diff --git a/.gitignore b/.gitignore index ded564f..37ce316 100644 --- a/.gitignore +++ b/.gitignore @@ -6,6 +6,7 @@ .libs Makefile Makefile.in +GNUmakefile /downloads diff --git a/Makefile.am b/Makefile.am index 425c3a2..29be2c3 100644 --- a/Makefile.am +++ b/Makefile.am @@ -1,7 +1,7 @@ # -*- Makefile -*- AUTOMAKE_OPTIONS = foreign subdir-objects -SUBDIRS = extensions +SUBDIRS = extensions extensions/ipset man_MANS := xtables-addons.8 diff --git a/configure.ac b/configure.ac index ee9e082..10f7234 100644 --- a/configure.ac +++ b/configure.ac @@ -65,4 +65,4 @@ AC_SUBST([kinclude_CFLAGS]) AC_SUBST([kbuilddir]) AC_SUBST([ksourcedir]) AC_SUBST([xtlibdir]) -AC_OUTPUT([Makefile extensions/GNUmakefile]) +AC_OUTPUT([Makefile extensions/GNUmakefile extensions/ipset/GNUmakefile]) diff --git a/extensions/.gitignore b/extensions/.gitignore index 254653f..91ce60c 100644 --- a/extensions/.gitignore +++ b/extensions/.gitignore @@ -8,7 +8,6 @@ modules.order /*.so /*.oo -/GNUmakefile /matches.man /targets.man /.manpages.lst diff --git a/extensions/Kbuild b/extensions/Kbuild index 540699b..e5c01e4 100644 --- a/extensions/Kbuild +++ b/extensions/Kbuild @@ -15,6 +15,7 @@ obj-${build_TEE} += xt_TEE.o obj-${build_condition} += xt_condition.o obj-${build_geoip} += xt_geoip.o obj-${build_ipp2p} += xt_ipp2p.o +obj-${build_ipset} += ipset/ obj-${build_portscan} += xt_portscan.o obj-${build_quota2} += xt_quota2.o diff --git a/extensions/ipset/.gitignore b/extensions/ipset/.gitignore new file mode 100644 index 0000000..9808551 --- /dev/null +++ b/extensions/ipset/.gitignore @@ -0,0 +1,3 @@ +*.oo +*.so +/ipset diff --git a/extensions/ipset/GNUmakefile.in b/extensions/ipset/GNUmakefile.in new file mode 100644 index 0000000..73d7b67 --- /dev/null +++ b/extensions/ipset/GNUmakefile.in @@ -0,0 +1,84 @@ +# -*- Makefile -*- + +top_srcdir := @top_srcdir@ +srcdir := @srcdir@ +abstop_srcdir := $(shell readlink -e ${top_srcdir}) +abssrcdir := $(shell readlink -e ${srcdir}) + +ifeq (${abstop_srcdir},) +$(error Path resolution of ${top_srcdir} failed) +endif +ifeq (${abssrcdir},) +$(error Path resolution of ${srcdir} failed) +endif + +prefix := @prefix@ +exec_prefix := @exec_prefix@ +sbindir := @sbindir@ +libdir := @libdir@ +libexecdir := @libexecdir@ +xtlibdir := @xtlibdir@ +kbuilddir := @kbuilddir@ +man8dir := @mandir@/man8 + +CC := @CC@ +CCLD := ${CC} +CFLAGS := @CFLAGS@ +LDFLAGS := @LDFLAGS@ +regular_CFLAGS := @regular_CFLAGS@ +kinclude_CFLAGS := @kinclude_CFLAGS@ +xtables_CFLAGS := @xtables_CFLAGS@ + +AM_CFLAGS := ${regular_CFLAGS} -I${top_srcdir}/include ${xtables_CFLAGS} ${kinclude_CFLAGS} -DXTABLES_LIBDIR=\"${xtlibdir}\" +AM_DEPFLAGS = -Wp,-MMD,$(@D)/.$(@F).d,-MT,$@ + +ifeq (${V},) +AM_LIBTOOL_SILENT = --silent +AM_VERBOSE_CC = @echo " CC " $@; +AM_VERBOSE_CCLD = @echo " CCLD " $@; +AM_VERBOSE_CXX = @echo " CXX " $@; +AM_VERBOSE_CXXLD = @echo " CXXLD " $@; +AM_VERBOSE_AR = @echo " AR " $@; +AM_VERBOSE_GEN = @echo " GEN " $@; +endif + +# +# Building blocks +# +targets := $(addsuffix .so,$(addprefix libipset_,iphash ipmap ipporthash iptree iptreemap macipmap nethash portmap)) + +.SECONDARY: + +.PHONY: all install clean distclean FORCE + +all: ipset ${targets} + +install: all + @mkdir -p "${DESTDIR}${sbindir}" "${DESTDIR}${xtlibdir}" "${DESTDIR}${man8dir}"; + install -pm0755 ipset "${DESTDIR}${sbindir}/"; + install -pm0755 ${targets} "${DESTDIR}${xtlibdir}/"; + install -pm0644 ipset.8 "${DESTDIR}${man8dir}/"; + +clean: + rm -f *.oo *.so *.o ipset; + +distclean: clean + rm -f .*.d; + +-include .*.d + + +ipset: ipset.o + ${AM_VERBOSE_CCLD} ${CCLD} ${AM_LDFLAGS} ${LDFLAGS} -o $@ $< -ldl -rdynamic; + +# +# Shared libraries +# +lib%.so: lib%.oo + ${AM_VERBOSE_CCLD} ${CCLD} ${AM_LDFLAGS} -shared ${LDFLAGS} -o $@ $<; + +libipset_%.oo: ${srcdir}/ipset_%.c + ${AM_VERBOSE_CC} ${CC} ${AM_DEPFLAGS} ${AM_CFLAGS} -DPIC -fPIC ${CFLAGS} -o $@ -c $<; + +%.o: %.c + ${AM_VERBOSE_CC} ${CC} ${AM_DEPFLAGS} ${AM_CFLAGS} ${CFLAGS} -o $@ -c $<; diff --git a/extensions/ipset/Kbuild b/extensions/ipset/Kbuild new file mode 100644 index 0000000..80ff927 --- /dev/null +++ b/extensions/ipset/Kbuild @@ -0,0 +1,6 @@ +# -*- Makefile -*- + +obj-m += ipt_set.o ipt_SET.o +obj-m += ip_set.o ip_set_ipmap.o ip_set_portmap.o ip_set_macipmap.o +obj-m += ip_set_iphash.o ip_set_nethash.o ip_set_ipporthash.o +obj-m += ip_set_iptree.o ip_set_iptreemap.o diff --git a/extensions/ipset/ip_set.c b/extensions/ipset/ip_set.c new file mode 100644 index 0000000..8be9274 --- /dev/null +++ b/extensions/ipset/ip_set.c @@ -0,0 +1,1981 @@ +/* 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 +#include +#include +#include + +#define ASSERT_READ_LOCK(x) +#define ASSERT_WRITE_LOCK(x) +#include "ip_set.h" + +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 DECLARE_MUTEX(ip_set_app_mutex); /* serializes user access */ +static ip_set_id_t ip_set_max = CONFIG_IP_NF_SET_MAX; +static ip_set_id_t ip_set_bindings_hash_size = CONFIG_IP_NF_SET_HASHSIZE; +static struct list_head *ip_set_hash; /* hash of bindings */ +static unsigned int ip_set_hash_random; /* random seed */ + +#define SETNAME_EQ(a,b) (strncmp(a,b,IP_SET_MAXNAMELEN) == 0) + +/* + * Sets are identified either by the index in ip_set_list or by id. + * The id never changes and is used to find a key in the hash. + * The index may change by swapping and used at all other places + * (set/SET netfilter modules, binding value, 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); +} + +/* + * Binding routines + */ + +static inline struct ip_set_hash * +__ip_set_find(u_int32_t key, ip_set_id_t id, ip_set_ip_t ip) +{ + struct ip_set_hash *set_hash; + + list_for_each_entry(set_hash, &ip_set_hash[key], list) + if (set_hash->id == id && set_hash->ip == ip) + return set_hash; + + return NULL; +} + +static ip_set_id_t +ip_set_find_in_hash(ip_set_id_t id, ip_set_ip_t ip) +{ + u_int32_t key = jhash_2words(id, ip, ip_set_hash_random) + % ip_set_bindings_hash_size; + struct ip_set_hash *set_hash; + + ASSERT_READ_LOCK(&ip_set_lock); + IP_SET_ASSERT(ip_set_list[id]); + DP("set: %s, ip: %u.%u.%u.%u", ip_set_list[id]->name, HIPQUAD(ip)); + + set_hash = __ip_set_find(key, id, ip); + + DP("set: %s, ip: %u.%u.%u.%u, binding: %s", ip_set_list[id]->name, + HIPQUAD(ip), + set_hash != NULL ? ip_set_list[set_hash->binding]->name : ""); + + return (set_hash != NULL ? set_hash->binding : IP_SET_INVALID_ID); +} + +static inline void +__set_hash_del(struct ip_set_hash *set_hash) +{ + ASSERT_WRITE_LOCK(&ip_set_lock); + IP_SET_ASSERT(ip_set_list[set_hash->binding]); + + __ip_set_put(set_hash->binding); + list_del(&set_hash->list); + kfree(set_hash); +} + +static int +ip_set_hash_del(ip_set_id_t id, ip_set_ip_t ip) +{ + u_int32_t key = jhash_2words(id, ip, ip_set_hash_random) + % ip_set_bindings_hash_size; + struct ip_set_hash *set_hash; + + IP_SET_ASSERT(ip_set_list[id]); + DP("set: %s, ip: %u.%u.%u.%u", ip_set_list[id]->name, HIPQUAD(ip)); + write_lock_bh(&ip_set_lock); + set_hash = __ip_set_find(key, id, ip); + DP("set: %s, ip: %u.%u.%u.%u, binding: %s", ip_set_list[id]->name, + HIPQUAD(ip), + set_hash != NULL ? ip_set_list[set_hash->binding]->name : ""); + + if (set_hash != NULL) + __set_hash_del(set_hash); + write_unlock_bh(&ip_set_lock); + return 0; +} + +static int +ip_set_hash_add(ip_set_id_t id, ip_set_ip_t ip, ip_set_id_t binding) +{ + u_int32_t key = jhash_2words(id, ip, ip_set_hash_random) + % ip_set_bindings_hash_size; + struct ip_set_hash *set_hash; + int ret = 0; + + IP_SET_ASSERT(ip_set_list[id]); + IP_SET_ASSERT(ip_set_list[binding]); + DP("set: %s, ip: %u.%u.%u.%u, binding: %s", ip_set_list[id]->name, + HIPQUAD(ip), ip_set_list[binding]->name); + write_lock_bh(&ip_set_lock); + set_hash = __ip_set_find(key, id, ip); + if (!set_hash) { + set_hash = kmalloc(sizeof(struct ip_set_hash), GFP_ATOMIC); + if (!set_hash) { + ret = -ENOMEM; + goto unlock; + } + INIT_LIST_HEAD(&set_hash->list); + set_hash->id = id; + set_hash->ip = ip; + list_add(&set_hash->list, &ip_set_hash[key]); + } else { + IP_SET_ASSERT(ip_set_list[set_hash->binding]); + DP("overwrite binding: %s", + ip_set_list[set_hash->binding]->name); + __ip_set_put(set_hash->binding); + } + set_hash->binding = binding; + __ip_set_get(set_hash->binding); + DP("stored: key %u, id %u (%s), ip %u.%u.%u.%u, binding %u (%s)", + key, id, ip_set_list[id]->name, + HIPQUAD(ip), binding, ip_set_list[binding]->name); + unlock: + write_unlock_bh(&ip_set_lock); + return ret; +} + +#define FOREACH_HASH_DO(fn, args...) \ +({ \ + ip_set_id_t __key; \ + struct ip_set_hash *__set_hash; \ + \ + for (__key = 0; __key < ip_set_bindings_hash_size; __key++) { \ + list_for_each_entry(__set_hash, &ip_set_hash[__key], list) \ + fn(__set_hash , ## args); \ + } \ +}) + +#define FOREACH_HASH_RW_DO(fn, args...) \ +({ \ + ip_set_id_t __key; \ + struct ip_set_hash *__set_hash, *__n; \ + \ + ASSERT_WRITE_LOCK(&ip_set_lock); \ + for (__key = 0; __key < ip_set_bindings_hash_size; __key++) { \ + list_for_each_entry_safe(__set_hash, __n, &ip_set_hash[__key], list)\ + fn(__set_hash , ## args); \ + } \ +}) + +/* Add, del and test set entries from kernel */ + +#define follow_bindings(index, set, ip) \ +((index = ip_set_find_in_hash((set)->id, ip)) != IP_SET_INVALID_ID \ + || (index = (set)->binding) != IP_SET_INVALID_ID) + +int +ip_set_testip_kernel(ip_set_id_t index, + const struct sk_buff *skb, + const u_int32_t *flags) +{ + struct ip_set *set; + ip_set_ip_t ip; + int res; + unsigned char i = 0; + + IP_SET_ASSERT(flags[i]); + read_lock_bh(&ip_set_lock); + do { + 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, &ip, flags, i++); + read_unlock_bh(&set->lock); + i += !!(set->type->features & IPSET_DATA_DOUBLE); + } while (res > 0 + && flags[i] + && follow_bindings(index, set, ip)); + read_unlock_bh(&ip_set_lock); + + return res; +} + +void +ip_set_addip_kernel(ip_set_id_t index, + const struct sk_buff *skb, + const u_int32_t *flags) +{ + struct ip_set *set; + ip_set_ip_t ip; + int res; + unsigned char i = 0; + + IP_SET_ASSERT(flags[i]); + retry: + read_lock_bh(&ip_set_lock); + do { + 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, &ip, flags, i++); + write_unlock_bh(&set->lock); + i += !!(set->type->features & IPSET_DATA_DOUBLE); + } while ((res == 0 || res == -EEXIST) + && flags[i] + && follow_bindings(index, set, ip)); + read_unlock_bh(&ip_set_lock); + + if (res == -EAGAIN + && set->type->retry + && (res = set->type->retry(set)) == 0) + goto retry; +} + +void +ip_set_delip_kernel(ip_set_id_t index, + const struct sk_buff *skb, + const u_int32_t *flags) +{ + struct ip_set *set; + ip_set_ip_t ip; + int res; + unsigned char i = 0; + + IP_SET_ASSERT(flags[i]); + read_lock_bh(&ip_set_lock); + do { + 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, &ip, flags, i++); + write_unlock_bh(&set->lock); + i += !!(set->type->features & IPSET_DATA_DOUBLE); + } while ((res == 0 || res == -EEXIST) + && flags[i] + && follow_bindings(index, set, ip)); + read_unlock_bh(&ip_set_lock); +} + +/* 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 (!strncmp(set_type->typename, name, IP_SET_MAXNAMELEN - 1)) + 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); + +} + +/* + * 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 + && SETNAME_EQ(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; +} + +/* + * 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(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 + && SETNAME_EQ(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, test, bind and unbind + */ + +static inline int +__ip_set_testip(struct ip_set *set, + const void *data, + size_t size, + ip_set_ip_t *ip) +{ + int res; + + read_lock_bh(&set->lock); + res = set->type->testip(set, data, size, ip); + read_unlock_bh(&set->lock); + + return res; +} + +static int +__ip_set_addip(ip_set_id_t index, + const void *data, + size_t size) +{ + struct ip_set *set = ip_set_list[index]; + ip_set_ip_t ip; + int res; + + IP_SET_ASSERT(set); + do { + write_lock_bh(&set->lock); + res = set->type->addip(set, data, size, &ip); + write_unlock_bh(&set->lock); + } while (res == -EAGAIN + && set->type->retry + && (res = set->type->retry(set)) == 0); + + return res; +} + +static int +ip_set_addip(ip_set_id_t index, + const void *data, + size_t size) +{ + + return __ip_set_addip(index, + data + sizeof(struct ip_set_req_adt), + size - sizeof(struct ip_set_req_adt)); +} + +static int +ip_set_delip(ip_set_id_t index, + const void *data, + size_t size) +{ + struct ip_set *set = ip_set_list[index]; + ip_set_ip_t ip; + int res; + + IP_SET_ASSERT(set); + write_lock_bh(&set->lock); + res = set->type->delip(set, + data + sizeof(struct ip_set_req_adt), + size - sizeof(struct ip_set_req_adt), + &ip); + write_unlock_bh(&set->lock); + + return res; +} + +static int +ip_set_testip(ip_set_id_t index, + const void *data, + size_t size) +{ + struct ip_set *set = ip_set_list[index]; + ip_set_ip_t ip; + int res; + + IP_SET_ASSERT(set); + res = __ip_set_testip(set, + data + sizeof(struct ip_set_req_adt), + size - sizeof(struct ip_set_req_adt), + &ip); + + return (res > 0 ? -EEXIST : res); +} + +static int +ip_set_bindip(ip_set_id_t index, + const void *data, + size_t size) +{ + struct ip_set *set = ip_set_list[index]; + const struct ip_set_req_bind *req_bind; + ip_set_id_t binding; + ip_set_ip_t ip; + int res; + + IP_SET_ASSERT(set); + if (size < sizeof(struct ip_set_req_bind)) + return -EINVAL; + + req_bind = data; + + if (SETNAME_EQ(req_bind->binding, IPSET_TOKEN_DEFAULT)) { + /* Default binding of a set */ + const char *binding_name; + + if (size != sizeof(struct ip_set_req_bind) + IP_SET_MAXNAMELEN) + return -EINVAL; + + binding_name = data + sizeof(struct ip_set_req_bind); + + binding = ip_set_find_byname(binding_name); + if (binding == IP_SET_INVALID_ID) + return -ENOENT; + + write_lock_bh(&ip_set_lock); + /* Sets as binding values are referenced */ + if (set->binding != IP_SET_INVALID_ID) + __ip_set_put(set->binding); + set->binding = binding; + __ip_set_get(set->binding); + write_unlock_bh(&ip_set_lock); + + return 0; + } + binding = ip_set_find_byname(req_bind->binding); + if (binding == IP_SET_INVALID_ID) + return -ENOENT; + + res = __ip_set_testip(set, + data + sizeof(struct ip_set_req_bind), + size - sizeof(struct ip_set_req_bind), + &ip); + DP("set %s, ip: %u.%u.%u.%u, binding %s", + set->name, HIPQUAD(ip), ip_set_list[binding]->name); + + if (res >= 0) + res = ip_set_hash_add(set->id, ip, binding); + + return res; +} + +#define FOREACH_SET_DO(fn, args...) \ +({ \ + ip_set_id_t __i; \ + struct ip_set *__set; \ + \ + for (__i = 0; __i < ip_set_max; __i++) { \ + __set = ip_set_list[__i]; \ + if (__set != NULL) \ + fn(__set , ##args); \ + } \ +}) + +static inline void +__set_hash_del_byid(struct ip_set_hash *set_hash, ip_set_id_t id) +{ + if (set_hash->id == id) + __set_hash_del(set_hash); +} + +static inline void +__unbind_default(struct ip_set *set) +{ + if (set->binding != IP_SET_INVALID_ID) { + /* Sets as binding values are referenced */ + __ip_set_put(set->binding); + set->binding = IP_SET_INVALID_ID; + } +} + +static int +ip_set_unbindip(ip_set_id_t index, + const void *data, + size_t size) +{ + struct ip_set *set; + const struct ip_set_req_bind *req_bind; + ip_set_ip_t ip; + int res; + + DP(""); + if (size < sizeof(struct ip_set_req_bind)) + return -EINVAL; + + req_bind = data; + + DP("%u %s", index, req_bind->binding); + if (index == IP_SET_INVALID_ID) { + /* unbind :all: */ + if (SETNAME_EQ(req_bind->binding, IPSET_TOKEN_DEFAULT)) { + /* Default binding of sets */ + write_lock_bh(&ip_set_lock); + FOREACH_SET_DO(__unbind_default); + write_unlock_bh(&ip_set_lock); + return 0; + } else if (SETNAME_EQ(req_bind->binding, IPSET_TOKEN_ALL)) { + /* Flush all bindings of all sets*/ + write_lock_bh(&ip_set_lock); + FOREACH_HASH_RW_DO(__set_hash_del); + write_unlock_bh(&ip_set_lock); + return 0; + } + DP("unreachable reached!"); + return -EINVAL; + } + + set = ip_set_list[index]; + IP_SET_ASSERT(set); + if (SETNAME_EQ(req_bind->binding, IPSET_TOKEN_DEFAULT)) { + /* Default binding of set */ + ip_set_id_t binding = ip_set_find_byindex(set->binding); + + if (binding == IP_SET_INVALID_ID) + return -ENOENT; + + write_lock_bh(&ip_set_lock); + /* Sets in hash values are referenced */ + __ip_set_put(set->binding); + set->binding = IP_SET_INVALID_ID; + write_unlock_bh(&ip_set_lock); + + return 0; + } else if (SETNAME_EQ(req_bind->binding, IPSET_TOKEN_ALL)) { + /* Flush all bindings */ + + write_lock_bh(&ip_set_lock); + FOREACH_HASH_RW_DO(__set_hash_del_byid, set->id); + write_unlock_bh(&ip_set_lock); + return 0; + } + + res = __ip_set_testip(set, + data + sizeof(struct ip_set_req_bind), + size - sizeof(struct ip_set_req_bind), + &ip); + + DP("set %s, ip: %u.%u.%u.%u", set->name, HIPQUAD(ip)); + if (res >= 0) + res = ip_set_hash_del(set->id, ip); + + return res; +} + +static int +ip_set_testbind(ip_set_id_t index, + const void *data, + size_t size) +{ + struct ip_set *set = ip_set_list[index]; + const struct ip_set_req_bind *req_bind; + ip_set_id_t binding; + ip_set_ip_t ip; + int res; + + IP_SET_ASSERT(set); + if (size < sizeof(struct ip_set_req_bind)) + return -EINVAL; + + req_bind = data; + + if (SETNAME_EQ(req_bind->binding, IPSET_TOKEN_DEFAULT)) { + /* Default binding of set */ + const char *binding_name; + + if (size != sizeof(struct ip_set_req_bind) + IP_SET_MAXNAMELEN) + return -EINVAL; + + binding_name = data + sizeof(struct ip_set_req_bind); + + binding = ip_set_find_byname(binding_name); + if (binding == IP_SET_INVALID_ID) + return -ENOENT; + + res = (set->binding == binding) ? -EEXIST : 0; + + return res; + } + binding = ip_set_find_byname(req_bind->binding); + if (binding == IP_SET_INVALID_ID) + return -ENOENT; + + + res = __ip_set_testip(set, + data + sizeof(struct ip_set_req_bind), + size - sizeof(struct ip_set_req_bind), + &ip); + DP("set %s, ip: %u.%u.%u.