes; i++) { + id = jhash_ip(map, i, ip) % map->hashsize; + DP("hash key: %u", id); + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); + if (*elem == ip) + return id; + /* No shortcut - there can be deleted entries. */ + } + return UINT_MAX; +} + +static inline __u32 +nethash_id(struct ip_set *set, ip_set_ip_t ip) +{ + const struct ip_set_nethash *map = set->data; + __u32 id = UINT_MAX; + int i; + + for (i = 0; i < 30 && map->cidr[i]; i++) { + id = nethash_id_cidr(map, ip, map->cidr[i]); + if (id != UINT_MAX) + break; + } + return id; +} + +static inline int +nethash_test_cidr(struct ip_set *set, ip_set_ip_t ip, uint8_t cidr) +{ + const struct ip_set_nethash *map = set->data; + + return (nethash_id_cidr(map, ip, cidr) != UINT_MAX); +} + +static inline int +nethash_test(struct ip_set *set, ip_set_ip_t ip) +{ + return (nethash_id(set, ip) != UINT_MAX); +} + +static int +nethash_utest(struct ip_set *set, const void *data, u_int32_t size) +{ + const struct ip_set_req_nethash *req = data; + + if (req->cidr <= 0 || req->cidr > 32) + return -EINVAL; + return (req->cidr == 32 ? nethash_test(set, req->ip) + : nethash_test_cidr(set, req->ip, req->cidr)); +} + +#define KADT_CONDITION + +KADT(nethash, test, ipaddr) + +static inline int +__nethash_add(struct ip_set_nethash *map, ip_set_ip_t *ip) +{ + __u32 probe; + u_int16_t i; + ip_set_ip_t *elem, *slot = NULL; + + for (i = 0; i < map->probes; i++) { + probe = jhash_ip(map, i, *ip) % map->hashsize; + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, probe); + if (*elem == *ip) + return -EEXIST; + if (!(slot || *elem)) + slot = elem; + /* There can be deleted entries, must check all slots */ + } + if (slot) { + *slot = *ip; + map->elements++; + return 0; + } + /* Trigger rehashing */ + return -EAGAIN; +} + +static inline int +nethash_add(struct ip_set *set, ip_set_ip_t ip, uint8_t cidr) +{ + struct ip_set_nethash *map = set->data; + int ret; + + if (map->elements >= limit || map->nets[cidr-1] == UINT16_MAX) + return -ERANGE; + if (cidr <= 0 || cidr >= 32) + return -EINVAL; + + ip = pack_ip_cidr(ip, cidr); + if (!ip) + return -ERANGE; + + ret = __nethash_add(map, &ip); + if (ret == 0) { + if (!map->nets[cidr-1]++) + add_cidr_size(map->cidr, cidr); + } + + return ret; +} + +#undef KADT_CONDITION +#define KADT_CONDITION \ + struct ip_set_nethash *map = set->data; \ + uint8_t cidr = map->cidr[0] ? map->cidr[0] : 31; + +UADT(nethash, add, req->cidr) +KADT(nethash, add, ipaddr, cidr) + +static inline void +__nethash_retry(struct ip_set_nethash *tmp, struct ip_set_nethash *map) +{ + memcpy(tmp->cidr, map->cidr, sizeof(tmp->cidr)); + memcpy(tmp->nets, map->nets, sizeof(tmp->nets)); +} + +HASH_RETRY(nethash, ip_set_ip_t) + +static inline int +nethash_del(struct ip_set *set, ip_set_ip_t ip, uint8_t cidr) +{ + struct ip_set_nethash *map = set->data; + ip_set_ip_t id, *elem; + + if (cidr <= 0 || cidr >= 32) + return -EINVAL; + + id = nethash_id_cidr(map, ip, cidr); + if (id == UINT_MAX) + return -EEXIST; + + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); + *elem = 0; + map->elements--; + if (!map->nets[cidr-1]--) + del_cidr_size(map->cidr, cidr); + return 0; +} + +UADT(nethash, del, req->cidr) +KADT(nethash, del, ipaddr, cidr) + +static inline int +__nethash_create(const struct ip_set_req_nethash_create *req, + struct ip_set_nethash *map) +{ + memset(map->cidr, 0, sizeof(map->cidr)); + memset(map->nets, 0, sizeof(map->nets)); + + return 0; +} + +HASH_CREATE(nethash, ip_set_ip_t) +HASH_DESTROY(nethash) + +HASH_FLUSH_CIDR(nethash, ip_set_ip_t) + +static inline void +__nethash_list_header(const struct ip_set_nethash *map, + struct ip_set_req_nethash_create *header) +{ +} + +HASH_LIST_HEADER(nethash) +HASH_LIST_MEMBERS_SIZE(nethash, ip_set_ip_t) +HASH_LIST_MEMBERS(nethash, ip_set_ip_t) + +IP_SET_RTYPE(nethash, IPSET_TYPE_IP | IPSET_DATA_SINGLE) + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("nethash type of IP sets"); +module_param(limit, int, 0600); +MODULE_PARM_DESC(limit, "maximal number of elements stored in the sets"); + +REGISTER_MODULE(nethash) diff --git