%u, binding %s", + set->name, HIPQUAD(ip), ip_set_list[binding]->name); + + if (res >= 0) + res = (ip_set_find_in_hash(set->id, ip) == binding) + ? -EEXIST : 0; + + return 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 (SETNAME_EQ(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, + size_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; + set->lock = RW_LOCK_UNLOCKED; + strncpy(set->name, name, IP_SET_MAXNAMELEN); + set->binding = IP_SET_INVALID_ID; + 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); + + /* + * 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); + FOREACH_HASH_RW_DO(__set_hash_del_byid, set->id); + if (set->binding != IP_SET_INVALID_ID) + __ip_set_put(set->binding); + 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 + FOREACH_SET_DO(ip_set_flush_set); + + 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 + && SETNAME_EQ(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. Artifical restriction. */ + 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 inline void +__set_hash_bindings_size_list(struct ip_set_hash *set_hash, + ip_set_id_t id, size_t *size) +{ + if (set_hash->id == id) + *size += sizeof(struct ip_set_hash_list); +} + +static inline void +__set_hash_bindings_size_save(struct ip_set_hash *set_hash, + ip_set_id_t id, size_t *size) +{ + if (set_hash->id == id) + *size += sizeof(struct ip_set_hash_save); +} + +static inline void +__set_hash_bindings(struct ip_set_hash *set_hash, + ip_set_id_t id, void *data, int *used) +{ + if (set_hash->id == id) { + struct ip_set_hash_list *hash_list = data + *used; + + hash_list->ip = set_hash->ip; + hash_list->binding = set_hash->binding; + *used += sizeof(struct ip_set_hash_list); + } +} + +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 %p %p", set->name, *used, data, data + *used); + + /* Get and ensure header size */ + if (*used + sizeof(struct ip_set_list) > len) + goto not_enough_mem; + *used += sizeof(struct ip_set_list); + + read_lock_bh(&set->lock); + /* Get and ensure set specific header size */ + set_list->header_size = 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 = set->binding; + 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); + if (*used + set_list->members_size > len) + goto unlock_set; + + /* Fill in set spefific members data */ + set->type->list_members(set, data + *used); + *used += set_list->members_size; + read_unlock_bh(&set->lock); + + /* Bindings */ + + /* Get and ensure set specific bindings size */ + set_list->bindings_size = 0; + FOREACH_HASH_DO(__set_hash_bindings_size_list, + set->id, &set_list->bindings_size); + if (*used + set_list->bindings_size > len) + goto not_enough_mem; + + /* Fill in set spefific bindings data */ + FOREACH_HASH_DO(__set_hash_bindings, set->id, data, used); + + return 0; + + unlock_set: + read_unlock_bh(&set->lock); + not_enough_mem: + DP("not enough mem, try again"); + return -EAGAIN; +} + +/* + * Save sets + */ +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 + sizeof(struct ip_set_save) > len) + goto not_enough_mem; + *used += sizeof(struct ip_set_save); + + set = ip_set_list[index]; + DP("set: %s, used: %u(%u) %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 = 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 = set->binding; + + /* Fill in set spefific header data */ + set->type->list_header(set, data + *used); + *used += set_save->header_size; + + DP("set header filled: %s, used: %u(%u) %p %p", set->name, *used, + set_save->header_size, data, data + *used); + /* Get and ensure set specific members size */ + set_save->members_size = set->type->list_members_size(set); + if (*used + set_save->members_size > len) + goto unlock_set; + + /* Fill in set spefific members data */ + set->type->list_members(set, data + *used); + *used += set_save->members_size; + read_unlock_bh(&set->lock); + DP("set members filled: %s, used: %u(%u) %p %p", set->name, *used, + 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; +} + +static inline void +__set_hash_save_bindings(struct ip_set_hash *set_hash, + ip_set_id_t id, + void *data, + int *used, + int len, + int *res) +{ + if (*res == 0 + && (id == IP_SET_INVALID_ID || set_hash->id == id)) { + struct ip_set_hash_save *hash_save = data + *used; + /* Ensure bindings size */ + if (*used + sizeof(struct ip_set_hash_save) > len) { + *res = -ENOMEM; + return; + } + hash_save->id = set_hash->id; + hash_save->ip = set_hash->ip; + hash_save->binding = set_hash->binding; + *used += sizeof(struct ip_set_hash_save); + } +} + +static int ip_set_save_bindings(ip_set_id_t index, + void *data, + int *used, + int len) +{ + int res = 0; + struct ip_set_save *set_save; + + DP("used %u, len %u", *used, len); + /* Get and ensure header size */ + if (*used + sizeof(struct ip_set_save) > len) + return -ENOMEM; + + /* Marker */ + set_save = data + *used; + set_save->index = IP_SET_INVALID_ID; + set_save->header_size = 0; + set_save->members_size = 0; + *used += sizeof(struct ip_set_save); + + DP("marker added used %u, len %u", *used, len); + /* Fill in bindings data */ + if (index != IP_SET_INVALID_ID) + /* Sets are identified by id in hash */ + index = ip_set_list[index]->id; + FOREACH_HASH_DO(__set_hash_save_bindings, index, data, used, len, &res); + + return res; +} + +/* + * 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_hash_save *hash_save; + struct ip_set_restore *set_restore; + ip_set_id_t index; + + /* Loop to restore sets */ + while (1) { + line++; + + DP("%u %u %u", used, sizeof(struct ip_set_restore), len); + /* Get and ensure header size */ + if (used + sizeof(struct ip_set_restore) > len) + return line; + set_restore = data + used; + used += 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 bindings; + } + + /* 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 += 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 %u reqsize %u", + set_restore->members_size, set->type->reqsize); + while (members_size + set->type->reqsize <= + set_restore->members_size) { + line++; + DP("members: %u, line %u", members_size, line); + res = __ip_set_addip(index, + data + used + members_size, + set->type->reqsize); + if (!(res == 0 || res == -EEXIST)) + return line; + members_size += set->type->reqsize; + } + + DP("members_size %u %u", + set_restore->members_size, members_size); + if (members_size != set_restore->members_size) + return line++; + used += set_restore->members_size; + } + + bindings: + /* Loop to restore bindings */ + while (used < len) { + line++; + + DP("restore binding, line %u", line); + /* Get and ensure size */ + if (used + sizeof(struct ip_set_hash_save) > len) + return line; + hash_save = data + used; + used += sizeof(struct ip_set_hash_save); + + /* hash_save->id is used to store the index */ + index = ip_set_find_byindex(hash_save->id); + DP("restore binding index %u, id %u, %u -> %u", + index, hash_save->id, hash_save->ip, hash_save->binding); + if (index != hash_save->id) + return line; + if (ip_set_find_byindex(hash_save->binding) == IP_SET_INVALID_ID) { + DP("corrupt binding set index %u", hash_save->binding); + return line; + } + set = ip_set_list[hash_save->id]; + /* Null valued IP means default binding */ + if (hash_save->ip) + res = ip_set_hash_add(set->id, + hash_save->ip, + hash_save->binding); + else { + IP_SET_ASSERT(set->binding == IP_SET_INVALID_ID); + write_lock_bh(&ip_set_lock); + set->binding = hash_save->binding; + __ip_set_get(set->binding); + write_unlock_bh(&ip_set_lock); + DP("default binding: %u", set->binding); + } + if (res != 0) + return line; + } + 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 */ + unsigned *op; + struct ip_set_req_adt *req_adt; + ip_set_id_t index = IP_SET_INVALID_ID; + int (*adtfn)(ip_set_id_t index, + const void *data, size_t size); + struct fn_table { + int (*fn)(ip_set_id_t index, + const void *data, size_t size); + } adtfn_table[] = + { { ip_set_addip }, { ip_set_delip }, { ip_set_testip}, + { ip_set_bindip}, { ip_set_unbindip }, { ip_set_testbind }, + }; + + 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 done; + } + if (down_interruptible(&ip_set_app_mutex)) { + res = -EINTR; + goto done; + } + + 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_VERSION) { + res = -EPROTO; + goto done; + } + } + + switch (*op) { + case IP_SET_OP_CREATE:{ + struct ip_set_req_create *req_create = data; + + if (len < sizeof(struct ip_set_req_create)) { + ip_set_printk("short CREATE data (want >=%zu, got %u)", + sizeof(struct ip_set_req_create), 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 + sizeof(struct ip_set_req_create), + len - sizeof(struct ip_set_req_create)); + 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 (SETNAME_EQ(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 (SETNAME_EQ(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_BIND_SET) { + res = -EBADMSG; + goto done; + } + adtfn = adtfn_table[*op - IP_SET_OP_ADD_IP].fn; + + if (len < sizeof(struct ip_set_req_adt)) { + ip_set_printk("short data in adt request (want >=%zu, got %u)", + sizeof(struct ip_set_req_adt), len); + res = -EINVAL; + goto done; + } + req_adt = data; + + /* -U :all: :all:|:default: uses IP_SET_INVALID_ID */ + if (!(*op == IP_SET_OP_UNBIND_SET + && req_adt->index == IP_SET_INVALID_ID)) { + index = ip_set_find_byindex(req_adt->index); + if (index == IP_SET_INVALID_ID) { + res = -ENOENT; + goto done; + } + } + res = adtfn(index, data, len); + + done: + up(&ip_set_app_mutex); + 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 done; + } + if (down_interruptible(&ip_set_app_mutex)) { + res = -EINTR; + goto done; + } + + 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_VERSION) { + res = -EPROTO; + goto done; + } + } + + 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 (SETNAME_EQ(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 < sizeof(struct ip_set_req_setnames)) { + ip_set_printk("short LIST_SIZE (want >=%zu, got %d)", + sizeof(struct ip_set_req_setnames), *len); + res = -EINVAL; + goto done; + } + + req_setnames->size = 0; + used = 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 += 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 */ + switch (*op) { + case IP_SET_OP_LIST_SIZE: { + req_setnames->size += sizeof(struct ip_set_list) + + set->type->header_size + + set->type->list_members_size(set); + /* Sets are identified by id in the hash */ + FOREACH_HASH_DO(__set_hash_bindings_size_list, + set->id, &req_setnames->size); + break; + } + case IP_SET_OP_SAVE_SIZE: { + req_setnames->size += sizeof(struct ip_set_save) + + set->type->header_size + + set->type->list_members_size(set); + FOREACH_HASH_DO(__set_hash_bindings_size_save, + set->id, &req_setnames->size); + break; + } + default: + break; + } + } + 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; + } + used = 0; + if (index == IP_SET_INVALID_ID) { + /* Save all sets */ + for (i = 0; i < ip_set_max && res == 0; i++) { + if (ip_set_list[i] != NULL) + res = ip_set_save_set(i, data, &used, *len); + } + } else { + /* Save an individual set */ + res = ip_set_save_set(index, data, &used, *len); + } + if (res == 0) + res = ip_set_save_bindings(index, 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; + int line; + + if (*len < sizeof(struct ip_set_req_setnames) + || *len != req_restore->size) { + ip_set_printk("invalid RESTORE (want =%zu, got %d)", + req_restore->size, *len); + res = -EINVAL; + goto done; + } + line = ip_set_restore(data + sizeof(struct ip_set_req_setnames), + req_restore->size - sizeof(struct ip_set_req_setnames)); + DP("ip_set_restore: %u", 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 %u", 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); + 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, hash_size; +module_param(max_sets, int, 0600); +MODULE_PARM_DESC(max_sets, "maximal number of sets"); +module_param(hash_size, int, 0600); +MODULE_PARM_DESC(hash_size, "hash size for bindings"); +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("module implementing core IP set support"); + +static int __init ip_set_init(void) +{ + int res; + ip_set_id_t i; + + get_random_bytes(&ip_set_hash_random, 4); + if (max_sets) + ip_set_max = max_sets; + 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); + if (hash_size) + ip_set_bindings_hash_size = hash_size; + ip_set_hash = vmalloc(sizeof(struct list_head) * ip_set_bindings_hash_size); + if (!ip_set_hash) { + printk(KERN_ERR "Unable to create ip_set_hash\n"); + vfree(ip_set_list); + return -ENOMEM; + } + for (i = 0; i < ip_set_bindings_hash_size; i++) + INIT_LIST_HEAD(&ip_set_hash[i]); + + 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); + vfree(ip_set_hash); + return res; + } + 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); + vfree(ip_set_hash); + 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); + +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/extensions/ipset/ip_set.h b/extensions/ipset/ip_set.h new file mode 100644 index 0000000..d6cf536 --- /dev/null +++ b/extensions/ipset/ip_set.h @@ -0,0 +1,506 @@ +#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. + */ + +#ifndef CONFIG_IP_NF_SET_MAX + /* from 2 to 65534 */ +# define CONFIG_IP_NF_SET_MAX 256 +#endif +#ifndef CONFIG_IP_NF_SET_HASHSIZE +# define CONFIG_IP_NF_SET_HASHSIZE 1024 +#endif + +#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_VERSION 2 + +#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 */ + +/* 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 */ + size_t size; /* size to get setdata/bindings */ + /* 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; + size_t header_size; /* Set header data of header_size */ + size_t members_size; /* Set members data of members_size */ + size_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; + size_t header_size; /* Set header data of header_size */ + size_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; + size_t header_size; /* Create data of header_size */ + size_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); +} + +#ifdef __KERNEL__ + +#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, + ip_set_ip_t *ip, + const u_int32_t *flags, + unsigned char index); + + /* 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, size_t size, + ip_set_ip_t *ip); + + /* + * Size of the data structure passed by when + * adding/deletin/testing an entry. + */ + size_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, size_t size, + ip_set_ip_t *ip); + + /* 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, + ip_set_ip_t *ip, + const u_int32_t *flags, + unsigned char index); + + /* 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, size_t size, + ip_set_ip_t *ip); + + /* 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, + ip_set_ip_t *ip, + const u_int32_t *flags, + unsigned char index); + + /* new set creation - allocated type specific items + */ + int (*create) (struct ip_set *set, + const void *data, size_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 + */ + size_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); + + /* 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 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 */ + ip_set_id_t binding; /* default binding for the set */ + atomic_t ref; /* in kernel and in hash references */ + struct ip_set_type *type; /* the set types */ + void *data; /* pooltype specific data */ +}; + +/* Structure to bind set elements to sets */ +struct ip_set_hash { + struct list_head list; /* list of clashing entries in hash */ + ip_set_ip_t ip; /* ip from set */ + ip_set_id_t id; /* set id */ + ip_set_id_t binding; /* set we bind the element to */ +}; + +/* 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 id); +extern void ip_set_put(ip_set_id_t id); + +/* API for iptables set match, and SET target */ +extern void ip_set_addip_kernel(ip_set_id_t id, + const struct sk_buff *skb, + const u_int32_t *flags); +extern void 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); + +#endif /* __KERNEL__ */ + +#endif /*_IP_SET_H*/ diff --git a/extensions/ipset/ip_set_iphash.c b/extensions/ipset/ip_set_iphash.c new file mode 100644 index 0000000..3108c1b --- /dev/null +++ b/extensions/ipset/ip_set_iphash.c @@ -0,0 +1,425 @@ +/* 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 implementing an ip hash set */ + +#include +#include +#include +#include +#include +#include +#include "ip_set.h" +#include +#include +#include +#include +#include +#include + +#include + +#include "ip_set_malloc.h" +#include "ip_set_iphash.h" + +static int limit = MAX_RANGE; + +static inline __u32 +jhash_ip(const struct ip_set_iphash *map, uint16_t i, ip_set_ip_t ip) +{ + return jhash_1word(ip, *(((uint32_t *) map->initval) + i)); +} + +static inline __u32 +hash_id(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) +{ + struct ip_set_iphash *map = set->data; + __u32 id; + u_int16_t i; + ip_set_ip_t *elem; + + *hash_ip = ip & map->netmask; + DP("set: %s, ip:%u.%u.%u.%u, %u.%u.%u.%u, %u.%u.%u.%u", + set->name, HIPQUAD(ip), HIPQUAD(*hash_ip), HIPQUAD(map->netmask)); + + for (i = 0; i < map->probes; i++) { + id = jhash_ip(map, i, *hash_ip) % map->hashsize; + DP("hash key: %u", id); + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); + if (*elem == *hash_ip) + return id; + /* No shortcut at testing - there can be deleted + * entries. */ + } + return UINT_MAX; +} + +static inline int +__testip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) +{ + return (ip && hash_id(set, ip, hash_ip) != UINT_MAX); +} + +static int +testip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_iphash *req = data; + + if (size != sizeof(struct ip_set_req_iphash)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_iphash), + size); + return -EINVAL; + } + return __testip(set, req->ip, hash_ip); +} + +static int +testip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + return __testip(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + hash_ip); +} + +static inline int +__addip(struct ip_set_iphash *map, ip_set_ip_t ip, ip_set_ip_t *hash_ip) +{ + __u32 probe; + u_int16_t i; + ip_set_ip_t *elem; + + if (!ip || map->elements >= limit) + return -ERANGE; + + *hash_ip = ip & map->netmask; + + for (i = 0; i < map->probes; i++) { + probe = jhash_ip(map, i, *hash_ip) % map->hashsize; + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, probe); + if (*elem == *hash_ip) + return -EEXIST; + if (!*elem) { + *elem = *hash_ip; + map->elements++; + return 0; + } + } + /* Trigger rehashing */ + return -EAGAIN; +} + +static int +addip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_iphash *req = data; + + if (size != sizeof(struct ip_set_req_iphash)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_iphash), + size); + return -EINVAL; + } + return __addip(set->data, req->ip, hash_ip); +} + +static int +addip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + return __addip((struct ip_set_iphash *) set->data, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + hash_ip); +} + +static int retry(struct ip_set *set) +{ + struct ip_set_iphash *map = set->data; + ip_set_ip_t hash_ip, *elem; + void *members; + u_int32_t i, hashsize = map->hashsize; + int res; + struct ip_set_iphash *tmp; + + 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 %u to %u", + set->name, map->hashsize, hashsize); + + tmp = kmalloc(sizeof(struct ip_set_iphash) + + map->probes * sizeof(uint32_t), GFP_ATOMIC); + if (!tmp) { + DP("out of memory for %d bytes", + sizeof(struct ip_set_iphash) + + map->probes * sizeof(uint32_t)); + return -ENOMEM; + } + tmp->members = harray_malloc(hashsize, sizeof(ip_set_ip_t), GFP_ATOMIC); + if (!tmp->members) { + DP("out of memory for %d bytes", hashsize * sizeof(ip_set_ip_t)); + kfree(tmp); + return -ENOMEM; + } + tmp->hashsize = hashsize; + tmp->elements = 0; + tmp->probes = map->probes; + tmp->resize = map->resize; + tmp->netmask = map->netmask; + memcpy(tmp->initval, map->initval, map->probes * sizeof(uint32_t)); + + write_lock_bh(&set->lock); + map = set->data; /* Play safe */ + for (i = 0; i < map->hashsize && res == 0; i++) { + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, i); + if (*elem) + res = __addip(tmp, *elem, &hash_ip); + } + 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; +} + +static inline int +__delip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) +{ + struct ip_set_iphash *map = set->data; + ip_set_ip_t id, *elem; + + if (!ip) + return -ERANGE; + + id = hash_id(set, ip, hash_ip); + if (id == UINT_MAX) + return -EEXIST; + + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); + *elem = 0; + map->elements--; + + return 0; +} + +static int +delip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_iphash *req = data; + + if (size != sizeof(struct ip_set_req_iphash)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_iphash), + size); + return -EINVAL; + } + return __delip(set, req->ip, hash_ip); +} + +static int +delip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + return __delip(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + hash_ip); +} + +static int create(struct ip_set *set, const void *data, size_t size) +{ + const struct ip_set_req_iphash_create *req = data; + struct ip_set_iphash *map; + uint16_t i; + + if (size != sizeof(struct ip_set_req_iphash_create)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_iphash_create), + size); + return -EINVAL; + } + + 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_iphash) + + req->probes * sizeof(uint32_t), GFP_KERNEL); + if (!map) { + DP("out of memory for %d bytes", + sizeof(struct ip_set_iphash) + + req->probes * sizeof(uint32_t)); + return -ENOMEM; + } + for (i = 0; i < req->probes; i++) + get_random_bytes(((uint32_t *) map->initval)+i, 4); + map->elements = 0; + map->hashsize = req->hashsize; + map->probes = req->probes; + map->resize = req->resize; + map->netmask = req->netmask; + map->members = harray_malloc(map->hashsize, sizeof(ip_set_ip_t), GFP_KERNEL); + if (!map->members) { + DP("out of memory for %d bytes", map->hashsize * sizeof(ip_set_ip_t)); + kfree(map); + return -ENOMEM; + } + + set->data = map; + return 0; +} + +static void destroy(struct ip_set *set) +{ + struct ip_set_iphash *map = set->data; + + harray_free(map->members); + kfree(map); + + set->data = NULL; +} + +static void flush(struct ip_set *set) +{ + struct ip_set_iphash *map = set->data; + harray_flush(map->members, map->hashsize, sizeof(ip_set_ip_t)); + map->elements = 0; +} + +static void list_header(const struct ip_set *set, void *data) +{ + struct ip_set_iphash *map = set->data; + struct ip_set_req_iphash_create *header = data; + + header->hashsize = map->hashsize; + header->probes = map->probes; + header->resize = map->resize; + header->netmask = map->netmask; +} + +static int list_members_size(const struct ip_set *set) +{ + const struct ip_set_iphash *map = set->data; + + return (map->hashsize * sizeof(ip_set_ip_t)); +} + +static void list_members(const struct ip_set *set, void *data) +{ + const struct ip_set_iphash *map = set->data; + ip_set_ip_t i, *elem; + + for (i = 0; i < map->hashsize; i++) { + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, i); + ((ip_set_ip_t *)data)[i] = *elem; + } +} + +static struct ip_set_type ip_set_iphash = { + .typename = SETTYPE_NAME, + .features = IPSET_TYPE_IP | IPSET_DATA_SINGLE, + .protocol_version = IP_SET_PROTOCOL_VERSION, + .create = &create, + .destroy = &destroy, + .flush = &flush, + .reqsize = sizeof(struct ip_set_req_iphash), + .addip = &addip, + .addip_kernel = &addip_kernel, + .retry = &retry, + .delip = &delip, + .delip_kernel = &delip_kernel, + .testip = &testip, + .testip_kernel = &testip_kernel, + .header_size = sizeof(struct ip_set_req_iphash_create), + .list_header = &list_header, + .list_members_size = &list_members_size, + .list_members = &list_members, + .me = THIS_MODULE, +}; + +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"); + +static int __init ip_set_iphash_init(void) +{ + init_max_page_size(); + return ip_set_register_set_type(&ip_set_iphash); +} + +static void __exit ip_set_iphash_fini(void) +{ + /* FIXME: possible race with ip_set_create() */ + ip_set_unregister_set_type(&ip_set_iphash); +} + +module_init(ip_set_iphash_init); +module_exit(ip_set_iphash_fini); diff --git a/extensions/ipset/ip_set_iphash.h b/extensions/ipset/ip_set_iphash.h new file mode 100644 index 0000000..38cf2f2 --- /dev/null +++ b/extensions/ipset/ip_set_iphash.h @@ -0,0 +1,30 @@ +#ifndef __IP_SET_IPHASH_H +#define __IP_SET_IPHASH_H + +#include "ip_set.h" + +#define SETTYPE_NAME "iphash" +#define MAX_RANGE 0x0000FFFF + +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 */ + void *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/extensions/ipset/ip_set_ipmap.c b/extensions/ipset/ip_set_ipmap.c new file mode 100644 index 0000000..394871e --- /dev/null +++ b/extensions/ipset/ip_set_ipmap.c @@ -0,0 +1,331 @@ +/* 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 implementing an IP set type: the single bitmap type */ + +#include +#include +#include +#include +#include +#include "ip_set.h" +#include +#include +#include +#include + +#include "ip_set_ipmap.h" + +static inline ip_set_ip_t +ip_to_id(const struct ip_set_ipmap *map, ip_set_ip_t ip) +{ + return (ip - map->first_ip)/map->hosts; +} + +static inline int +__testip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) +{ + struct ip_set_ipmap *map = set->data; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + *hash_ip = ip & map->netmask; + DP("set: %s, ip:%u.%u.%u.%u, %u.%u.%u.%u", + set->name, HIPQUAD(ip), HIPQUAD(*hash_ip)); + return !!test_bit(ip_to_id(map, *hash_ip), map->members); +} + +static int +testip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_ipmap *req = data; + + if (size != sizeof(struct ip_set_req_ipmap)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_ipmap), + size); + return -EINVAL; + } + return __testip(set, req->ip, hash_ip); +} + +static int +testip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + int res = __testip(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + hash_ip); + return (res < 0 ? 0 : res); +} + +static inline int +__addip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) +{ + struct ip_set_ipmap *map = set->data; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + *hash_ip = ip & map->netmask; + DP("%u.%u.%u.%u, %u.%u.%u.%u", HIPQUAD(ip), HIPQUAD(*hash_ip)); + if (test_and_set_bit(ip_to_id(map, *hash_ip), map->members)) + return -EEXIST; + + return 0; +} + +static int +addip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_ipmap *req = data; + + if (size != sizeof(struct ip_set_req_ipmap)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_ipmap), + size); + return -EINVAL; + } + DP("%u.%u.%u.%u", HIPQUAD(req->ip)); + return __addip(set, req->ip, hash_ip); +} + +static int +addip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + return __addip(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + hash_ip); +} + +static inline int +__delip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) +{ + struct ip_set_ipmap *map = set->data; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + *hash_ip = ip & map->netmask; + DP("%u.%u.%u.%u, %u.%u.%u.%u", HIPQUAD(ip), HIPQUAD(*hash_ip)); + if (!test_and_clear_bit(ip_to_id(map, *hash_ip), map->members)) + return -EEXIST; + + return 0; +} + +static int +delip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_ipmap *req = data; + + if (size != sizeof(struct ip_set_req_ipmap)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_ipmap), + size); + return -EINVAL; + } + return __delip(set, req->ip, hash_ip); +} + +static int +delip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + return __delip(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + hash_ip); +} + +static int create(struct ip_set *set, const void *data, size_t size) +{ + int newbytes; + const struct ip_set_req_ipmap_create *req = data; + struct ip_set_ipmap *map; + + if (size != sizeof(struct ip_set_req_ipmap_create)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_ipmap_create), + size); + return -EINVAL; + } + + DP("from %u.%u.%u.%u to %u.%u.%u.%u", + HIPQUAD(req->from), HIPQUAD(req->to)); + + if (req->from > req->to) { + DP("bad ip range"); + return -ENOEXEC; + } + + map = kmalloc(sizeof(struct ip_set_ipmap), GFP_KERNEL); + if (!map) { + DP("out of memory for %d bytes", + sizeof(struct ip_set_ipmap)); + return -ENOMEM; + } + map->first_ip = req->from; + map->last_ip = req->to; + 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 (max %d addresses)", + MAX_RANGE+1); + kfree(map); + return -ENOEXEC; + } + DP("hosts %u, sizeid %u", map->hosts, map->sizeid); + newbytes = bitmap_bytes(0, map->sizeid - 1); + map->members = kmalloc(newbytes, GFP_KERNEL); + if (!map->members) { + DP("out of memory for %d bytes", newbytes); + kfree(map); + return -ENOMEM; + } + memset(map->members, 0, newbytes); + + set->data = map; + return 0; +} + +static void destroy(struct ip_set *set) +{ + struct ip_set_ipmap *map = set->data; + + kfree(map->members); + kfree(map); + + set->data = NULL; +} + +static void flush(struct ip_set *set) +{ + struct ip_set_ipmap *map = set->data; + memset(map->members, 0, bitmap_bytes(0, map->sizeid - 1)); +} + +static void list_header(const struct ip_set *set, void *data) +{ + const struct ip_set_ipmap *map = set->data; + struct ip_set_req_ipmap_create *header = data; + + header->from = map->first_ip; + header->to = map->last_ip; + header->netmask = map->netmask; +} + +static int list_members_size(const struct ip_set *set) +{ + const struct ip_set_ipmap *map = set->data; + + return bitmap_bytes(0, map->sizeid - 1); +} + +static void list_members(const struct ip_set *set, void *data) +{ + const struct ip_set_ipmap *map = set->data; + int bytes = bitmap_bytes(0, map->sizeid - 1); + + memcpy(data, map->members, bytes); +} + +static struct ip_set_type ip_set_ipmap = { + .typename = SETTYPE_NAME, + .features = IPSET_TYPE_IP | IPSET_DATA_SINGLE, + .protocol_version = IP_SET_PROTOCOL_VERSION, + .create = &create, + .destroy = &destroy, + .flush = &flush, + .reqsize = sizeof(struct ip_set_req_ipmap), + .addip = &addip, + .addip_kernel = &addip_kernel, + .delip = &delip, + .delip_kernel = &delip_kernel, + .testip = &testip, + .testip_kernel = &testip_kernel, + .header_size = sizeof(struct ip_set_req_ipmap_create), + .list_header = &list_header, + .list_members_size = &list_members_size, + .list_members = &list_members, + .me = THIS_MODULE, +}; + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("ipmap type of IP sets"); + +static int __init ip_set_ipmap_init(void) +{ + return ip_set_register_set_type(&ip_set_ipmap); +} + +static void __exit ip_set_ipmap_fini(void) +{ + /* FIXME: possible race with ip_set_create() */ + ip_set_unregister_set_type(&ip_set_ipmap); +} + +module_init(ip_set_ipmap_init); +module_exit(ip_set_ipmap_fini); diff --git a/extensions/ipset/ip_set_ipmap.h b/extensions/ipset/ip_set_ipmap.h new file mode 100644 index 0000000..ebe67fc --- /dev/null +++ b/extensions/ipset/ip_set_ipmap.h @@ -0,0 +1,56 @@ +#ifndef __IP_SET_IPMAP_H +#define __IP_SET_IPMAP_H + +#include "ip_set.h" + +#define SETTYPE_NAME "ipmap" +#define MAX_RANGE 0x0000FFFF + +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 */ +}; + +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 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 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/extensions/ipset/ip_set_ipporthash.c b/extensions/ipset/ip_set_ipporthash.c new file mode 100644 index 0000000..07b290e --- /dev/null +++ b/extensions/ipset/ip_set_ipporthash.c @@ -0,0 +1,575 @@ +/* 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 implementing an ip+port hash set */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include "ip_set.h" +#include +#include +#include +#include +#include +#include + +#include + +#include "ip_set_malloc.h" +#include "ip_set_ipporthash.h" + +static int limit = MAX_RANGE; + +/* We must handle non-linear skbs */ +static inline ip_set_ip_t +get_port(const struct sk_buff *skb, u_int32_t flags) +{ +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + struct iphdr *iph = ip_hdr(skb); +#else + struct iphdr *iph = skb->nh.iph; +#endif + 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 LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + if (skb_copy_bits(skb, ip_hdr(skb)->ihl*4, &tcph, sizeof(tcph)) < 0) +#else + if (skb_copy_bits(skb, skb->nh.iph->ihl*4, &tcph, sizeof(tcph)) < 0) +#endif + /* No choice either */ + return INVALID_PORT; + + return ntohs(flags & IPSET_SRC ? + tcph.source : tcph.dest); + } + case IPPROTO_UDP: { + struct udphdr udph; + + if (offset) + return INVALID_PORT; + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + if (skb_copy_bits(skb, ip_hdr(skb)->ihl*4, &udph, sizeof(udph)) < 0) +#else + if (skb_copy_bits(skb, skb->nh.iph->ihl*4, &udph, sizeof(udph)) < 0) +#endif + /* No choice either */ + return INVALID_PORT; + + return ntohs(flags & IPSET_SRC ? + udph.source : udph.dest); + } + default: + return INVALID_PORT; + } +} + +static inline __u32 +jhash_ip(const struct ip_set_ipporthash *map, uint16_t i, ip_set_ip_t ip) +{ + return jhash_1word(ip, *(((uint32_t *) map->initval) + i)); +} + +#define HASH_IP(map, ip, port) (port + ((ip - ((map)->first_ip)) << 16)) + +static inline __u32 +hash_id(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t port, + ip_set_ip_t *hash_ip) +{ + struct ip_set_ipporthash *map = set->data; + __u32 id; + u_int16_t i; + ip_set_ip_t *elem; + + *hash_ip = HASH_IP(map, ip, port); + DP("set: %s, ipport:%u.%u.%u.%u:%u, %u.%u.%u.%u", + set->name, HIPQUAD(ip), port, HIPQUAD(*hash_ip)); + + for (i = 0; i < map->probes; i++) { + id = jhash_ip(map, i, *hash_ip) % map->hashsize; + DP("hash key: %u", id); + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); + if (*elem == *hash_ip) + return id; + /* No shortcut at testing - there can be deleted + * entries. */ + } + return UINT_MAX; +} + +static inline int +__testip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t port, + ip_set_ip_t *hash_ip) +{ + struct ip_set_ipporthash *map = set->data; + + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + return (hash_id(set, ip, port, hash_ip) != UINT_MAX); +} + +static int +testip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_ipporthash *req = data; + + if (size != sizeof(struct ip_set_req_ipporthash)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_ipporthash), + size); + return -EINVAL; + } + return __testip(set, req->ip, req->port, hash_ip); +} + +static int +testip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + ip_set_ip_t port; + int res; + + if (flags[index+1] == 0) + return 0; + + port = get_port(skb, flags[index+1]); + + DP("flag: %s src: %u.%u.%u.%u dst: %u.%u.%u.%u", + flags[index] & IPSET_SRC ? "SRC" : "DST", +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + NIPQUAD(ip_hdr(skb)->saddr), + NIPQUAD(ip_hdr(skb)->daddr)); +#else + NIPQUAD(skb->nh.iph->saddr), + NIPQUAD(skb->nh.iph->daddr)); +#endif + DP("flag %s port %u", + flags[index+1] & IPSET_SRC ? "SRC" : "DST", + port); + if (port == INVALID_PORT) + return 0; + + res = __testip(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + port, + hash_ip); + return (res < 0 ? 0 : res); + +} + +static inline int +__add_haship(struct ip_set_ipporthash *map, ip_set_ip_t hash_ip) +{ + __u32 probe; + u_int16_t i; + ip_set_ip_t *elem; + + for (i = 0; i < map->probes; i++) { + probe = jhash_ip(map, i, hash_ip) % map->hashsize; + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, probe); + if (*elem == hash_ip) + return -EEXIST; + if (!*elem) { + *elem = hash_ip; + map->elements++; + return 0; + } + } + /* Trigger rehashing */ + return -EAGAIN; +} + +static inline int +__addip(struct ip_set_ipporthash *map, ip_set_ip_t ip, ip_set_ip_t port, + ip_set_ip_t *hash_ip) +{ + if (map->elements > limit) + return -ERANGE; + if (ip < map->first_ip || ip > map->last_ip) + return -ERANGE; + + *hash_ip = HASH_IP(map, ip, port); + + return __add_haship(map, *hash_ip); +} + +static int +addip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_ipporthash *req = data; + + if (size != sizeof(struct ip_set_req_ipporthash)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_ipporthash), + size); + return -EINVAL; + } + return __addip(set->data, req->ip, req->port, hash_ip); +} + +static int +addip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + ip_set_ip_t port; + + if (flags[index+1] == 0) + return -EINVAL; + + port = get_port(skb, flags[index+1]); + + DP("flag: %s src: %u.%u.%u.%u dst: %u.%u.%u.