a/net/ipv4/netfilter/ip_set_portmap.c b/net/ipv4/netfilter/ip_set_portmap.c new file mode 100644 index 0000000..8bb6e76 --- /dev/null +++ b/net/ipv4/netfilter/ip_set_portmap.c @@ -0,0 +1,130 @@ +/* Copyright (C) 2003-2008 Jozsef Kadlecsik + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +/* Kernel module implementing a port set type as a bitmap */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include + +#include +#include + +static inline int +portmap_test(const struct ip_set *set, ip_set_ip_t port) +{ + const struct ip_set_portmap *map = set->data; + + if (port < map->first_ip || port > map->last_ip) + return -ERANGE; + + DP("set: %s, port: %u", set->name, port); + return !!test_bit(port - map->first_ip, map->members); +} + +#define KADT_CONDITION \ + if (ip == INVALID_PORT) \ + return 0; + +UADT(portmap, test) +KADT(portmap, test, get_port) + +static inline int +portmap_add(struct ip_set *set, ip_set_ip_t port) +{ + struct ip_set_portmap *map = set->data; + + if (port < map->first_ip || port > map->last_ip) + return -ERANGE; + if (test_and_set_bit(port - map->first_ip, map->members)) + return -EEXIST; + + DP("set: %s, port %u", set->name, port); + return 0; +} + +UADT(portmap, add) +KADT(portmap, add, get_port) + +static inline int +portmap_del(struct ip_set *set, ip_set_ip_t port) +{ + struct ip_set_portmap *map = set->data; + + if (port < map->first_ip || port > map->last_ip) + return -ERANGE; + if (!test_and_clear_bit(port - map->first_ip, map->members)) + return -EEXIST; + + DP("set: %s, port %u", set->name, port); + return 0; +} + +UADT(portmap, del) +KADT(portmap, del, get_port) + +static inline int +__portmap_create(const struct ip_set_req_portmap_create *req, + struct ip_set_portmap *map) +{ + if (req->to - req->from > MAX_RANGE) { + ip_set_printk("range too big, %d elements (max %d)", + req->to - req->from + 1, MAX_RANGE+1); + return -ENOEXEC; + } + return bitmap_bytes(req->from, req->to); +} + +BITMAP_CREATE(portmap) +BITMAP_DESTROY(portmap) +BITMAP_FLUSH(portmap) + +static inline void +__portmap_list_header(const struct ip_set_portmap *map, + struct ip_set_req_portmap_create *header) +{ +} + +BITMAP_LIST_HEADER(portmap) +BITMAP_LIST_MEMBERS_SIZE(portmap, ip_set_ip_t, (map->last_ip - map->first_ip + 1), + test_bit(i, map->members)) + +static void +portmap_list_members(const struct ip_set *set, void *data, char dont_align) +{ + const struct ip_set_portmap *map = set->data; + uint32_t i, n = 0; + ip_set_ip_t *d; + + if (dont_align) { + memcpy(data, map->members, map->size); + return; + } + + for (i = 0; i < map->last_ip - map->first_ip + 1; i++) + if (test_bit(i, map->members)) { + d = data + n * IPSET_ALIGN(sizeof(ip_set_ip_t)); + *d = map->first_ip + i; + n++; + } +} + +IP_SET_TYPE(portmap, IPSET_TYPE_PORT | IPSET_DATA_SINGLE) + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("portmap type of IP sets"); + +REGISTER_MODULE(portmap) diff --git a/net/ipv4/netfilter/ip_set_setlist.c b/net/ipv4/netfilter/ip_set_setlist.c new file mode 100644 index 0000000..3cfdae8 --- /dev/null +++ b/net/ipv4/netfilter/ip_set_setlist.c @@ -0,0 +1,324 @@ +/* Copyright (C) 2008 Jozsef Kadlecsik + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +/* Kernel module implementing an IP set type: the setlist type */ + +#include +#include +#include +#include + +#include +#include +#include + +/* + * before ==> index, ref + * after ==> ref, index + */ + +static inline int +next_index_eq(const struct ip_set_setlist *map, int i, ip_set_id_t index) +{ + return i < map->size && map->index[i] == index; +} + +static int +setlist_utest(struct ip_set *set, const void *data, u_int32_t size) +{ + const struct ip_set_setlist *map = set->data; + const struct ip_set_req_setlist *req = data; + ip_set_id_t index, ref = IP_SET_INVALID_ID; + int i, res = 0; + struct ip_set *s; + + if (req->before && req->ref[0] == '\0') + return 0; + + index = __ip_set_get_byname(req->name, &s); + if (index == IP_SET_INVALID_ID) + return 0; + if (req->ref[0] != '\0') { + ref = __ip_set_get_byname(req->ref, &s); + if (ref == IP_SET_INVALID_ID) + goto finish; + } + for (i = 0; i < map->size + && map->index[i] != IP_SET_INVALID_ID; i++) { + if (req->before && map->index[i] == index) { + res = next_index_eq(map, i + 1, ref); + break; + } else if (!req->before) { + if ((ref == IP_SET_INVALID_ID + && map->index[i] == index) + || (map->index[i] == ref + && next_index_eq(map, i + 1, index))) { + res = 1; + break; + } + } + } + if (ref != IP_SET_INVALID_ID) + __ip_set_put_byindex(ref); +finish: + __ip_set_put_byindex(index); + return res; +} + +static int +setlist_ktest(struct ip_set *set, + const struct sk_buff *skb, + const u_int32_t *flags) +{ + struct ip_set_setlist *map = set->data; + int i, res = 0; + + for (i = 0; i < map->size + && map->index[i] != IP_SET_INVALID_ID + && res == 0; i++) + res = ip_set_testip_kernel(map->index[i], skb, flags); + return res; +} + +static inline int +insert_setlist(struct ip_set_setlist *map, int i, ip_set_id_t index) +{ + ip_set_id_t tmp; + int j; + + DP("i: %u, last %u\n", i, map->index[map->size - 1]); + if (i >= map->size || map->index[map->size - 1] != IP_SET_INVALID_ID) + return -ERANGE; + + for (j = i; j < map->size + && index != IP_SET_INVALID_ID; j++) { + tmp = map->index[j]; + map->index[j] = index; + index = tmp; + } + return 0; +} + +static int +setlist_uadd(struct ip_set *set, const void *data, u_int32_t size) +{ + struct ip_set_setlist *map = set->data; + const struct ip_set_req_setlist *req = data; + ip_set_id_t index, ref = IP_SET_INVALID_ID; + int i, res = -ERANGE; + struct ip_set *s; + + if (req->before && req->ref[0] == '\0') + return -EINVAL; + + index = __ip_set_get_byname(req->name, &s); + if (index == IP_SET_INVALID_ID) + return -EEXIST; + /* "Loop detection" */ + if (strcmp(s->type->typename, "setlist") == 0) + goto finish; + + if (req->ref[0] != '\0') { + ref = __ip_set_get_byname(req->ref, &s); + if (ref == IP_SET_INVALID_ID) { + res = -EEXIST; + goto finish; + } + } + for (i = 0; i < map->size; i++) { + if (map->index[i] != ref) + continue; + if (req->before) + res = insert_setlist(map, i, index); + else + res = insert_setlist(map, + ref == IP_SET_INVALID_ID ? i : i + 1, + index); + break; + } + if (ref != IP_SET_INVALID_ID) + __ip_set_put_byindex(ref); + /* In case of success, we keep the reference to the set */ +finish: + if (res != 0) + __ip_set_put_byindex(index); + return res; +} + +static int +setlist_kadd(struct ip_set *set, + const struct sk_buff *skb, + const u_int32_t *flags) +{ + struct ip_set_setlist *map = set->data; + int i, res = -EINVAL; + + for (i = 0; i < map->size + && map->index[i] != IP_SET_INVALID_ID + && res != 0; i++) + res = ip_set_addip_kernel(map->index[i], skb, flags); + return res; +} + +static inline int +unshift_setlist(struct ip_set_setlist *map, int i) +{ + int j; + + for (j = i; j < map->size - 1; j++) + map->index[j] = map->index[j+1]; + map->index[map->size-1] = IP_SET_INVALID_ID; + return 0; +} + +static int +setlist_udel(struct ip_set *set, const void *data, u_int32_t size) +{ + struct ip_set_setlist *map = set->data; + const struct ip_set_req_setlist *req = data; + ip_set_id_t index, ref = IP_SET_INVALID_ID; + int i, res = -EEXIST; + struct ip_set *s; + + if (req->before && req->ref[0] == '\0') + return -EINVAL; + + index = __ip_set_get_byname(req->name, &s); + if (index == IP_SET_INVALID_ID) + return -EEXIST; + if (req->ref[0] != '\0') { + ref = __ip_set_get_byname(req->ref, &s); + if (ref == IP_SET_INVALID_ID) + goto finish; + } + for (i = 0; i < map->size + && map->index[i] != IP_SET_INVALID_ID; i++) { + if (req->before) { + if (map->index[i] == index + && next_index_eq(map, i + 1, ref)) { + res = unshift_setlist(map, i); + break; + } + } else if (ref == IP_SET_INVALID_ID) { + if (map->index[i] == index) { + res = unshift_setlist(map, i); + break; + } + } else if (map->index[i] == ref + && next_index_eq(map, i + 1, index)) { + res = unshift_setlist(map, i + 1); + break; + } + } + if (ref != IP_SET_INVALID_ID) + __ip_set_put_byindex(ref); +finish: + __ip_set_put_byindex(index); + /* In case of success, release the reference to the set */ + if (res == 0) + __ip_set_put_byindex(index); + return