%u", + flags[index] & IPSET_SRC ? "SRC" : "DST", +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + NIPQUAD(ip_hdr(skb)->saddr), + NIPQUAD(ip_hdr(skb)->daddr)); +#else + NIPQUAD(skb->nh.iph->saddr), + NIPQUAD(skb->nh.iph->daddr)); +#endif + DP("flag %s port %u", + flags[index+1] & IPSET_SRC ? "SRC" : "DST", + port); + if (port == INVALID_PORT) + return -EINVAL; + + return __addip(set->data, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + port, + hash_ip); +} + +static int retry(struct ip_set *set) +{ + struct ip_set_ipporthash *map = set->data; + ip_set_ip_t *elem; + void *members; + u_int32_t i, hashsize = map->hashsize; + int res; + struct ip_set_ipporthash *tmp; + + 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 %u to %u", + set->name, map->hashsize, hashsize); + + tmp = kmalloc(sizeof(struct ip_set_ipporthash) + + map->probes * sizeof(uint32_t), GFP_ATOMIC); + if (!tmp) { + DP("out of memory for %d bytes", + sizeof(struct ip_set_ipporthash) + + map->probes * sizeof(uint32_t)); + return -ENOMEM; + } + tmp->members = harray_malloc(hashsize, sizeof(ip_set_ip_t), GFP_ATOMIC); + if (!tmp->members) { + DP("out of memory for %d bytes", hashsize * sizeof(ip_set_ip_t)); + kfree(tmp); + return -ENOMEM; + } + tmp->hashsize = hashsize; + tmp->elements = 0; + tmp->probes = map->probes; + tmp->resize = map->resize; + tmp->first_ip = map->first_ip; + tmp->last_ip = map->last_ip; + memcpy(tmp->initval, map->initval, map->probes * sizeof(uint32_t)); + + write_lock_bh(&set->lock); + map = set->data; /* Play safe */ + for (i = 0; i < map->hashsize && res == 0; i++) { + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, i); + if (*elem) + res = __add_haship(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; +} + +static inline int +__delip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t port, + ip_set_ip_t *hash_ip) +{ + 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 = hash_id(set, ip, port, hash_ip); + + if (id == UINT_MAX) + return -EEXIST; + + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); + *elem = 0; + map->elements--; + + return 0; +} + +static int +delip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_ipporthash *req = data; + + if (size != sizeof(struct ip_set_req_ipporthash)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_ipporthash), + size); + return -EINVAL; + } + return __delip(set, req->ip, req->port, hash_ip); +} + +static int +delip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + ip_set_ip_t port; + + if (flags[index+1] == 0) + return -EINVAL; + + port = get_port(skb, flags[index+1]); + + DP("flag: %s src: %u.%u.%u.%u dst: %u.%u.%u.%u", + flags[index] & IPSET_SRC ? "SRC" : "DST", +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + NIPQUAD(ip_hdr(skb)->saddr), + NIPQUAD(ip_hdr(skb)->daddr)); +#else + NIPQUAD(skb->nh.iph->saddr), + NIPQUAD(skb->nh.iph->daddr)); +#endif + DP("flag %s port %u", + flags[index+1] & IPSET_SRC ? "SRC" : "DST", + port); + if (port == INVALID_PORT) + return -EINVAL; + + return __delip(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + port, + hash_ip); +} + +static int create(struct ip_set *set, const void *data, size_t size) +{ + const struct ip_set_req_ipporthash_create *req = data; + struct ip_set_ipporthash *map; + uint16_t i; + + if (size != sizeof(struct ip_set_req_ipporthash_create)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_ipporthash_create), + size); + return -EINVAL; + } + + 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_ipporthash) + + req->probes * sizeof(uint32_t), GFP_KERNEL); + if (!map) { + DP("out of memory for %d bytes", + sizeof(struct ip_set_ipporthash) + + req->probes * sizeof(uint32_t)); + return -ENOMEM; + } + for (i = 0; i < req->probes; i++) + get_random_bytes(((uint32_t *) map->initval)+i, 4); + map->elements = 0; + map->hashsize = req->hashsize; + map->probes = req->probes; + map->resize = req->resize; + map->first_ip = req->from; + map->last_ip = req->to; + map->members = harray_malloc(map->hashsize, sizeof(ip_set_ip_t), GFP_KERNEL); + if (!map->members) { + DP("out of memory for %d bytes", map->hashsize * sizeof(ip_set_ip_t)); + kfree(map); + return -ENOMEM; + } + + set->data = map; + return 0; +} + +static void destroy(struct ip_set *set) +{ + struct ip_set_ipporthash *map = set->data; + + harray_free(map->members); + kfree(map); + + set->data = NULL; +} + +static void flush(struct ip_set *set) +{ + struct ip_set_ipporthash *map = set->data; + harray_flush(map->members, map->hashsize, sizeof(ip_set_ip_t)); + map->elements = 0; +} + +static void list_header(const struct ip_set *set, void *data) +{ + const struct ip_set_ipporthash *map = set->data; + struct ip_set_req_ipporthash_create *header = data; + + header->hashsize = map->hashsize; + header->probes = map->probes; + header->resize = map->resize; + header->from = map->first_ip; + header->to = map->last_ip; +} + +static int list_members_size(const struct ip_set *set) +{ + const struct ip_set_ipporthash *map = set->data; + + return (map->hashsize * sizeof(ip_set_ip_t)); +} + +static void list_members(const struct ip_set *set, void *data) +{ + const struct ip_set_ipporthash *map = set->data; + ip_set_ip_t i, *elem; + + for (i = 0; i < map->hashsize; i++) { + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, i); + ((ip_set_ip_t *)data)[i] = *elem; + } +} + +static struct ip_set_type ip_set_ipporthash = { + .typename = SETTYPE_NAME, + .features = IPSET_TYPE_IP | IPSET_TYPE_PORT | IPSET_DATA_DOUBLE, + .protocol_version = IP_SET_PROTOCOL_VERSION, + .create = &create, + .destroy = &destroy, + .flush = &flush, + .reqsize = sizeof(struct ip_set_req_ipporthash), + .addip = &addip, + .addip_kernel = &addip_kernel, + .retry = &retry, + .delip = &delip, + .delip_kernel = &delip_kernel, + .testip = &testip, + .testip_kernel = &testip_kernel, + .header_size = sizeof(struct ip_set_req_ipporthash_create), + .list_header = &list_header, + .list_members_size = &list_members_size, + .list_members = &list_members, + .me = THIS_MODULE, +}; + +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"); + +static int __init ip_set_ipporthash_init(void) +{ + init_max_page_size(); + return ip_set_register_set_type(&ip_set_ipporthash); +} + +static void __exit ip_set_ipporthash_fini(void) +{ + /* FIXME: possible race with ip_set_create() */ + ip_set_unregister_set_type(&ip_set_ipporthash); +} + +module_init(ip_set_ipporthash_init); +module_exit(ip_set_ipporthash_fini); diff --git a/extensions/ipset/ip_set_ipporthash.h b/extensions/ipset/ip_set_ipporthash.h new file mode 100644 index 0000000..ba838e3 --- /dev/null +++ b/extensions/ipset/ip_set_ipporthash.h @@ -0,0 +1,34 @@ +#ifndef __IP_SET_IPPORTHASH_H +#define __IP_SET_IPPORTHASH_H + +#include "ip_set.h" + +#define SETTYPE_NAME "ipporthash" +#define MAX_RANGE 0x0000FFFF +#define INVALID_PORT (MAX_RANGE + 1) + +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 */ + void *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/extensions/ipset/ip_set_iptree.c b/extensions/ipset/ip_set_iptree.c new file mode 100644 index 0000000..5170294 --- /dev/null +++ b/extensions/ipset/ip_set_iptree.c @@ -0,0 +1,607 @@ +/* Copyright (C) 2005 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 "ip_set.h" +#include +#include +#include +#include + +/* Backward compatibility */ +#ifndef __nocast +#define __nocast +#endif + +#include "ip_set_iptree.h" + +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 + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20) +static struct kmem_cache *branch_cachep; +static struct kmem_cache *leaf_cachep; +#else +static kmem_cache_t *branch_cachep; +static kmem_cache_t *leaf_cachep; +#endif + +#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 +__testip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_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; + + *hash_ip = ip; + ABCD(a, b, c, d, hash_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)); +} + +static int +testip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_iptree *req = data; + + if (size != sizeof(struct ip_set_req_iptree)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_iptree), + size); + return -EINVAL; + } + return __testip(set, req->ip, hash_ip); +} + +static int +testip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + int res; + + DP("flag: %s src: %u.%u.%u.%u dst: %u.%u.%u.%u", + flags[index] & IPSET_SRC ? "SRC" : "DST", +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + NIPQUAD(ip_hdr(skb)->saddr), + NIPQUAD(ip_hdr(skb)->daddr)); +#else + NIPQUAD(skb->nh.iph->saddr), + NIPQUAD(skb->nh.iph->daddr)); +#endif + + res = __testip(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + hash_ip); + return (res < 0 ? 0 : res); +} + +#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 +__addip(struct ip_set *set, ip_set_ip_t ip, unsigned int timeout, + ip_set_ip_t *hash_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; + int ret = 0; + + if (!ip || map->elements >= limit) + /* We could call the garbage collector + * but it's probably overkill */ + return -ERANGE; + + *hash_ip = ip; + ABCD(a, b, c, d, hash_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; + 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; +} + +static int +addip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + struct ip_set_iptree *map = set->data; + const struct ip_set_req_iptree *req = data; + + if (size != sizeof(struct ip_set_req_iptree)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_iptree), + size); + return -EINVAL; + } + DP("%u.%u.%u.%u %u", HIPQUAD(req->ip), req->timeout); + return __addip(set, req->ip, + req->timeout ? req->timeout : map->timeout, + hash_ip); +} + +static int +addip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + struct ip_set_iptree *map = set->data; + + return __addip(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + map->timeout, + hash_ip); +} + +#define DELIP_WALK(map, elem, branch) do { \ + if ((map)->tree[elem]) { \ + branch = (map)->tree[elem]; \ + } else \ + return -EEXIST; \ +} while (0) + +static inline int +__delip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_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; + + *hash_ip = ip; + ABCD(a, b, c, d, hash_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; +} + +static int +delip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_iptree *req = data; + + if (size != sizeof(struct ip_set_req_iptree)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_iptree), + size); + return -EINVAL; + } + return __delip(set, req->ip, hash_ip); +} + +static int +delip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + return __delip(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + hash_ip); +} + +#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 create(struct ip_set *set, const void *data, size_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 %zu)", + sizeof(struct ip_set_req_iptree_create), + size); + return -EINVAL; + } + + map = kmalloc(sizeof(struct ip_set_iptree), GFP_KERNEL); + if (!map) { + DP("out of memory for %d 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 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 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 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 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 list_members_size(const struct ip_set *set) +{ + 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 * sizeof(struct ip_set_req_iptree)); +} + +static void list_members(const struct ip_set *set, void *data) +{ + 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; + struct ip_set_req_iptree *entry; + + 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 += sizeof(struct ip_set_req_iptree); + } + } + LOOP_WALK_END; + LOOP_WALK_END; + LOOP_WALK_END; +} + +static struct ip_set_type ip_set_iptree = { + .typename = SETTYPE_NAME, + .features = IPSET_TYPE_IP | IPSET_DATA_SINGLE, + .protocol_version = IP_SET_PROTOCOL_VERSION, + .create = &create, + .destroy = &destroy, + .flush = &flush, + .reqsize = sizeof(struct ip_set_req_iptree), + .addip = &addip, + .addip_kernel = &addip_kernel, + .delip = &delip, + .delip_kernel = &delip_kernel, + .testip = &testip, + .testip_kernel = &testip_kernel, + .header_size = sizeof(struct ip_set_req_iptree_create), + .list_header = &list_header, + .list_members_size = &list_members_size, + .list_members = &list_members, + .me = THIS_MODULE, +}; + +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; + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) + branch_cachep = kmem_cache_create("ip_set_iptreeb", + sizeof(struct ip_set_iptreeb), + 0, 0, NULL); +#else + branch_cachep = kmem_cache_create("ip_set_iptreeb", + sizeof(struct ip_set_iptreeb), + 0, 0, NULL, NULL); +#endif + if (!branch_cachep) { + printk(KERN_ERR "Unable to create ip_set_iptreeb slab cache\n"); + ret = -ENOMEM; + goto out; + } +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) + leaf_cachep = kmem_cache_create("ip_set_iptreed", + sizeof(struct ip_set_iptreed), + 0, 0, NULL); +#else + leaf_cachep = kmem_cache_create("ip_set_iptreed", + sizeof(struct ip_set_iptreed), + 0, 0, NULL, NULL); +#endif + 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/extensions/ipset/ip_set_iptree.h b/extensions/ipset/ip_set_iptree.h new file mode 100644 index 0000000..f33affe --- /dev/null +++ b/extensions/ipset/ip_set_iptree.h @@ -0,0 +1,40 @@ +#ifndef __IP_SET_IPTREE_H +#define __IP_SET_IPTREE_H + +#include "ip_set.h" + +#define SETTYPE_NAME "iptree" +#define MAX_RANGE 0x0000FFFF + +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/extensions/ipset/ip_set_iptreemap.c b/extensions/ipset/ip_set_iptreemap.c new file mode 100644 index 0000000..bc102fc --- /dev/null +++ b/extensions/ipset/ip_set_iptreemap.c @@ -0,0 +1,827 @@ +/* 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 "ip_set.h" +#include +#include +#include +#include + +#include "ip_set_iptreemap.h" + +#define IPTREEMAP_DEFAULT_GC_TIME (5 * 60) +#define IPTREEMAP_DESTROY_SLEEP (100) + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,20) +static struct kmem_cache *cachep_b; +static struct kmem_cache *cachep_c; +static struct kmem_cache *cachep_d; +#else +static kmem_cache_t *cachep_b; +static kmem_cache_t *cachep_c; +static kmem_cache_t *cachep_d; +#endif + +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, flags) \ + do { \ + branch = (map)->tree[elem]; \ + if (!branch) { \ + return -EEXIST; \ + } else if (branch == full) { \ + branch = kmem_cache_alloc(cachep, flags); \ + 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, flags) \ + 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, flags); \ + 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 +__testip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_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; + + *hash_ip = ip; + + ABCD(a, b, c, d, hash_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); +} + +static int +testip(struct ip_set *set, const void *data, size_t size, ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_iptreemap *req = data; + + if (size != sizeof(struct ip_set_req_iptreemap)) { + ip_set_printk("data length wrong (want %zu, have %zu)", sizeof(struct ip_set_req_iptreemap), size); + return -EINVAL; + } + + return __testip(set, req->start, hash_ip); +} + +static int +testip_kernel(struct ip_set *set, const struct sk_buff *skb, ip_set_ip_t *hash_ip, const u_int32_t *flags, unsigned char index) +{ + int res; + + res = __testip(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + hash_ip); + + return (res < 0 ? 0 : res); +} + +static inline int +__addip_single(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_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; + + *hash_ip = ip; + + ABCD(a, b, c, d, hash_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 +__addip_range(struct ip_set *set, ip_set_ip_t start, ip_set_ip_t end, ip_set_ip_t *hash_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 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, hash_ip); + + *hash_ip = 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; +} + +static int +addip(struct ip_set *set, const void *data, size_t size, ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_iptreemap *req = data; + + if (size != sizeof(struct ip_set_req_iptreemap)) { + ip_set_printk("data length wrong (want %zu, have %zu)", sizeof(struct ip_set_req_iptreemap), size); + return -EINVAL; + } + + return __addip_range(set, min(req->start, req->end), max(req->start, req->end), hash_ip); +} + +static int +addip_kernel(struct ip_set *set, const struct sk_buff *skb, ip_set_ip_t *hash_ip, const u_int32_t *flags, unsigned char index) +{ + + return __addip_single(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + hash_ip); +} + +static inline int +__delip_single(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip, unsigned int __nocast flags) +{ + 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; + + *hash_ip = ip; + + ABCD(a, b, c, d, hash_ip); + + DELIP_WALK(map, a, btree, cachep_b, fullbitmap_b, flags); + DELIP_WALK(btree, b, ctree, cachep_c, fullbitmap_c, flags); + DELIP_WALK(ctree, c, dtree, cachep_d, fullbitmap_d, flags); + + if (!