res; +} + +static int +setlist_kdel(struct ip_set *set, + const struct sk_buff *skb, + const u_int32_t *flags) +{ + struct ip_set_setlist *map = set->data; + int i, res = -EINVAL; + + for (i = 0; i < map->size + && map->index[i] != IP_SET_INVALID_ID + && res != 0; i++) + res = ip_set_delip_kernel(map->index[i], skb, flags); + return res; +} + +static int +setlist_create(struct ip_set *set, const void *data, u_int32_t size) +{ + struct ip_set_setlist *map; + const struct ip_set_req_setlist_create *req = data; + int i; + + map = kmalloc(sizeof(struct ip_set_setlist) + + req->size * sizeof(ip_set_id_t), GFP_KERNEL); + if (!map) + return -ENOMEM; + map->size = req->size; + for (i = 0; i < map->size; i++) + map->index[i] = IP_SET_INVALID_ID; + + set->data = map; + return 0; +} + +static void +setlist_destroy(struct ip_set *set) +{ + struct ip_set_setlist *map = set->data; + int i; + + for (i = 0; i < map->size + && map->index[i] != IP_SET_INVALID_ID; i++) + __ip_set_put_byindex(map->index[i]); + + kfree(map); + set->data = NULL; +} + +static void +setlist_flush(struct ip_set *set) +{ + struct ip_set_setlist *map = set->data; + int i; + + for (i = 0; i < map->size + && map->index[i] != IP_SET_INVALID_ID; i++) { + __ip_set_put_byindex(map->index[i]); + map->index[i] = IP_SET_INVALID_ID; + } +} + +static void +setlist_list_header(const struct ip_set *set, void *data) +{ + const struct ip_set_setlist *map = set->data; + struct ip_set_req_setlist_create *header = data; + + header->size = map->size; +} + +static int +setlist_list_members_size(const struct ip_set *set, char dont_align) +{ + const struct ip_set_setlist *map = set->data; + + return map->size * IPSET_VALIGN(sizeof(ip_set_id_t), dont_align); +} + +static void +setlist_list_members(const struct ip_set *set, void *data, char dont_align) +{ + struct ip_set_setlist *map = set->data; + ip_set_id_t *d; + int i; + + for (i = 0; i < map->size; i++) { + d = data + i * IPSET_VALIGN(sizeof(ip_set_id_t), dont_align); + *d = ip_set_id(map->index[i]); + } +} + +IP_SET_TYPE(setlist, IPSET_TYPE_SETNAME | IPSET_DATA_SINGLE) + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("setlist type of IP sets"); + +REGISTER_MODULE(setlist) diff --git a/net/ipv4/netfilter/ipt_SET.c b/net/ipv4/netfilter/ipt_SET.c new file mode 100644 index 0000000..952b5d9 --- /dev/null +++ b/net/ipv4/netfilter/ipt_SET.c @@ -0,0 +1,256 @@ +/* Copyright (C) 2000-2002 Joakim Axelsson + * Patrick Schaaf + * Martin Josefsson + * Copyright (C) 2003-2004 Jozsef Kadlecsik + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +/* ipt_SET.c - netfilter target to manipulate IP sets */ + +#include +#include +#include +#include + +#include +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,16) +#include +#define xt_register_target ipt_register_target +#define xt_unregister_target ipt_unregister_target +#define xt_target ipt_target +#define XT_CONTINUE IPT_CONTINUE +#else +#include +#endif +#include + +static unsigned int +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) +target(struct sk_buff **pskb, + unsigned int hooknum, + const struct net_device *in, + const struct net_device *out, + const void *targinfo, + void *userinfo) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17) +target(struct sk_buff **pskb, + const struct net_device *in, + const struct net_device *out, + unsigned int hooknum, + const void *targinfo, + void *userinfo) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) +target(struct sk_buff **pskb, + const struct net_device *in, + const struct net_device *out, + unsigned int hooknum, + const struct xt_target *target, + const void *targinfo, + void *userinfo) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,24) +target(struct sk_buff **pskb, + const struct net_device *in, + const struct net_device *out, + unsigned int hooknum, + const struct xt_target *target, + const void *targinfo) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,28) +target(struct sk_buff *skb, + const struct net_device *in, + const struct net_device *out, + unsigned int hooknum, + const struct xt_target *target, + const void *targinfo) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,35) +target(struct sk_buff *skb, + const struct xt_target_param *par) +#else /* LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35) */ +target(struct sk_buff *skb, + const struct xt_action_param *par) +#endif +{ +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,28) + const struct ipt_set_info_target *info = targinfo; +#else + const struct ipt_set_info_target *info = par->targinfo; +#endif +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,24) + struct sk_buff *skb = *pskb; +#endif + + + if (info->add_set.index != IP_SET_INVALID_ID) + ip_set_addip_kernel(info->add_set.index, + skb, + info->add_set.flags); + if (info->del_set.index != IP_SET_INVALID_ID) + ip_set_delip_kernel(info->del_set.index, + skb, + info->del_set.flags); + + return XT_CONTINUE; +} + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,35) +#define CHECK_OK 1 +#define CHECK_FAIL 0 +#else /* LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35) */ +#define CHECK_OK 0 +#define CHECK_FAIL -EINVAL +#endif + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,16) +static int +checkentry(const char *tablename, + const struct ipt_entry *e, + void *targinfo, + unsigned int targinfosize, + unsigned int hook_mask) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17) +static int +checkentry(const char *tablename, + const void *e, + void *targinfo, + unsigned int targinfosize, + unsigned int hook_mask) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) +static int +checkentry(const char *tablename, + const void *e, + const struct xt_target *target, + void *targinfo, + unsigned int targinfosize, + unsigned int hook_mask) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23) +static int +checkentry(const char *tablename, + const void *e, + const struct xt_target *target, + void *targinfo, + unsigned int hook_mask) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,28) +static bool +checkentry(const char *tablename, + const void *e, + const struct xt_target *target, + void *targinfo, + unsigned int hook_mask) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,35) +static bool +checkentry(const struct xt_tgchk_param *par) +#else /* LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35) */ +static int +checkentry(const struct xt_tgchk_param *par) +#endif +{ +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,28) + const struct ipt_set_info_target *info = targinfo; +#else + const struct ipt_set_info_target *info = par->targinfo; +#endif + ip_set_id_t index; + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17) + if (targinfosize != IPT_ALIGN(sizeof(*info))) { + DP("bad target info size %u", targinfosize); + return CHECK_FAIL; + } +#endif + + if (info->add_set.index != IP_SET_INVALID_ID) { + index = ip_set_get_byindex(info->add_set.index); + if (index == IP_SET_INVALID_ID) { + ip_set_printk("cannot find add_set index %u as target", + info->add_set.index); + return CHECK_FAIL; /* error */ + } + } + + if (info->del_set.index != IP_SET_INVALID_ID) { + index = ip_set_get_byindex(info->del_set.index); + if (index == IP_SET_INVALID_ID) { + ip_set_printk("cannot find del_set index %u as target", + info->del_set.index); + return CHECK_FAIL; /* error */ + } + } + if (info->add_set.flags[IP_SET_MAX_BINDINGS] != 0 + || info->del_set.flags[IP_SET_MAX_BINDINGS] != 0) { + ip_set_printk("That's nasty!"); + return CHECK_FAIL; /* error */ + } + + return CHECK_OK; +} + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17) +static void destroy(void *targetinfo, + unsigned int targetsize) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) +static void destroy(const struct xt_target *target, + void *targetinfo, + unsigned int targetsize) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,28) +static void destroy(const struct xt_target *target, + void *targetinfo) +#else /* LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,28) */ +static void destroy(const struct xt_tgdtor_param *par) +#endif +{ +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,28) + const