__test_and_clear_bit(d, (void *) dtree->bitmap)) + return -EEXIST; + + __set_bit(b, (void *) btree->dirty); + + return 0; +} + +static inline int +__delip_range(struct ip_set *set, ip_set_ip_t start, ip_set_ip_t end, ip_set_ip_t *hash_ip, unsigned int __nocast flags) +{ + 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, hash_ip, flags); + + *hash_ip = 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, flags) { + 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, flags) { + 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, flags) { + 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; +} + +static int +delip(struct ip_set *set, const void *data, size_t size, ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_iptreemap *req = data; + + if (size != sizeof(struct ip_set_req_iptreemap)) { + ip_set_printk("data length wrong (want %zu, have %zu)", sizeof(struct ip_set_req_iptreemap), size); + return -EINVAL; + } + + return __delip_range(set, min(req->start, req->end), max(req->start, req->end), hash_ip, GFP_KERNEL); +} + +static int +delip_kernel(struct ip_set *set, const struct sk_buff *skb, ip_set_ip_t *hash_ip, const u_int32_t *flags, unsigned char index) +{ + return __delip_single(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + hash_ip, + GFP_ATOMIC); +} + +/* 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 create(struct ip_set *set, const void *data, size_t size) +{ + const struct ip_set_req_iptreemap_create *req = data; + struct ip_set_iptreemap *map; + + if (size != sizeof(struct ip_set_req_iptreemap_create)) { + ip_set_printk("data length wrong (want %zu, have %zu)", sizeof(struct ip_set_req_iptreemap_create), size); + return -EINVAL; + } + + 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 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 flush(struct ip_set *set) +{ + struct ip_set_iptreemap *map = set->data; + + while (!del_timer(&map->gc)) + msleep(IPTREEMAP_DESTROY_SLEEP); + + __flush(map); + + memset(map, 0, sizeof(*map)); + + init_gc_timer(set); +} + +static void 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 list_members_size(const struct ip_set *set) +{ + 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 * sizeof(struct ip_set_req_iptreemap)); +} + +static inline size_t 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->start = start; + entry->end = end; + + return sizeof(*entry); +} + +static void list_members(const struct ip_set *set, void *data) +{ + 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; + ip_set_ip_t start = 0, end = 0, ip; + + 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) { + offset += add_member(data, offset, start, end); + start = ip; + } + end = ip; + } else if (inrange) { + offset += add_member(data, offset, start, end); + inrange = 0; + } + } + } LOOP_WALK_END(); + } LOOP_WALK_END(); + } LOOP_WALK_END(); + + if (inrange) + add_member(data, offset, start, end); +} + +static struct ip_set_type ip_set_iptreemap = { + .typename = SETTYPE_NAME, + .features = IPSET_TYPE_IP | IPSET_DATA_SINGLE, + .protocol_version = IP_SET_PROTOCOL_VERSION, + .create = create, + .destroy = destroy, + .flush = flush, + .reqsize = sizeof(struct ip_set_req_iptreemap), + .addip = addip, + .addip_kernel = addip_kernel, + .delip = delip, + .delip_kernel = delip_kernel, + .testip = testip, + .testip_kernel = testip_kernel, + .header_size = sizeof(struct ip_set_req_iptreemap_create), + .list_header = list_header, + .list_members_size = list_members_size, + .list_members = list_members, + .me = THIS_MODULE, +}; + +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; + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) + cachep_b = kmem_cache_create("ip_set_iptreemap_b", + sizeof(struct ip_set_iptreemap_b), + 0, 0, NULL); +#else + cachep_b = kmem_cache_create("ip_set_iptreemap_b", + sizeof(struct ip_set_iptreemap_b), + 0, 0, NULL, NULL); +#endif + if (!cachep_b) { + ip_set_printk("Unable to create ip_set_iptreemap_b slab cache"); + goto out; + } + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) + cachep_c = kmem_cache_create("ip_set_iptreemap_c", + sizeof(struct ip_set_iptreemap_c), + 0, 0, NULL); +#else + cachep_c = kmem_cache_create("ip_set_iptreemap_c", + sizeof(struct ip_set_iptreemap_c), + 0, 0, NULL, NULL); +#endif + if (!cachep_c) { + ip_set_printk("Unable to create ip_set_iptreemap_c slab cache"); + goto outb; + } + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) + cachep_d = kmem_cache_create("ip_set_iptreemap_d", + sizeof(struct ip_set_iptreemap_d), + 0, 0, NULL); +#else + cachep_d = kmem_cache_create("ip_set_iptreemap_d", + sizeof(struct ip_set_iptreemap_d), + 0, 0, NULL, NULL); +#endif + 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/extensions/ipset/ip_set_iptreemap.h b/extensions/ipset/ip_set_iptreemap.h new file mode 100644 index 0000000..2f6c071 --- /dev/null +++ b/extensions/ipset/ip_set_iptreemap.h @@ -0,0 +1,40 @@ +#ifndef __IP_SET_IPTREEMAP_H +#define __IP_SET_IPTREEMAP_H + +#include "ip_set.h" + +#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 start; + ip_set_ip_t end; +}; + +#endif /* __IP_SET_IPTREEMAP_H */ diff --git a/extensions/ipset/ip_set_jhash.h b/extensions/ipset/ip_set_jhash.h new file mode 100644 index 0000000..25c6b97 --- /dev/null +++ b/extensions/ipset/ip_set_jhash.h @@ -0,0 +1,148 @@ +#ifndef _LINUX_IPSET_JHASH_H +#define _LINUX_IPSET_JHASH_H + +/* This is a copy of linux/jhash.h but the types u32/u8 are changed + * to __u32/__u8 so that the header file can be included into + * userspace code as well. Jozsef Kadlecsik (kadlec@blackhole.kfki.hu) + */ + +/* jhash.h: Jenkins hash support. + * + * Copyright (C) 1996 Bob Jenkins (bob_jenkins@burtleburtle.net) + * + * http://burtleburtle.net/bob/hash/ + * + * These are the credits from Bob's sources: + * + * lookup2.c, by Bob Jenkins, December 1996, Public Domain. + * hash(), hash2(), hash3, and mix() 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 has no warranty. + * + * Copyright (C) 2003 David S. Miller (davem@redhat.com) + * + * I've modified Bob's hash to be useful in the Linux kernel, and + * any bugs present are surely my fault. -DaveM + */ + +/* NOTE: Arguments are modified. */ +#define __jhash_mix(a, b, c) \ +{ \ + a -= b; a -= c; a ^= (c>>13); \ + b -= c; b -= a; b ^= (a<<8); \ + c -= a; c -= b; c ^= (b>>13); \ + a -= b; a -= c; a ^= (c>>12); \ + b -= c; b -= a; b ^= (a<<16); \ + c -= a; c -= b; c ^= (b>>5); \ + a -= b; a -= c; a ^= (c>>3); \ + b -= c; b -= a; b ^= (a<<10); \ + c -= a; c -= b; c ^= (b>>15); \ +} + +/* The golden ration: an arbitrary value */ +#define JHASH_GOLDEN_RATIO 0x9e3779b9 + +/* The most generic version, hashes an arbitrary sequence + * of bytes. No alignment or length assumptions are made about + * the input key. + */ +static inline __u32 jhash(void *key, __u32 length, __u32 initval) +{ + __u32 a, b, c, len; + __u8 *k = key; + + len = length; + a = b = JHASH_GOLDEN_RATIO; + c = initval; + + while (len >= 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); + + k += 12; + len -= 12; + } + + c += length; + switch (len) { + case 11: c += ((__u32)k[10]<<24); + case 10: c += ((__u32)k[9]<<16); + case 9 : c += ((__u32)k[8]<<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_mix(a,b,c); + + 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(__u32 *k, __u32 length, __u32 initval) +{ + __u32 a, b, c, len; + + a = b = JHASH_GOLDEN_RATIO; + c = initval; + len = length; + + while (len >= 3) { + a += k[0]; + b += k[1]; + c += k[2]; + __jhash_mix(a, b, c); + k += 3; len -= 3; + } + + c += length * 4; + + switch (len) { + case 2 : b += k[1]; + case 1 : a += k[0]; + }; + + __jhash_mix(a,b,c); + + return c; +} + + +/* A special ultra-optimized versions that knows they are hashing exactly + * 3, 2 or 1 word(s). + * + * NOTE: In partilar the "c += length; __jhash_mix(a,b,c);" normally + * done at the end is not done here. + */ +static inline __u32 jhash_3words(__u32 a, __u32 b, __u32 c, __u32 initval) +{ + a += JHASH_GOLDEN_RATIO; + b += JHASH_GOLDEN_RATIO; + c += initval; + + __jhash_mix(a, b, c); + + return c; +} + +static inline __u32 jhash_2words(__u32 a, __u32 b, __u32 initval) +{ + return jhash_3words(a, b, 0, initval); +} + +static inline __u32 jhash_1word(__u32 a, __u32 initval) +{ + return jhash_3words(a, 0, 0, initval); +} + +#endif /* _LINUX_IPSET_JHASH_H */ diff --git a/extensions/ipset/ip_set_macipmap.c b/extensions/ipset/ip_set_macipmap.c new file mode 100644 index 0000000..76c4c4a --- /dev/null +++ b/extensions/ipset/ip_set_macipmap.c @@ -0,0 +1,360 @@ +/* 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 implementing an IP set type: the macipmap type */ + +#include +#include +#include +#include +#include +#include "ip_set.h" +#include +#include +#include +#include +#include +#include + +#include "ip_set_malloc.h" +#include "ip_set_macipmap.h" + +static int +testip(struct ip_set *set, const void *data, size_t size, ip_set_ip_t *hash_ip) +{ + struct ip_set_macipmap *map = set->data; + struct ip_set_macip *table = map->members; + const struct ip_set_req_macipmap *req = data; + + if (size != sizeof(struct ip_set_req_macipmap)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_macipmap), + size); + return -EINVAL; + } + + if (req->ip < map->first_ip || req->ip > map->last_ip) + return -ERANGE; + + *hash_ip = req->ip; + DP("set: %s, ip:%u.%u.%u.%u, %u.%u.%u.%u", + set->name, HIPQUAD(req->ip), HIPQUAD(*hash_ip)); + if (test_bit(IPSET_MACIP_ISSET, + (void *) &table[req->ip - map->first_ip].flags)) { + 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 +testip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + struct ip_set_macipmap *map = set->data; + struct ip_set_macip *table = map->members; + ip_set_ip_t ip; + + ip = ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr); +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr); +#endif + + if (ip < map->first_ip || ip > map->last_ip) + return 0; + + *hash_ip = ip; + DP("set: %s, ip:%u.%u.%u.%u, %u.%u.%u.%u", + set->name, HIPQUAD(ip), HIPQUAD(*hash_ip)); + if (test_bit(IPSET_MACIP_ISSET, + (void *) &table[ip - map->first_ip].flags)) { + /* Is mac pointer valid? + * If so, compare... */ +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + return (skb_mac_header(skb) >= skb->head + && (skb_mac_header(skb) + ETH_HLEN) <= skb->data +#else + return (skb->mac.raw >= skb->head + && (skb->mac.raw + ETH_HLEN) <= skb->data +#endif + && (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 +__addip(struct ip_set *set, + ip_set_ip_t ip, const unsigned char *ethernet, ip_set_ip_t *hash_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 (test_and_set_bit(IPSET_MACIP_ISSET, + (void *) &table[ip - map->first_ip].flags)) + return -EEXIST; + + *hash_ip = ip; + DP("%u.%u.%u.%u, %u.%u.%u.%u", HIPQUAD(ip), HIPQUAD(*hash_ip)); + memcpy(&table[ip - map->first_ip].ethernet, ethernet, ETH_ALEN); + return 0; +} + +static int +addip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_macipmap *req = data; + + if (size != sizeof(struct ip_set_req_macipmap)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_macipmap), + size); + return -EINVAL; + } + return __addip(set, req->ip, req->ethernet, hash_ip); +} + +static int +addip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + ip_set_ip_t ip; + + ip = ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr); +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr); +#endif + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + if (!(skb_mac_header(skb) >= skb->head + && (skb_mac_header(skb) + ETH_HLEN) <= skb->data)) +#else + if (!(skb->mac.raw >= skb->head + && (skb->mac.raw + ETH_HLEN) <= skb->data)) +#endif + return -EINVAL; + + return __addip(set, ip, eth_hdr(skb)->h_source, hash_ip); +} + +static inline int +__delip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_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 (!test_and_clear_bit(IPSET_MACIP_ISSET, + (void *)&table[ip - map->first_ip].flags)) + return -EEXIST; + + *hash_ip = ip; + DP("%u.%u.%u.%u, %u.%u.%u.%u", HIPQUAD(ip), HIPQUAD(*hash_ip)); + return 0; +} + +static int +delip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_macipmap *req = data; + + if (size != sizeof(struct ip_set_req_macipmap)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_macipmap), + size); + return -EINVAL; + } + return __delip(set, req->ip, hash_ip); +} + +static int +delip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + return __delip(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + hash_ip); +} + +static inline size_t members_size(ip_set_ip_t from, ip_set_ip_t to) +{ + return (size_t)((to - from + 1) * sizeof(struct ip_set_macip)); +} + +static int create(struct ip_set *set, const void *data, size_t size) +{ + size_t newbytes; + const struct ip_set_req_macipmap_create *req = data; + struct ip_set_macipmap *map; + + if (size != sizeof(struct ip_set_req_macipmap_create)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_macipmap_create), + size); + return -EINVAL; + } + + DP("from %u.%u.%u.%u to %u.%u.%u.%u", + HIPQUAD(req->from), HIPQUAD(req->to)); + + if (req->from > req->to) { + DP("bad ip range"); + return -ENOEXEC; + } + + if (req->to - req->from > MAX_RANGE) { + ip_set_printk("range too big (max %d addresses)", + MAX_RANGE+1); + return -ENOEXEC; + } + + map = kmalloc(sizeof(struct ip_set_macipmap), GFP_KERNEL); + if (!map) { + DP("out of memory for %d bytes", + sizeof(struct ip_set_macipmap)); + return -ENOMEM; + } + map->flags = req->flags; + map->first_ip = req->from; + map->last_ip = req->to; + newbytes = members_size(map->first_ip, map->last_ip); + map->members = ip_set_malloc(newbytes); + DP("members: %u %p", newbytes, map->members); + if (!map->members) { + DP("out of memory for %d bytes", newbytes); + kfree(map); + return -ENOMEM; + } + memset(map->members, 0, newbytes); + + set->data = map; + return 0; +} + +static void destroy(struct ip_set *set) +{ + struct ip_set_macipmap *map = set->data; + + ip_set_free(map->members, members_size(map->first_ip, map->last_ip)); + kfree(map); + + set->data = NULL; +} + +static void flush(struct ip_set *set) +{ + struct ip_set_macipmap *map = set->data; + memset(map->members, 0, members_size(map->first_ip, map->last_ip)); +} + +static void list_header(const struct ip_set *set, void *data) +{ + const struct ip_set_macipmap *map = set->data; + struct ip_set_req_macipmap_create *header = data; + + DP("list_header %x %x %u", map->first_ip, map->last_ip, + map->flags); + + header->from = map->first_ip; + header->to = map->last_ip; + header->flags = map->flags; +} + +static int list_members_size(const struct ip_set *set) +{ + const struct ip_set_macipmap *map = set->data; + + DP("%u", members_size(map->first_ip, map->last_ip)); + return members_size(map->first_ip, map->last_ip); +} + +static void list_members(const struct ip_set *set, void *data) +{ + const struct ip_set_macipmap *map = set->data; + + int bytes = members_size(map->first_ip, map->last_ip); + + DP("members: %u %p", bytes, map->members); + memcpy(data, map->members, bytes); +} + +static struct ip_set_type ip_set_macipmap = { + .typename = SETTYPE_NAME, + .features = IPSET_TYPE_IP | IPSET_DATA_SINGLE, + .protocol_version = IP_SET_PROTOCOL_VERSION, + .create = &create, + .destroy = &destroy, + .flush = &flush, + .reqsize = sizeof(struct ip_set_req_macipmap), + .addip = &addip, + .addip_kernel = &addip_kernel, + .delip = &delip, + .delip_kernel = &delip_kernel, + .testip = &testip, + .testip_kernel = &testip_kernel, + .header_size = sizeof(struct ip_set_req_macipmap_create), + .list_header = &list_header, + .list_members_size = &list_members_size, + .list_members = &list_members, + .me = THIS_MODULE, +}; + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("macipmap type of IP sets"); + +static int __init ip_set_macipmap_init(void) +{ + init_max_page_size(); + return ip_set_register_set_type(&ip_set_macipmap); +} + +static void __exit ip_set_macipmap_fini(void) +{ + /* FIXME: possible race with ip_set_create() */ + ip_set_unregister_set_type(&ip_set_macipmap); +} + +module_init(ip_set_macipmap_init); +module_exit(ip_set_macipmap_fini); diff --git a/extensions/ipset/ip_set_macipmap.h b/extensions/ipset/ip_set_macipmap.h new file mode 100644 index 0000000..9f9389b --- /dev/null +++ b/extensions/ipset/ip_set_macipmap.h @@ -0,0 +1,38 @@ +#ifndef __IP_SET_MACIPMAP_H +#define __IP_SET_MACIPMAP_H + +#include "ip_set.h" + +#define SETTYPE_NAME "macipmap" +#define MAX_RANGE 0x0000FFFF + +/* 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; +}; + +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 flags; + unsigned char ethernet[ETH_ALEN]; +}; + +#endif /* __IP_SET_MACIPMAP_H */ diff --git a/extensions/ipset/ip_set_malloc.h b/extensions/ipset/ip_set_malloc.h new file mode 100644 index 0000000..ddf6389 --- /dev/null +++ b/extensions/ipset/ip_set_malloc.h @@ -0,0 +1,143 @@ +#ifndef _IP_SET_MALLOC_H +#define _IP_SET_MALLOC_H + +#ifdef __KERNEL__ + +static size_t max_malloc_size = 0, max_page_size = 0; + +static inline unsigned int init_max_page_size(void) +{ + 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; + } + return 0; +} + +struct harray { + size_t max_elements; + void *arrays[0]; +}; + +static inline void * +__harray_malloc(size_t hashsize, size_t typesize, int 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, int 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 > 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 > max_malloc_size) + vfree(data); + else + kfree(data); +} + +#endif /* __KERNEL__ */ + +#endif /*_IP_SET_MALLOC_H*/ diff --git a/extensions/ipset/ip_set_nethash.c b/extensions/ipset/ip_set_nethash.c new file mode 100644 index 0000000..9e3cd0e --- /dev/null +++ b/extensions/ipset/ip_set_nethash.c @@ -0,0 +1,490 @@ +/* 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 implementing a cidr nethash set */ + +#include +#include +#include +#include +#include +#include +#include "ip_set.h" +#include +#include +#include +#include +#include +#include + +#include + +#include "ip_set_malloc.h" +#include "ip_set_nethash.h" + +static int limit = MAX_RANGE; + +static inline __u32 +jhash_ip(const struct ip_set_nethash *map, uint16_t i, ip_set_ip_t ip) +{ + return jhash_1word(ip, *(((uint32_t *) map->initval) + i)); +} + +static inline __u32 +hash_id_cidr(struct ip_set_nethash *map, + ip_set_ip_t ip, + unsigned char cidr, + ip_set_ip_t *hash_ip) +{ + __u32 id; + u_int16_t i; + ip_set_ip_t *elem; + + *hash_ip = pack(ip, cidr); + + for (i = 0; i < map->probes; i++) { + id = jhash_ip(map, i, *hash_ip) % map->hashsize; + DP("hash key: %u", id); + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); + if (*elem == *hash_ip) + return id; + } + return UINT_MAX; +} + +static inline __u32 +hash_id(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) +{ + struct ip_set_nethash *map = set->data; + __u32 id = UINT_MAX; + int i; + + for (i = 0; i < 30 && map->cidr[i]; i++) { + id = hash_id_cidr(map, ip, map->cidr[i], hash_ip); + if (id != UINT_MAX) + break; + } + return id; +} + +static inline int +__testip_cidr(struct ip_set *set, ip_set_ip_t ip, unsigned char cidr, + ip_set_ip_t *hash_ip) +{ + struct ip_set_nethash *map = set->data; + + return (ip && hash_id_cidr(map, ip, cidr, hash_ip) != UINT_MAX); +} + +static inline int +__testip(struct ip_set *set, ip_set_ip_t ip, ip_set_ip_t *hash_ip) +{ + return (ip && hash_id(set, ip, hash_ip) != UINT_MAX); +} + +static int +testip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_nethash *req = data; + + if (size != sizeof(struct ip_set_req_nethash)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_nethash), + size); + return -EINVAL; + } + return (req->cidr == 32 ? __testip(set, req->ip, hash_ip) + : __testip_cidr(set, req->ip, req->cidr, hash_ip)); +} + +static int +testip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + return __testip(set, + ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr), +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr), +#endif + hash_ip); +} + +static inline int +__addip_base(struct ip_set_nethash *map, ip_set_ip_t ip) +{ + __u32 probe; + u_int16_t i; + ip_set_ip_t *elem; + + 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 (!*elem) { + *elem = ip; + map->elements++; + return 0; + } + } + /* Trigger rehashing */ + return -EAGAIN; +} + +static inline int +__addip(struct ip_set_nethash *map, ip_set_ip_t ip, unsigned char cidr, + ip_set_ip_t *hash_ip) +{ + if (!ip || map->elements >= limit) + return -ERANGE; + + *hash_ip = pack(ip, cidr); + DP("%u.%u.%u.%u/%u, %u.%u.%u.