struct ipt_set_info_target *info = targetinfo; +#else + const struct ipt_set_info_target *info = par->targinfo; +#endif + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17) + if (targetsize != IPT_ALIGN(sizeof(struct ipt_set_info_target))) { + ip_set_printk("invalid targetsize %d", targetsize); + return; + } +#endif + if (info->add_set.index != IP_SET_INVALID_ID) + ip_set_put_byindex(info->add_set.index); + if (info->del_set.index != IP_SET_INVALID_ID) + ip_set_put_byindex(info->del_set.index); +} + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17) +static struct xt_target SET_target = { + .name = "SET", + .target = target, + .checkentry = checkentry, + .destroy = destroy, + .me = THIS_MODULE +}; +#else /* LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) */ +static struct xt_target SET_target = { + .name = "SET", + .family = AF_INET, + .target = target, + .targetsize = sizeof(struct ipt_set_info_target), + .checkentry = checkentry, + .destroy = destroy, + .me = THIS_MODULE +}; +#endif + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("iptables IP set target module"); + +static int __init ipt_SET_init(void) +{ + return xt_register_target(&SET_target); +} + +static void __exit ipt_SET_fini(void) +{ + xt_unregister_target(&SET_target); +} + +module_init(ipt_SET_init); +module_exit(ipt_SET_fini); diff --git a/net/ipv4/netfilter/ipt_set.c b/net/ipv4/netfilter/ipt_set.c new file mode 100644 index 0000000..3c661be --- /dev/null +++ b/net/ipv4/netfilter/ipt_set.c @@ -0,0 +1,253 @@ +/* Copyright (C) 2000-2002 Joakim Axelsson + * Patrick Schaaf + * Martin Josefsson + * Copyright (C) 2003-2004 Jozsef Kadlecsik + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +/* Kernel module to match an IP set. */ + +#include +#include +#include +#include + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,16) +#include +#define xt_register_match ipt_register_match +#define xt_unregister_match ipt_unregister_match +#define xt_match ipt_match +#else +#include +#endif +#include +#include + +static inline int +match_set(const struct ipt_set_info *info, + const struct sk_buff *skb, + int inv) +{ + if (ip_set_testip_kernel(info->index, skb, info->flags)) + inv = !inv; + return inv; +} + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,0) +static int +match(const struct sk_buff *skb, + const struct net_device *in, + const struct net_device *out, + const void *matchinfo, + int offset, + const void *hdr, + u_int16_t datalen, + int *hotdrop) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,16) +static int +match(const struct sk_buff *skb, + const struct net_device *in, + const struct net_device *out, + const void *matchinfo, + int offset, + int *hotdrop) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17) +static int +match(const struct sk_buff *skb, + const struct net_device *in, + const struct net_device *out, + const void *matchinfo, + int offset, + unsigned int protoff, + int *hotdrop) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23) +static int +match(const struct sk_buff *skb, + const struct net_device *in, + const struct net_device *out, + const struct xt_match *match, + const void *matchinfo, + int offset, + unsigned int protoff, + int *hotdrop) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,28) +static bool +match(const struct sk_buff *skb, + const struct net_device *in, + const struct net_device *out, + const struct xt_match *match, + const void *matchinfo, + int offset, + unsigned int protoff, + bool *hotdrop) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,35) +static bool +match(const struct sk_buff *skb, + const struct xt_match_param *par) +#else /* LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35) */ +static bool +match(const struct sk_buff *skb, + struct xt_action_param *par) +#endif +{ +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,28) + const struct ipt_set_info_match *info = matchinfo; +#else + const struct ipt_set_info_match *info = par->matchinfo; +#endif + + return