%u", HIPQUAD(ip), cidr, HIPQUAD(*hash_ip)); + + return __addip_base(map, *hash_ip); +} + +static void +update_cidr_sizes(struct ip_set_nethash *map, unsigned char cidr) +{ + unsigned char next; + int i; + + for (i = 0; i < 30 && map->cidr[i]; i++) { + if (map->cidr[i] == cidr) { + return; + } else if (map->cidr[i] < cidr) { + next = map->cidr[i]; + map->cidr[i] = cidr; + cidr = next; + } + } + if (i < 30) + map->cidr[i] = cidr; +} + +static int +addip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_nethash *req = data; + int ret; + + if (size != sizeof(struct ip_set_req_nethash)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_nethash), + size); + return -EINVAL; + } + ret = __addip(set->data, req->ip, req->cidr, hash_ip); + + if (ret == 0) + update_cidr_sizes(set->data, req->cidr); + + return ret; +} + +static int +addip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + struct ip_set_nethash *map = set->data; + int ret = -ERANGE; + ip_set_ip_t ip = ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr); +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr); +#endif + + if (map->cidr[0]) + ret = __addip(map, ip, map->cidr[0], hash_ip); + + return ret; +} + +static int retry(struct ip_set *set) +{ + struct ip_set_nethash *map = set->data; + ip_set_ip_t *elem; + void *members; + u_int32_t i, hashsize = map->hashsize; + int res; + struct ip_set_nethash *tmp; + + if (map->resize == 0) + return -ERANGE; + + again: + res = 0; + + /* Calculate new parameters */ + hashsize += (hashsize * map->resize)/100; + if (hashsize == map->hashsize) + hashsize++; + + ip_set_printk("rehashing of set %s triggered: " + "hashsize grows from %u to %u", + set->name, map->hashsize, hashsize); + + tmp = kmalloc(sizeof(struct ip_set_nethash) + + map->probes * sizeof(uint32_t), GFP_ATOMIC); + if (!tmp) { + DP("out of memory for %d bytes", + sizeof(struct ip_set_nethash) + + map->probes * sizeof(uint32_t)); + return -ENOMEM; + } + tmp->members = harray_malloc(hashsize, sizeof(ip_set_ip_t), GFP_ATOMIC); + if (!tmp->members) { + DP("out of memory for %d bytes", hashsize * sizeof(ip_set_ip_t)); + 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(uint32_t)); + memcpy(tmp->cidr, map->cidr, 30 * sizeof(unsigned char)); + + write_lock_bh(&set->lock); + map = set->data; /* Play safe */ + for (i = 0; i < map->hashsize && res == 0; i++) { + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, i); + if (*elem) + res = __addip_base(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; +} + +static inline int +__delip(struct ip_set_nethash *map, ip_set_ip_t ip, unsigned char cidr, + ip_set_ip_t *hash_ip) +{ + ip_set_ip_t id, *elem; + + if (!ip) + return -ERANGE; + + id = hash_id_cidr(map, ip, cidr, hash_ip); + if (id == UINT_MAX) + return -EEXIST; + + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, id); + *elem = 0; + map->elements--; + return 0; +} + +static int +delip(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_ip) +{ + const struct ip_set_req_nethash *req = data; + + if (size != sizeof(struct ip_set_req_nethash)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_nethash), + size); + return -EINVAL; + } + /* TODO: no garbage collection in map->cidr */ + return __delip(set->data, req->ip, req->cidr, hash_ip); +} + +static int +delip_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_ip, + const u_int32_t *flags, + unsigned char index) +{ + struct ip_set_nethash *map = set->data; + int ret = -ERANGE; + ip_set_ip_t ip = ntohl(flags[index] & IPSET_SRC +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + ? ip_hdr(skb)->saddr + : ip_hdr(skb)->daddr); +#else + ? skb->nh.iph->saddr + : skb->nh.iph->daddr); +#endif + + if (map->cidr[0]) + ret = __delip(map, ip, map->cidr[0], hash_ip); + + return ret; +} + +static int create(struct ip_set *set, const void *data, size_t size) +{ + const struct ip_set_req_nethash_create *req = data; + struct ip_set_nethash *map; + uint16_t i; + + if (size != sizeof(struct ip_set_req_nethash_create)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_nethash_create), + size); + return -EINVAL; + } + + 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_nethash) + + req->probes * sizeof(uint32_t), GFP_KERNEL); + if (!map) { + DP("out of memory for %d bytes", + sizeof(struct ip_set_nethash) + + req->probes * sizeof(uint32_t)); + return -ENOMEM; + } + for (i = 0; i < req->probes; i++) + get_random_bytes(((uint32_t *) map->initval)+i, 4); + map->elements = 0; + map->hashsize = req->hashsize; + map->probes = req->probes; + map->resize = req->resize; + memset(map->cidr, 0, 30 * sizeof(unsigned char)); + map->members = harray_malloc(map->hashsize, sizeof(ip_set_ip_t), GFP_KERNEL); + if (!map->members) { + DP("out of memory for %d bytes", map->hashsize * sizeof(ip_set_ip_t)); + kfree(map); + return -ENOMEM; + } + + set->data = map; + return 0; +} + +static void destroy(struct ip_set *set) +{ + struct ip_set_nethash *map = set->data; + + harray_free(map->members); + kfree(map); + + set->data = NULL; +} + +static void flush(struct ip_set *set) +{ + struct ip_set_nethash *map = set->data; + harray_flush(map->members, map->hashsize, sizeof(ip_set_ip_t)); + memset(map->cidr, 0, 30 * sizeof(unsigned char)); + map->elements = 0; +} + +static void list_header(const struct ip_set *set, void *data) +{ + const struct ip_set_nethash *map = set->data; + struct ip_set_req_nethash_create *header = data; + + header->hashsize = map->hashsize; + header->probes = map->probes; + header->resize = map->resize; +} + +static int list_members_size(const struct ip_set *set) +{ + struct ip_set_nethash *map = set->data; + + return (map->hashsize * sizeof(ip_set_ip_t)); +} + +static void list_members(const struct ip_set *set, void *data) +{ + const struct ip_set_nethash *map = set->data; + ip_set_ip_t i, *elem; + + for (i = 0; i < map->hashsize; i++) { + elem = HARRAY_ELEM(map->members, ip_set_ip_t *, i); + ((ip_set_ip_t *)data)[i] = *elem; + } +} + +static struct ip_set_type ip_set_nethash = { + .typename = SETTYPE_NAME, + .features = IPSET_TYPE_IP | IPSET_DATA_SINGLE, + .protocol_version = IP_SET_PROTOCOL_VERSION, + .create = &create, + .destroy = &destroy, + .flush = &flush, + .reqsize = sizeof(struct ip_set_req_nethash), + .addip = &addip, + .addip_kernel = &addip_kernel, + .retry = &retry, + .delip = &delip, + .delip_kernel = &delip_kernel, + .testip = &testip, + .testip_kernel = &testip_kernel, + .header_size = sizeof(struct ip_set_req_nethash_create), + .list_header = &list_header, + .list_members_size = &list_members_size, + .list_members = &list_members, + .me = THIS_MODULE, +}; + +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"); + +static int __init ip_set_nethash_init(void) +{ + init_max_page_size(); + return ip_set_register_set_type(&ip_set_nethash); +} + +static void __exit ip_set_nethash_fini(void) +{ + /* FIXME: possible race with ip_set_create() */ + ip_set_unregister_set_type(&ip_set_nethash); +} + +module_init(ip_set_nethash_init); +module_exit(ip_set_nethash_fini); diff --git a/extensions/ipset/ip_set_nethash.h b/extensions/ipset/ip_set_nethash.h new file mode 100644 index 0000000..2f347b9 --- /dev/null +++ b/extensions/ipset/ip_set_nethash.h @@ -0,0 +1,55 @@ +#ifndef __IP_SET_NETHASH_H +#define __IP_SET_NETHASH_H + +#include "ip_set.h" + +#define SETTYPE_NAME "nethash" +#define MAX_RANGE 0x0000FFFF + +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 */ + unsigned char cidr[30]; /* CIDR sizes */ + void *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; + unsigned char cidr; +}; + +static unsigned char shifts[] = {255, 253, 249, 241, 225, 193, 129, 1}; + +static inline ip_set_ip_t +pack(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_NETHASH_H */ diff --git a/extensions/ipset/ip_set_portmap.c b/extensions/ipset/ip_set_portmap.c new file mode 100644 index 0000000..c44b8db --- /dev/null +++ b/extensions/ipset/ip_set_portmap.c @@ -0,0 +1,341 @@ +/* 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 implementing a port set type as a bitmap */ + +#include +#include +#include +#include +#include +#include +#include +#include "ip_set.h" +#include +#include +#include +#include + +#include + +#include "ip_set_portmap.h" + +/* We must handle non-linear skbs */ +static inline ip_set_ip_t +get_port(const struct sk_buff *skb, u_int32_t flags) +{ +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + struct iphdr *iph = ip_hdr(skb); +#else + struct iphdr *iph = skb->nh.iph; +#endif + 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 LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + if (skb_copy_bits(skb, ip_hdr(skb)->ihl*4, &tcph, sizeof(tcph)) < 0) +#else + if (skb_copy_bits(skb, skb->nh.iph->ihl*4, &tcph, sizeof(tcph)) < 0) +#endif + /* No choice either */ + return INVALID_PORT; + + return ntohs(flags & IPSET_SRC ? + tcph.source : tcph.dest); + } + case IPPROTO_UDP: { + struct udphdr udph; + + if (offset) + return INVALID_PORT; + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,22) + if (skb_copy_bits(skb, ip_hdr(skb)->ihl*4, &udph, sizeof(udph)) < 0) +#else + if (skb_copy_bits(skb, skb->nh.iph->ihl*4, &udph, sizeof(udph)) < 0) +#endif + /* No choice either */ + return INVALID_PORT; + + return ntohs(flags & IPSET_SRC ? + udph.source : udph.dest); + } + default: + return INVALID_PORT; + } +} + +static inline int +__testport(struct ip_set *set, ip_set_ip_t port, ip_set_ip_t *hash_port) +{ + struct ip_set_portmap *map = set->data; + + if (port < map->first_port || port > map->last_port) + return -ERANGE; + + *hash_port = port; + DP("set: %s, port:%u, %u", set->name, port, *hash_port); + return !!test_bit(port - map->first_port, map->members); +} + +static int +testport(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_port) +{ + const struct ip_set_req_portmap *req = data; + + if (size != sizeof(struct ip_set_req_portmap)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_portmap), + size); + return -EINVAL; + } + return __testport(set, req->port, hash_port); +} + +static int +testport_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_port, + const u_int32_t *flags, + unsigned char index) +{ + int res; + ip_set_ip_t port = get_port(skb, flags[index]); + + DP("flag %s port %u", flags[index] & IPSET_SRC ? "SRC" : "DST", port); + if (port == INVALID_PORT) + return 0; + + res = __testport(set, port, hash_port); + + return (res < 0 ? 0 : res); +} + +static inline int +__addport(struct ip_set *set, ip_set_ip_t port, ip_set_ip_t *hash_port) +{ + struct ip_set_portmap *map = set->data; + + if (port < map->first_port || port > map->last_port) + return -ERANGE; + if (test_and_set_bit(port - map->first_port, map->members)) + return -EEXIST; + + *hash_port = port; + DP("port %u", port); + return 0; +} + +static int +addport(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_port) +{ + const struct ip_set_req_portmap *req = data; + + if (size != sizeof(struct ip_set_req_portmap)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_portmap), + size); + return -EINVAL; + } + return __addport(set, req->port, hash_port); +} + +static int +addport_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_port, + const u_int32_t *flags, + unsigned char index) +{ + ip_set_ip_t port = get_port(skb, flags[index]); + + if (port == INVALID_PORT) + return -EINVAL; + + return __addport(set, port, hash_port); +} + +static inline int +__delport(struct ip_set *set, ip_set_ip_t port, ip_set_ip_t *hash_port) +{ + struct ip_set_portmap *map = set->data; + + if (port < map->first_port || port > map->last_port) + return -ERANGE; + if (!test_and_clear_bit(port - map->first_port, map->members)) + return -EEXIST; + + *hash_port = port; + DP("port %u", port); + return 0; +} + +static int +delport(struct ip_set *set, const void *data, size_t size, + ip_set_ip_t *hash_port) +{ + const struct ip_set_req_portmap *req = data; + + if (size != sizeof(struct ip_set_req_portmap)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_portmap), + size); + return -EINVAL; + } + return __delport(set, req->port, hash_port); +} + +static int +delport_kernel(struct ip_set *set, + const struct sk_buff *skb, + ip_set_ip_t *hash_port, + const u_int32_t *flags, + unsigned char index) +{ + ip_set_ip_t port = get_port(skb, flags[index]); + + if (port == INVALID_PORT) + return -EINVAL; + + return __delport(set, port, hash_port); +} + +static int create(struct ip_set *set, const void *data, size_t size) +{ + int newbytes; + const struct ip_set_req_portmap_create *req = data; + struct ip_set_portmap *map; + + if (size != sizeof(struct ip_set_req_portmap_create)) { + ip_set_printk("data length wrong (want %zu, have %zu)", + sizeof(struct ip_set_req_portmap_create), + size); + return -EINVAL; + } + + DP("from %u to %u", req->from, req->to); + + if (req->from > req->to) { + DP("bad port range"); + return -ENOEXEC; + } + + if (req->to - req->from > MAX_RANGE) { + ip_set_printk("range too big (max %d ports)", + MAX_RANGE+1); + return -ENOEXEC; + } + + map = kmalloc(sizeof(struct ip_set_portmap), GFP_KERNEL); + if (!map) { + DP("out of memory for %d bytes", + sizeof(struct ip_set_portmap)); + return -ENOMEM; + } + map->first_port = req->from; + map->last_port = req->to; + newbytes = bitmap_bytes(req->from, req->to); + map->members = kmalloc(newbytes, GFP_KERNEL); + if (!map->members) { + DP("out of memory for %d bytes", newbytes); + kfree(map); + return -ENOMEM; + } + memset(map->members, 0, newbytes); + + set->data = map; + return 0; +} + +static void destroy(struct ip_set *set) +{ + struct ip_set_portmap *map = set->data; + + kfree(map->members); + kfree(map); + + set->data = NULL; +} + +static void flush(struct ip_set *set) +{ + struct ip_set_portmap *map = set->data; + memset(map->members, 0, bitmap_bytes(map->first_port, map->last_port)); +} + +static void list_header(const struct ip_set *set, void *data) +{ + const struct ip_set_portmap *map = set->data; + struct ip_set_req_portmap_create *header = data; + + DP("list_header %u %u", map->first_port, map->last_port); + + header->from = map->first_port; + header->to = map->last_port; +} + +static int list_members_size(const struct ip_set *set) +{ + const struct ip_set_portmap *map = set->data; + + return bitmap_bytes(map->first_port, map->last_port); +} + +static void list_members(const struct ip_set *set, void *data) +{ + const struct ip_set_portmap *map = set->data; + int bytes = bitmap_bytes(map->first_port, map->last_port); + + memcpy(data, map->members, bytes); +} + +static struct ip_set_type ip_set_portmap = { + .typename = SETTYPE_NAME, + .features = IPSET_TYPE_PORT | IPSET_DATA_SINGLE, + .protocol_version = IP_SET_PROTOCOL_VERSION, + .create = &create, + .destroy = &destroy, + .flush = &flush, + .reqsize = sizeof(struct ip_set_req_portmap), + .addip = &addport, + .addip_kernel = &addport_kernel, + .delip = &delport, + .delip_kernel = &delport_kernel, + .testip = &testport, + .testip_kernel = &testport_kernel, + .header_size = sizeof(struct ip_set_req_portmap_create), + .list_header = &list_header, + .list_members_size = &list_members_size, + .list_members = &list_members, + .me = THIS_MODULE, +}; + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("portmap type of IP sets"); + +static int __init ip_set_portmap_init(void) +{ + return ip_set_register_set_type(&ip_set_portmap); +} + +static void __exit ip_set_portmap_fini(void) +{ + /* FIXME: possible race with ip_set_create() */ + ip_set_unregister_set_type(&ip_set_portmap); +} + +module_init(ip_set_portmap_init); +module_exit(ip_set_portmap_fini); diff --git a/extensions/ipset/ip_set_portmap.h b/extensions/ipset/ip_set_portmap.h new file mode 100644 index 0000000..0c08974 --- /dev/null +++ b/extensions/ipset/ip_set_portmap.h @@ -0,0 +1,25 @@ +#ifndef __IP_SET_PORTMAP_H +#define __IP_SET_PORTMAP_H + +#include "ip_set.h" + +#define SETTYPE_NAME "portmap" +#define MAX_RANGE 0x0000FFFF +#define INVALID_PORT (MAX_RANGE + 1) + +struct ip_set_portmap { + void *members; /* the portmap proper */ + ip_set_ip_t first_port; /* host byte order, included in range */ + ip_set_ip_t last_port; /* host byte order, included in range */ +}; + +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 port; +}; + +#endif /* __IP_SET_PORTMAP_H */ diff --git a/extensions/ipset/ipset.8 b/extensions/ipset/ipset.8 new file mode 100644 index 0000000..2c4edcb --- /dev/null +++ b/extensions/ipset/ipset.8 @@ -0,0 +1,470 @@ +.TH IPSET 8 "Feb 05, 2004" "" "" +.\" +.\" Man page written by Jozsef Kadlecsik +.\" +.\" This program is free software; you can redistribute it and/or modify +.\" it under the terms of the GNU General Public License as published by +.\" the Free Software Foundation; either version 2 of the License, or +.\" (at your option) any later version. +.\" +.\" This program is distributed in the hope that it will be useful, +.\" but WITHOUT ANY WARRANTY; without even the implied warranty of +.\" MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +.\" GNU General Public License for more details. +.\" +.\" You should have received a copy of the GNU General Public License +.\" along with this program; if not, write to the Free Software +.\" Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. +.\" +.\" +.SH NAME +ipset \- administration tool for IP sets +.SH SYNOPSIS +.BR "ipset -N " "set type-specification [options]" +.br +.BR "ipset -[XFLSHh] " "[set] [options]" +.br +.BR "ipset -[EW] " "from-set to-set" +.br +.BR "ipset -[ADU] " "set entry" +.br +.BR "ipset -B " "set entry -b binding" +.br +.BR "ipset -T " "set entry [-b binding]" +.br +.BR "ipset -R " +.SH DESCRIPTION +.B ipset +is used to set up, maintain and inspect so called IP sets in the Linux +kernel. Depending on the type, an IP set may store IP addresses, (TCP/UDP) +port numbers or additional informations besides IP addresses: the word IP +means a general term here. See the set type definitions below. +.P +Any entry in a set can be bound to another set, which forms a relationship +between a set element and the set it is bound to. In order to define a +binding it is not required that the entry be already added to the set. +The sets may have a default binding, which is valid for every set element +for which there is no binding defined at all. +.P +IP set bindings pointing to sets and iptables matches and targets +referring to sets creates references, which protects the given sets in +the kernel. A set cannot be removed (destroyed) while there is a single +reference pointing to it. +.SH OPTIONS +The options that are recognized by +.B ipset +can be divided into several different groups. +.SS COMMANDS +These options specify the specific action to perform. Only one of them +can be specified on the command line unless otherwise specified +below. For all the long versions of the command and option names, you +need to use only enough letters to ensure that +.B ipset +can differentiate it from all other options. +.TP +.BI "-N, --create " "\fIsetname\fP type type-specific-options" +Create a set identified with setname and specified type. +Type-specific options must be supplied. +.TP +.BI "-X, --destroy " "[\fIsetname\fP]" +Destroy the specified set, or all sets if none or the keyword +.B +:all: +is specified. +Before destroying the set, all bindings belonging to the +set elements and the default binding of the set are removed. + +If the set has got references, nothing is done. +.TP +.BI "-F, --flush " "[\fIsetname\fP]" +Delete all entries from the specified set, or flush +all sets if none or the keyword +.B +:all: +is given. Bindings are not affected by the flush operation. +.TP +.BI "-E, --rename " "\fIfrom-setname\fP \fIto-setname\fP" +Rename a set. Set identified by to-setname must not exist. +.TP +.BI "-W, --swap " "\fIfrom-setname\fP \fIto-setname\fP" +Swap two sets as they referenced in the Linux kernel. +.B +iptables +rules or +.B +ipset +bindings pointing to the content of from-setname will point to +the content of to-setname and vice versa. Both sets must exist. +.TP +.BI "-L, --list " "[\fIsetname\fP]" +List the entries and bindings for the specified set, or for +all sets if none or the keyword +.B +:all: +is given. The +.B "-n, --numeric" +option can be used to suppress name lookups and generate numeric +output. When the +.B "-s, --sorted" +option is given, the entries are listed sorted (if the given set +type supports the operation). +.TP +.BI "-S, --save " "[\fIsetname\fP]" +Save the given set, or all sets if none or the keyword +.B +:all: +is specified to stdout in a format that --restore can read. +.TP +.BI "-R, --restore " +Restore a saved session generated by --save. The saved session +can be fed from stdin. + +When generating a session file please note that the supported commands +(create set, add element, bind) must appear in a strict order: first create +the set, then add all elements. Then create the next set, add all its elements +and so on. Finally you can list all binding commands. Also, it is a restore +operation, so the sets being restored must not exist. +.TP +.BI "-A, --add " "\fIsetname\fP \fIIP\fP" +Add an IP to a set. +.TP +.BI "-D, --del " "\fIsetname\fP \fIIP\fP" +Delete an IP from a set. +.TP +.BI "-T, --test " "\fIsetname\fP \fIIP +Test wether an IP is in a set or not. Exit status number is zero +if the tested IP is in the set and nonzero if it is missing from +the set. +.TP +.BI "-T, --test " "\fIsetname\fP \fIIP\fP \fI--binding\fP \fIto-setname\fP" +Test wether the IP belonging to the set points to the specified binding. +Exit status number is zero if the binding points to the specified set, +otherwise it is nonzero. The keyword +.B +:default: +can be used to test the default binding of the set. +.TP +.BI "-B, --bind " "\fIsetname\fP \fIIP\fP \fI--binding\fP \fIto-setname\fP" +Bind the IP in setname to to-setname. +.TP +.BI "-U, --unbind " "\fIsetname\fP \fIIP\fP" +Delete the binding belonging to IP in set setname. +.TP +.BI "-H, --help " "[settype]" +Print help and settype specific help if settype specified. +.P +At the +.B +-B, -U +and +.B +-T +commands you can use the token +.B +:default: +to bind, unbind or test the default binding of a set instead +of an IP. At the +.B +-U +command you can use the token +.B +:all: +to destroy the bindings of all elements of a set. +.SS "OTHER OPTIONS" +The following additional options can be specified: +.TP +.B "-b, --binding setname" +The option specifies the value of the binding for the +.B "-B" +binding command, for which it is a mandatory option. +You can use it in the +.B "-T" +test command as well to test bindings. +.TP +.B "-s, --sorted" +Sorted output. When listing sets, entries are listed sorted. +.TP +.B "-n, --numeric" +Numeric output. When listing sets, bindings, IP addresses and +port numbers will be printed in numeric format. By default the +program will try to display them as host names, network names +or services (whenever applicable), which can trigger +.B +slow +DNS +lookups. +.TP +.B "-q, --quiet" +Suppress any output to stdout and stderr. ipset will still return +possible errors. +.SH SET TYPES +ipset supports the following set types: +.SS ipmap +The ipmap set type uses a memory range, where each bit represents +one IP address. An ipmap set can store up to 65536 (B-class network) +IP addresses. The ipmap set type is very fast and memory cheap, great +for use when one want to match certain IPs in a range. Using the +.B "--netmask" +option with a CIDR netmask value between 0-32 when creating an ipmap +set, you will be able to store and match network addresses: i.e an +IP address will be in the set if the value resulted by masking the address +with the specified netmask can be found in the set. +.P +Options to use when creating an ipmap set: +.TP +.BR "--from " from-IP +.TP +.BR "--to " to-IP +Create an ipmap set from the specified range. +.TP +.BR "--network " IP/mask +Create an ipmap set from the specified network. +.TP +.BR "--netmask " CIDR-netmask +When the optional +.B "--netmask" +parameter specified, network addresses will be +stored in the set instead of IP addresses, and the from-IP parameter +must be a network address. +.SS macipmap +The macipmap set type uses a memory range, where each 8 bytes +represents one IP and a MAC addresses. A macipmap set type can store +up to 65536 (B-class network) IP addresses with MAC. +When adding an entry to a macipmap set, you must specify the entry as +.I IP:MAC. +When deleting or testing macipmap entries, the +.I :MAC +part is not mandatory. (The old "%" separation token instead of ":", i.e +IP%MAC is accepted as well.) +.P +Options to use when creating an macipmap set: +.TP +.BR "--from " from-IP +.TP +.BR "--to " to-IP +Create a macipmap set from the specified range. +.TP +.BR "--network " IP/mask +Create a macipmap set from the specified network. +.TP +.BR "--matchunset" +When the optional +.B "--matchunset" +parameter specified, IP addresses which could be stored +in the set but not set yet, will always match. +.P +Please note, the +.I +set +and +.I +SET +netfilter kernel modules +.B +always +use the source MAC address from the packet to match, add or delete +entries from a macipmap type of set. +.SS portmap +The portmap set type uses a memory range, where each bit represents +one port. A portmap set type can store up to 65536 ports. +The portmap set type is very fast and memory cheap. +.P +Options to use when creating an portmap set: +.TP +.BR "--from " from-port +.TP +.BR "--to " to-port +Create a portmap set from the specified range. +.SS iphash +The iphash set type uses a hash to store IP addresses. +In order to avoid clashes in the hash double-hashing, and as a last +resort, dynamic growing of the hash performed. The iphash set type is +great to store random addresses. By supplyig the +.B "--netmask" +option with a CIDR netmask value between 0-32 at creating the set, +you will be able to store and match network addresses instead: i.e +an IP address will be in the set if the value of the address +masked with the specified netmask can be found in the set. +.P +Options to use when creating an iphash set: +.TP +.BR "--hashsize " hashsize +The initial hash size (default 1024) +.TP +.BR "--probes " probes +How many times try to resolve clashing at adding an IP to the hash +by double-hashing (default 8). +.TP +.BR "--resize " percent +Increase the hash size by this many percent (default 50) when adding +an IP to the hash could not be performed after +.B +probes +number of double-hashing. +.TP +.BR "--netmask " CIDR-netmask +When the optional +.B "--netmask" +parameter specified, network addresses will be +stored in the set instead of IP addresses. +.P +The iphash type of sets can store up to 65536 entries. If a set is full, +no new entries can be added to it. +.P +Sets created by zero valued resize parameter won't be resized at all. +The lookup time in an iphash type of set approximately linearly grows with +the value of the +.B +probes +parameter. At the same time higher +.B +probes +values result a better utilized hash while smaller values +produce a larger, sparse hash. +.SS nethash +The nethash set type uses a hash to store different size of +network addresses. The +.I +IP +"address" used in the ipset commands must be in the form +.I +IP-address/cidr-size +where the CIDR block size must be in the inclusive range of 1-31. +In order to avoid clashes in the hash +double-hashing, and as a last resort, dynamic growing of the hash performed. +.P +Options to use when creating an nethash set: +.TP +.BR "--hashsize " hashsize +The initial hash size (default 1024) +.TP +.BR "--probes " probes +How many times try to resolve clashing at adding an IP to the hash +by double-hashing (default 4). +.TP +.BR "--resize " percent +Increase the hash size by this many percent (default 50) when adding +an IP to the hash could not be performed after +.P +The nethash type of sets can store up to 65536 entries. If a set is full, +no new entries can be added to it. +.P +An IP address will be in a nethash type of set if it is in any of the +netblocks added to the set and the matching always start from the smallest +size of netblock (most specific netmask) to the biggest ones (least +specific netmasks). When adding/deleting IP addresses +to a nethash set by the +.I +SET +netfilter kernel module, it will be added/deleted by the smallest +netblock size which can be found in the set. +.P +The lookup time in a nethash type of set is approximately linearly +grows with the times of the +.B +probes +parameter and the number of different mask parameters in the hash. +Otherwise the same speed and memory efficiency comments applies here +as at the iphash type. +.SS ipporthash +The ipporthash set type uses a hash to store IP address and port pairs. +In order to avoid clashes in the hash double-hashing, and as a last +resort, dynamic growing of the hash performed. An ipporthash set can +store up to 65536 (B-class network) IP addresses with all possible port +values. When adding, deleting and testing values in an ipporthash type of +set, the entries must be specified as +.B +"IP:port". +(Old "IP%port" format accepted as well.) +.P +The ipporthash types of sets evaluates two src/dst parameters of the +.I +set +match and +.I +SET +target. +.P +Options to use when creating an ipporthash set: +.TP +.BR "--from " from-IP +.TP +.BR "--to " to-IP +Create an ipporthash set from the specified range. +.TP +.BR "--network " IP/mask +Create an ipporthash set from the specified network. +.TP +.BR "--hashsize " hashsize +The initial hash size (default 1024) +.TP +.BR "--probes " probes +How many times try to resolve clashing at adding an IP to the hash +by double-hashing (default 8). +.TP +.BR "--resize " percent +Increase the hash size by this many percent (default 50) when adding +an IP to the hash could not be performed after +.B +probes +number of double-hashing. +.P +The same resizing, speed and memory efficiency comments applies here +as at the iphash type. +.SS iptree +The iptree set type uses a tree to store IP addresses, optionally +with timeout values. +.P +Options to use when creating an iptree set: +.TP +.BR "--timeout " value +The timeout value for the entries in seconds (default 0) +.P +If a set was created with a nonzero valued +.B "--timeout" +parameter then one may add IP addresses to the set with a specific +timeout value using the syntax +.I IP:timeout-value. +Similarly to the hash types, the iptree type of sets can store up to 65536 +entries. +.SS iptreemap +The iptreemap set type uses a tree to store IP addresses or networks, +where the last octet of an IP address are stored in a bitmap. +As input entry, you can add IP addresses, CIDR blocks or network ranges +to the set. Network ranges can be specified in the format +.I IP1:IP2 +.P +Options to use when creating an iptreemap set: +.TP +.BR "--gc " value +How often the garbage collection should be called, in seconds (default 300) +.SH GENERAL RESTRICTIONS +Setnames starting with colon (:) cannot be defined. Zero valued set +entries cannot be used with hash type of sets. +.SH COMMENTS +If you want to store same size subnets from a given network +(say /24 blocks from a /8 network), use the ipmap set type. +If you want to store random same size networks (say random /24 blocks), +use the iphash set type. If you have got random size of netblocks, +use nethash. +.SH DIAGNOSTICS +Various error messages are printed to standard error. The exit code +is 0 for correct functioning. Errors which appear to be caused by +invalid or abused command line parameters cause an exit code of 2, and +other errors cause an exit code of 1. +.SH BUGS +Bugs? No, just funny features. :-) +OK, just kidding... +.SH SEE ALSO +.BR iptables (8), +.SH AUTHORS +Jozsef Kadlecsik wrote ipset, which is based on ippool by +Joakim Axelsson, Patrick Schaaf and Martin Josefsson. +.P +Sven Wegener wrote the iptreemap type. +.SH LAST REMARK +.BR "I stand on the shoulder of giants." +.\" .. and did I mention that we are incredibly cool people? +.\" .. sexy, too .. +.\" .. witty, charming, powerful .. +.\" .. and most of all, modest .. diff --git a/extensions/ipset/ipset.c b/extensions/ipset/ipset.c new file mode 100644 index 0000000..2180493 --- /dev/null +++ b/extensions/ipset/ipset.c @@ -0,0 +1,2309 @@ +/* Copyright 2000-2002 Joakim Axelsson (gozem@linux.nu) + * Patrick Schaaf (bof@bof.de) + * Copyright 2003-2004 Jozsef Kadlecsik (kadlec@blackhole.kfki.hu) + * + * 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. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +/* #include */ + +#include "ipset.h" + +#ifndef PROC_SYS_MODPROBE +#define PROC_SYS_MODPROBE "/proc/sys/kernel/modprobe" +#endif + +#define IPSET_VERSION "2.3.2a" + +char program_name[] = "ipset"; +char program_version[] = IPSET_VERSION; +const char *xtables_libdir = XTABLES_LIBDIR; + +/* The list of loaded set types */ +static struct settype *all_settypes = NULL; + +/* Array of sets */ +struct set **set_list = NULL; +ip_set_id_t max_sets = 0; + +/* Suppress output to stdout and stderr? */ +static int option_quiet = 0; + +/* Data for restore mode */ +static int restore = 0; +void *restore_data = NULL; +struct ip_set_restore *restore_set = NULL; +size_t restore_offset = 0; +socklen_t restore_size; +unsigned int g_line = 0; + +#define TEMPFILE_PATTERN "/ipsetXXXXXX" + +#ifdef IPSET_DEBUG +int option_debug = 0; +#endif + +#define OPTION_OFFSET 256 +static unsigned int global_option_offset = 0; + +/* Most of these command parsing functions are borrowed from iptables.c */ + +static const char cmdflags[] = { ' ', /* CMD_NONE */ + 'N', 'X', 'F', 'E', 'W', 'L', 'S', 'R', + 'A', 'D', 'T', 'B', 'U', 'H', 'V', +}; + +/* Options */ +#define OPT_NONE 0x0000U +#define OPT_NUMERIC 0x0001U /* -n */ +#define OPT_SORTED 0x0002U /* -s */ +#define OPT_QUIET 0x0004U /* -q */ +#define OPT_DEBUG 0x0008U /* -z */ +#define OPT_BINDING 0x0010U /* -b */ +#define NUMBER_OF_OPT 5 +static const char optflags[] = + { 'n', 's', 'q', 'z', 'b' }; + +static struct option opts_long[] = { + /* set operations */ + {"create", 1, 0, 'N'}, + {"destroy", 2, 0, 'X'}, + {"flush", 2, 0, 'F'}, + {"rename", 1, 0, 'E'}, + {"swap", 1, 0, 'W'}, + {"list", 2, 0, 'L'}, + + {"save", 2, 0, 'S'}, + {"restore", 0, 0, 'R'}, + + /* ip in set operations */ + {"add", 1, 0, 'A'}, + {"del", 1, 0, 'D'}, + {"test", 1, 0, 'T'}, + + /* binding operations */ + {"bind", 1, 0, 'B'}, + {"unbind", 1, 0, 'U'}, + + /* free options */ + {"numeric", 0, 0, 'n'}, + {"sorted", 0, 0, 's'}, + {"quiet", 0, 0, 'q'}, + {"binding", 1, 0, 'b'}, + +#ifdef IPSET_DEBUG + /* debug (if compiled with it) */ + {"debug", 0, 0, 'z'}, +#endif + + /* version and help */ + {"version", 0, 0, 'V'}, + {"help", 2, 0, 'H'}, + + /* end */ + {0} +}; + +static char opts_short[] = + "-N:X::F::E:W:L::S::RA:D:T:B:U:nsqzb:Vh::H::"; + +/* Table of legal combinations of commands and options. If any of the + * given commands make an option legal, that option is legal. + * Key: + * + compulsory + * x illegal + * optional + */ + +static char commands_v_options[NUMBER_OF_CMD][NUMBER_OF_OPT] = { + /* -n -s -q -z -b */ + /*CREATE*/ {'x', 'x', ' ', ' ', 'x'}, + /*DESTROY*/ {'x', 'x', ' ', ' ', 'x'}, + /*FLUSH*/ {'x', 'x', ' ', ' ', 'x'}, + /*RENAME*/ {'x', 'x', ' ', ' ', 'x'}, + /*SWAP*/ {'x', 'x', ' ', ' ', 'x'}, + /*LIST*/ {' ', ' ', 'x', ' ', 'x'}, + /*SAVE*/ {'x', 'x', ' ', ' ', 'x'}, + /*RESTORE*/ {'x', 'x', ' ', ' ', 'x'}, + /*ADD*/ {'x', 'x', ' ', ' ', 'x'}, + /*DEL*/ {'x', 'x', ' ', ' ', 'x'}, + /*TEST*/ {'x', 'x', ' ', ' ', ' '}, + /*BIND*/ {'x', 'x', ' ', ' ', '+'}, + /*UNBIND*/ {'x', 'x', ' ', ' ', 'x'}, + /*HELP*/ {'x', 'x', 'x', ' ', 'x'}, + /*VERSION*/ {'x', 'x', 'x', ' ', 'x'}, +}; + +/* Main parser function */ +int parse_commandline(int argc, char *argv[]); + +static void exit_tryhelp(int status) +{ + fprintf(stderr, + "Try `%s -H' or '%s --help' for more information.\n", + program_name, program_name); + exit(status); +} + +void exit_error(enum exittype status, const char *msg, ...) +{ + va_list args; + + if (!option_quiet) { + va_start(args, msg); + fprintf(stderr, "%s v%s: ", program_name, program_version); + vfprintf(stderr, msg, args); + va_end(args); + fprintf(stderr, "\n"); + if (g_line) + fprintf(stderr, "Restore failed at line %u:\n", g_line); + if (status == PARAMETER_PROBLEM) + exit_tryhelp(status); + if (status == VERSION_PROBLEM) + fprintf(stderr, + "Perhaps %s or your kernel needs to be upgraded.\n", + program_name); + } + + exit(status); +} + +static void ipset_printf(const char *msg, ...) +{ + va_list args; + + if (!option_quiet) { + va_start(args, msg); + vfprintf(stdout, msg, args); + va_end(args); + fprintf(stdout, "\n"); + } +} + +static void generic_opt_check(int command, int options) +{ + int i, j, legal = 0; + + /* Check that commands are valid with options. Complicated by the + * fact that if an option is legal with *any* command given, it is + * legal overall (ie. -z and -l). + */ + for (i = 0; i < NUMBER_OF_OPT; i++) { + legal = 0; /* -1 => illegal, 1 => legal, 0 => undecided. */ + + for (j = 1; j <= NUMBER_OF_CMD; j++) { + if (command != j) + continue; + + if (!