match_set(&info->match_set, + skb, + info->match_set.flags[0] & IPSET_MATCH_INV); +} + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,35) +#define CHECK_OK 1 +#define CHECK_FAIL 0 +#else /* LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35) */ +#define CHECK_OK 0 +#define CHECK_FAIL -EINVAL +#endif + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,16) +static int +checkentry(const char *tablename, + const struct ipt_ip *ip, + void *matchinfo, + unsigned int matchsize, + unsigned int hook_mask) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17) +static int +checkentry(const char *tablename, + const void *inf, + void *matchinfo, + unsigned int matchsize, + unsigned int hook_mask) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) +static int +checkentry(const char *tablename, + const void *inf, + const struct xt_match *match, + void *matchinfo, + unsigned int matchsize, + unsigned int hook_mask) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,23) +static int +checkentry(const char *tablename, + const void *inf, + const struct xt_match *match, + void *matchinfo, + unsigned int hook_mask) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,28) +static bool +checkentry(const char *tablename, + const void *inf, + const struct xt_match *match, + void *matchinfo, + unsigned int hook_mask) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,35) +static bool +checkentry(const struct xt_mtchk_param *par) +#else /* LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,35) */ +static int +checkentry(const struct xt_mtchk_param *par) +#endif +{ +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,28) + struct ipt_set_info_match *info = matchinfo; +#else + struct ipt_set_info_match *info = par->matchinfo; +#endif + ip_set_id_t index; + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17) + if (matchsize != IPT_ALIGN(sizeof(struct ipt_set_info_match))) { + ip_set_printk("invalid matchsize %d", matchsize); + return CHECK_FAIL; + } +#endif + + index = ip_set_get_byindex(info->match_set.index); + + if (index == IP_SET_INVALID_ID) { + ip_set_printk("Cannot find set indentified by id %u to match", + info->match_set.index); + return CHECK_FAIL; /* error */ + } + if (info->match_set.flags[IP_SET_MAX_BINDINGS] != 0) { + ip_set_printk("That's nasty!"); + return CHECK_FAIL; /* error */ + } + + return CHECK_OK; +} + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17) +static void destroy(void *matchinfo, + unsigned int matchsize) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) +static void destroy(const struct xt_match *match, + void *matchinfo, + unsigned int matchsize) +#elif LINUX_VERSION_CODE < KERNEL_VERSION(2,6,28) +static void destroy(const struct xt_match *match, + void *matchinfo) +#else /* LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,28) */ +static void destroy(const struct xt_mtdtor_param *par) +#endif +{ +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,28) + struct ipt_set_info_match *info = matchinfo; +#else + struct ipt_set_info_match *info = par->matchinfo; +#endif + + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17) + if (matchsize != IPT_ALIGN(sizeof(struct ipt_set_info_match))) { + ip_set_printk("invalid matchsize %d", matchsize); + return; + } +#endif + ip_set_put_byindex(info->match_set.index); +} + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,17) +static struct xt_match set_match = { + .name = "set", + .match = &match, + .checkentry = &checkentry, + .destroy = &destroy, + .me = THIS_MODULE +}; +#else /* LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) */ +static struct xt_match set_match = { + .name = "set", + .family = AF_INET, + .match = &match, + .matchsize = sizeof(struct ipt_set_info_match), + .checkentry = &checkentry, + .destroy = &destroy, + .me = THIS_MODULE +}; +#endif + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("iptables IP set match module"); + +static int __init ipt_ipset_init(void) +{ + return xt_register_match(&set_match); +} + +static void __exit ipt_ipset_fini(void) +{ + xt_unregister_match(&set_match); +} + +module_init(ipt_ipset_init); +module_exit(ipt_ipset_fini);