(options & (1 << i))) { + if (commands_v_options[j-1][i] == '+') + exit_error(PARAMETER_PROBLEM, + "You need to supply the `-%c' " + "option for this command\n", + optflags[i]); + } else { + if (commands_v_options[j-1][i] != 'x') + legal = 1; + else if (legal == 0) + legal = -1; + } + } + if (legal == -1) + exit_error(PARAMETER_PROBLEM, + "Illegal option `-%c' with this command\n", + optflags[i]); + } +} + +static char opt2char(int option) +{ + const char *ptr; + for (ptr = optflags; option > 1; option >>= 1, ptr++); + + return *ptr; +} + +static char cmd2char(int option) +{ + if (option <= CMD_NONE || option > NUMBER_OF_CMD) + return ' '; + + return cmdflags[option]; +} + +/* From iptables.c ... */ +static char *get_modprobe(void) +{ + int procfile; + char *ret; + +#define PROCFILE_BUFSIZ 1024 + procfile = open(PROC_SYS_MODPROBE, O_RDONLY); + if (procfile < 0) + return NULL; + + ret = (char *) malloc(PROCFILE_BUFSIZ); + if (ret) { + memset(ret, 0, PROCFILE_BUFSIZ); + switch (read(procfile, ret, PROCFILE_BUFSIZ)) { + case -1: goto fail; + case PROCFILE_BUFSIZ: goto fail; /* Partial read. Wierd */ + } + if (ret[strlen(ret)-1]=='\n') + ret[strlen(ret)-1]=0; + close(procfile); + return ret; + } + fail: + free(ret); + close(procfile); + return NULL; +} + +static int ipset_insmod(const char *modname, const char *modprobe) +{ + char *buf = NULL; + char *argv[3]; + struct stat junk; + int status; + + if (!stat(modprobe, &junk)) { + /* Try to read out of the kernel */ + buf = get_modprobe(); + if (!buf) + return -1; + modprobe = buf; + } + + switch (fork()) { + case 0: + argv[0] = (char *)modprobe; + argv[1] = (char *)modname; + argv[2] = NULL; + execv(argv[0], argv); + + /* Should not reach */ + exit(1); + case -1: + return -1; + + default: /* parent */ + wait(&status); + } + + free(buf); + + if (WIFEXITED(status) && WEXITSTATUS(status) == 0) + return 0; + return -1; +} + +static int kernel_getsocket(void) +{ + int sockfd = -1; + + sockfd = socket(AF_INET, SOCK_RAW, IPPROTO_RAW); + if (sockfd < 0) + exit_error(OTHER_PROBLEM, + "You need to be root to perform this command."); + + return sockfd; +} + +static void kernel_error(unsigned cmd, int err) +{ + unsigned int i; + struct translate_error { + int err; + unsigned cmd; + char *message; + } table[] = + { /* Generic error codes */ + { EPERM, 0, "Missing capability" }, + { EBADF, 0, "Invalid socket option" }, + { EINVAL, 0, "Size mismatch for expected socket data" }, + { ENOMEM, 0, "Not enough memory" }, + { EFAULT, 0, "Failed to copy data" }, + { EPROTO, 0, "ipset kernel/userspace version mismatch" }, + { EBADMSG, 0, "Unknown command" }, + /* Per command error codes */ + /* Reserved ones for add/del/test to handle internally: + * EEXIST + */ + { ENOENT, CMD_CREATE, "Unknown set type" }, + { ENOENT, 0, "Unknown set" }, + { EAGAIN, 0, "Sets are busy, try again later" }, + { ERANGE, CMD_CREATE, "No free slot remained to add a new set" }, + { ERANGE, 0, "IP/port is outside of the set" }, + { ENOEXEC, CMD_CREATE, "Invalid parameters to create a set" }, + { ENOEXEC, CMD_SWAP, "Sets with different types cannot be swapped" }, + { EEXIST, CMD_CREATE, "Set already exists" }, + { EEXIST, CMD_RENAME, "Set with new name already exists" }, + { EBUSY, 0, "Set is in use, operation not permitted" }, + }; + for (i = 0; i < sizeof(table)/sizeof(struct translate_error); i++) { + if ((table[i].cmd == cmd || table[i].cmd == 0) + && table[i].err == err) + exit_error(err == EPROTO ? VERSION_PROBLEM + : OTHER_PROBLEM, + table[i].message); + } + exit_error(OTHER_PROBLEM, "Error from kernel: %s", strerror(err)); +} + +static inline int wrapped_getsockopt(void *data, socklen_t *size) +{ + int res; + int sockfd = kernel_getsocket(); + + /* Send! */ + res = getsockopt(sockfd, SOL_IP, SO_IP_SET, data, size); + if (res != 0 + && errno == ENOPROTOOPT + && ipset_insmod("ip_set", "/sbin/modprobe") == 0) + res = getsockopt(sockfd, SOL_IP, SO_IP_SET, data, size); + DP("res=%d errno=%d", res, errno); + + return res; +} + +static inline int wrapped_setsockopt(void *data, socklen_t size) +{ + int res; + int sockfd = kernel_getsocket(); + + /* Send! */ + res = setsockopt(sockfd, SOL_IP, SO_IP_SET, data, size); + if (res != 0 + && errno == ENOPROTOOPT + && ipset_insmod("ip_set", "/sbin/modprobe") == 0) + res = setsockopt(sockfd, SOL_IP, SO_IP_SET, data, size); + DP("res=%d errno=%d", res, errno); + + return res; +} + +static void kernel_getfrom(unsigned cmd, void *data, socklen_t * size) +{ + int res = wrapped_getsockopt(data, size); + + if (res != 0) + kernel_error(cmd, errno); +} + +static int kernel_sendto_handleerrno(unsigned cmd, unsigned op, + void *data, socklen_t size) +{ + int res = wrapped_setsockopt(data, size); + + if (res != 0) { + if (errno == EEXIST) + return -1; + else + kernel_error(cmd, errno); + } + + return 0; /* all ok */ +} + +static void kernel_sendto(unsigned cmd, void *data, size_t size) +{ + int res = wrapped_setsockopt(data, size); + + if (res != 0) + kernel_error(cmd, errno); +} + +static int kernel_getfrom_handleerrno(unsigned cmd, void *data, size_t * size) +{ + int res = wrapped_getsockopt(data, size); + + if (res != 0) { + if (errno == EAGAIN) + return -1; + else + kernel_error(cmd, errno); + } + + return 0; /* all ok */ +} + +static void check_protocolversion(void) +{ + struct ip_set_req_version req_version; + socklen_t size = sizeof(struct ip_set_req_version); + int sockfd = kernel_getsocket(); + int res; + + req_version.op = IP_SET_OP_VERSION; + res = getsockopt(sockfd, SOL_IP, SO_IP_SET, &req_version, &size); + + if (res != 0) { + ipset_printf("I'm of protocol version %u.\n" + "Kernel module is not loaded in, " + "cannot verify kernel version.", + IP_SET_PROTOCOL_VERSION); + return; + } + if (req_version.version != IP_SET_PROTOCOL_VERSION) + exit_error(OTHER_PROBLEM, + "Kernel ipset code is of protocol version %u." + "I'm of protocol version %u.\n" + "Please upgrade your kernel and/or ipset(8) utillity.", + req_version.version, IP_SET_PROTOCOL_VERSION); +} + +static void set_command(unsigned *cmd, const int newcmd) +{ + if (*cmd != CMD_NONE) + exit_error(PARAMETER_PROBLEM, "Can't use -%c with -%c\n", + cmd2char(*cmd), cmd2char(newcmd)); + *cmd = newcmd; +} + +static void add_option(unsigned int *options, unsigned int option) +{ + if (*options & option) + exit_error(PARAMETER_PROBLEM, + "multiple -%c flags not allowed", + opt2char(option)); + *options |= option; +} + +void *ipset_malloc(size_t size) +{ + void *p; + + if (size == 0) + return NULL; + + if ((p = malloc(size)) == NULL) { + perror("ipset: not enough memory"); + exit(1); + } + return p; +} + +char *ipset_strdup(const char *s) +{ + char *p; + + if ((p = strdup(s)) == NULL) { + perror("ipset: not enough memory"); + exit(1); + } + return p; +} + +void ipset_free(void **data) +{ + if (*data == NULL) + return; + + free(*data); + *data = NULL; +} + +static struct option *merge_options(struct option *oldopts, + const struct option *newopts, + unsigned int *option_offset) +{ + unsigned int num_old, num_new, i; + struct option *merge; + + for (num_old = 0; oldopts[num_old].name; num_old++); + for (num_new = 0; newopts[num_new].name; num_new++); + + global_option_offset += OPTION_OFFSET; + *option_offset = global_option_offset; + + merge = ipset_malloc(sizeof(struct option) * (num_new + num_old + 1)); + memcpy(merge, oldopts, num_old * sizeof(struct option)); + for (i = 0; i < num_new; i++) { + merge[num_old + i] = newopts[i]; + merge[num_old + i].val += *option_offset; + } + memset(merge + num_old + num_new, 0, sizeof(struct option)); + + return merge; +} + +static char *ip_tohost(const struct in_addr *addr) +{ + struct hostent *host; + + if ((host = gethostbyaddr((char *) addr, + sizeof(struct in_addr), + AF_INET)) != NULL) { + DP("%s", host->h_name); + return (char *) host->h_name; + } + + return (char *) NULL; +} + +static char *ip_tonetwork(const struct in_addr *addr) +{ + struct netent *net; + + if ((net = getnetbyaddr((long) ntohl(addr->s_addr), + AF_INET)) != NULL) { + DP("%s", net->n_name); + return (char *) net->n_name; + } + + return (char *) NULL; +} + +/* Return a string representation of an IP address. + * Please notice that a pointer to static char* area is returned. + */ +char *ip_tostring(ip_set_ip_t ip, unsigned options) +{ + struct in_addr addr; + addr.s_addr = htonl(ip); + + if (!(options & OPT_NUMERIC)) { + char *name; + if ((name = ip_tohost(&addr)) != NULL || + (name = ip_tonetwork(&addr)) != NULL) + return name; + } + + return inet_ntoa(addr); +} + +char *binding_ip_tostring(struct set *set, ip_set_ip_t ip, unsigned options) +{ + return ip_tostring(ip, options); +} +char *ip_tostring_numeric(ip_set_ip_t ip) +{ + return ip_tostring(ip, OPT_NUMERIC); +} + +/* Fills the 'ip' with the parsed ip or host in host byte order */ +void parse_ip(const char *str, ip_set_ip_t * ip) +{ + struct hostent *host; + struct in_addr addr; + + DP("%s", str); + + if (inet_aton(str, &addr) != 0) { + *ip = ntohl(addr.s_addr); /* We want host byte order */ + return; + } + + host = gethostbyname(str); + if (host != NULL) { + if (host->h_addrtype != AF_INET || + host->h_length != sizeof(struct in_addr)) + exit_error(PARAMETER_PROBLEM, + "host/network `%s' not an internet name", + str); + if (host->h_addr_list[1] != 0) + exit_error(PARAMETER_PROBLEM, + "host/network `%s' resolves to serveral ip-addresses. " + "Please specify one.", str); + + *ip = ntohl(((struct in_addr *) host->h_addr_list[0])->s_addr); + return; + } + + exit_error(PARAMETER_PROBLEM, "host/network `%s' not found", str); +} + +/* Fills 'mask' with the parsed mask in host byte order */ +void parse_mask(const char *str, ip_set_ip_t * mask) +{ + struct in_addr addr; + unsigned int bits; + + DP("%s", str); + + if (str == NULL) { + /* no mask at all defaults to 32 bits */ + *mask = 0xFFFFFFFF; + return; + } + if (strchr(str, '.') && inet_aton(str, &addr) != 0) { + *mask = ntohl(addr.s_addr); /* We want host byte order */ + return; + } + if (sscanf(str, "%d", &bits) != 1 || bits < 0 || bits > 32) + exit_error(PARAMETER_PROBLEM, + "invalid mask `%s' specified", str); + + DP("bits: %d", bits); + + *mask = bits != 0 ? 0xFFFFFFFF << (32 - bits) : 0L; +} + +/* Combines parse_ip and parse_mask */ +void +parse_ipandmask(const char *str, ip_set_ip_t * ip, ip_set_ip_t * mask) +{ + char buf[256]; + char *p; + + strncpy(buf, str, sizeof(buf) - 1); + buf[255] = '\0'; + + if ((p = strrchr(buf, '/')) != NULL) { + *p = '\0'; + parse_mask(p + 1, mask); + } else + parse_mask(NULL, mask); + + /* if a null mask is given, the name is ignored, like in "any/0" */ + if (*mask == 0U) + *ip = 0U; + else + parse_ip(buf, ip); + + DP("%s ip: %08X (%s) mask: %08X", + str, *ip, ip_tostring_numeric(*ip), *mask); + + /* Apply the netmask */ + *ip &= *mask; + + DP("%s ip: %08X (%s) mask: %08X", + str, *ip, ip_tostring_numeric(*ip), *mask); +} + +/* Return a string representation of a port + * Please notice that a pointer to static char* area is returned + * and we assume TCP protocol. + */ +char *port_tostring(ip_set_ip_t port, unsigned options) +{ + struct servent *service; + static char name[] = "65535"; + + if (!(options & OPT_NUMERIC)) { + if ((service = getservbyport(htons(port), "tcp"))) + return service->s_name; + } + sprintf(name, "%u", port); + return name; +} + +int +string_to_number(const char *str, unsigned int min, unsigned int max, + ip_set_ip_t *port) +{ + long number; + char *end; + + /* Handle hex, octal, etc. */ + errno = 0; + number = strtol(str, &end, 0); + if (*end == '\0' && end != str) { + /* we parsed a number, let's see if we want this */ + if (errno != ERANGE && min <= number && number <= max) { + *port = number; + return 0; + } + } + return -1; +} + +static int +string_to_port(const char *str, ip_set_ip_t *port) +{ + struct servent *service; + + if ((service = getservbyname(str, "tcp")) != NULL) { + *port = ntohs((unsigned short) service->s_port); + return 0; + } + + return -1; +} + +/* Fills the 'ip' with the parsed port in host byte order */ +void parse_port(const char *str, ip_set_ip_t *port) +{ + if ((string_to_number(str, 0, 65535, port) != 0) + && (string_to_port(str, port) != 0)) + exit_error(PARAMETER_PROBLEM, + "Invalid TCP port `%s' specified", str); +} + +/* + * Settype functions + */ +static struct settype *settype_find(const char *typename) +{ + struct settype *runner = all_settypes; + + DP("%s", typename); + + while (runner != NULL) { + if (strncmp(runner->typename, typename, + IP_SET_MAXNAMELEN) == 0) + return runner; + + runner = runner->next; + } + + return NULL; /* not found */ +} + +static struct settype *settype_load(const char *typename) +{ + char path[256]; + struct settype *settype; + + /* do some search in list */ + settype = settype_find(typename); + if (settype != NULL) + return settype; /* found */ + + /* Else we have to load it */ + snprintf(path, sizeof(path), "%s/libipset_%s.so", + xtables_libdir, typename); + + if (dlopen(path, RTLD_NOW)) { + /* Found library. */ + + settype = settype_find(typename); + + if (settype != NULL) + return settype; + } + + /* Can't load the settype */ + exit_error(PARAMETER_PROBLEM, + "Couldn't load settype `%s':%s\n", + typename, dlerror()); + + return NULL; /* Never executed, but keep compilers happy */ +} + +static char *check_set_name(char *setname) +{ + if (strlen(setname) > IP_SET_MAXNAMELEN - 1) + exit_error(PARAMETER_PROBLEM, + "Setname '%s' too long, max %d characters.", + setname, IP_SET_MAXNAMELEN - 1); + + return setname; +} + +static struct settype *check_set_typename(const char *typename) +{ + if (strlen(typename) > IP_SET_MAXNAMELEN - 1) + exit_error(PARAMETER_PROBLEM, + "Typename '%s' too long, max %d characters.", + typename, IP_SET_MAXNAMELEN - 1); + + return settype_load(typename); +} + +#define MAX(a,b) ((a) > (b) ? (a) : (b)) + +/* Register a new set type */ +void settype_register(struct settype *settype) +{ + struct settype *chk; + size_t size; + + DP("%s", settype->typename); + + /* Check if this typename already exists */ + chk = settype_find(settype->typename); + + if (chk != NULL) + exit_error(OTHER_PROBLEM, + "Set type '%s' already registered!\n", + settype->typename); + + /* Check version */ + if (settype->protocol_version != IP_SET_PROTOCOL_VERSION) + exit_error(OTHER_PROBLEM, + "Set type %s is of wrong protocol version %u!" + " I'm of version %u.\n", settype->typename, + settype->protocol_version, + IP_SET_PROTOCOL_VERSION); + + /* Initialize internal data */ + settype->header = ipset_malloc(settype->header_size); + size = MAX(settype->create_size, settype->adt_size); + settype->data = ipset_malloc(size); + + /* Insert first */ + settype->next = all_settypes; + all_settypes = settype; + + DP("%s registered", settype->typename); +} + +/* Find set functions */ +static struct set *set_find_byid(ip_set_id_t id) +{ + struct set *set = NULL; + ip_set_id_t i; + + for (i = 0; i < max_sets; i++) + if (set_list[i] && set_list[i]->id == id) { + set = set_list[i]; + break; + } + + if (set == NULL) + exit_error(PARAMETER_PROBLEM, + "Set identified by id %u is not found", id); + return set; +} + +static struct set *set_find_byname(const char *name) +{ + struct set *set = NULL; + ip_set_id_t i; + + for (i = 0; i < max_sets; i++) + if (set_list[i] + && strncmp(set_list[i]->name, name, + IP_SET_MAXNAMELEN) == 0) { + set = set_list[i]; + break; + } + if (set == NULL) + exit_error(PARAMETER_PROBLEM, + "Set %s is not found", name); + return set; +} + +static ip_set_id_t set_find_free_index(const char *name) +{ + ip_set_id_t i, idx = IP_SET_INVALID_ID; + + for (i = 0; i < max_sets; i++) { + if (idx == IP_SET_INVALID_ID + && set_list[i] == NULL) + idx = i; + if (set_list[i] != NULL + && strncmp(set_list[i]->name, name, + IP_SET_MAXNAMELEN) == 0) + exit_error(PARAMETER_PROBLEM, + "Set %s is already defined, cannot be restored", + name); + } + + if (idx == IP_SET_INVALID_ID) + exit_error(PARAMETER_PROBLEM, + "Set %s cannot be restored, " + "max number of set %u reached", + name, max_sets); + + return idx; +} + +/* + * Send create set order to kernel + */ +static void set_create(const char *name, struct settype *settype) +{ + struct ip_set_req_create req_create; + size_t size; + void *data; + + DP("%s %s", name, settype->typename); + + req_create.op = IP_SET_OP_CREATE; + req_create.version = IP_SET_PROTOCOL_VERSION; + strcpy(req_create.name, name); + strcpy(req_create.typename, settype->typename); + + /* Final checks */ + settype->create_final(settype->data, settype->flags); + + /* Alloc memory for the data to send */ + size = sizeof(struct ip_set_req_create) + settype->create_size; + data = ipset_malloc(size); + + /* Add up ip_set_req_create and the settype data */ + memcpy(data, &req_create, sizeof(struct ip_set_req_create)); + memcpy(data + sizeof(struct ip_set_req_create), + settype->data, settype->create_size); + + kernel_sendto(CMD_CREATE, data, size); + free(data); +} + +static void set_restore_create(const char *name, struct settype *settype) +{ + struct set *set; + + DP("%s %s %u %u %u %u", name, settype->typename, + restore_offset, sizeof(struct ip_set_restore), + settype->create_size, restore_size); + + /* Sanity checking */ + if (restore_offset + + sizeof(struct ip_set_restore) + + settype->create_size > restore_size) + exit_error(PARAMETER_PROBLEM, + "Giving up, restore file is screwed up!"); + + /* Final checks */ + settype->create_final(settype->data, settype->flags); + + /* Fill out restore_data */ + restore_set = (struct ip_set_restore *) + (restore_data + restore_offset); + strcpy(restore_set->name, name); + strcpy(restore_set->typename, settype->typename); + restore_set->index = set_find_free_index(name); + restore_set->header_size = settype->create_size; + restore_set->members_size = 0; + + DP("name %s, restore index %u", restore_set->name, restore_set->index); + /* Add settype data */ + + memcpy(restore_data + restore_offset + sizeof(struct ip_set_restore), + settype->data, settype->create_size); + + restore_offset += sizeof(struct ip_set_restore) + + settype->create_size; + + /* Add set to set_list */ + set = ipset_malloc(sizeof(struct set)); + strcpy(set->name, name); + set->settype = settype; + set->index = restore_set->index; + set_list[restore_set->index] = set; +} + +/* + * Send destroy/flush order to kernel for one or all sets + */ +static void set_destroy(const char *name, unsigned op, unsigned cmd) +{ + struct ip_set_req_std req; + + DP("%s %s", cmd == CMD_DESTROY ? "destroy" : "flush", name); + + req.op = op; + req.version = IP_SET_PROTOCOL_VERSION; + strcpy(req.name, name); + + kernel_sendto(cmd, &req, sizeof(struct ip_set_req_std)); +} + +/* + * Send rename/swap order to kernel + */ +static void set_rename(const char *name, const char *newname, + unsigned op, unsigned cmd) +{ + struct ip_set_req_create req; + + DP("%s %s %s", cmd == CMD_RENAME ? "rename" : "swap", + name, newname); + + req.op = op; + req.version = IP_SET_PROTOCOL_VERSION; + strcpy(req.name, name); + strcpy(req.typename, newname); + + kernel_sendto(cmd, &req, + sizeof(struct ip_set_req_create)); +} + +/* + * Send MAX_SETS, LIST_SIZE and/or SAVE_SIZE orders to kernel + */ +static size_t load_set_list(const char name[IP_SET_MAXNAMELEN], + ip_set_id_t *idx, + unsigned op, unsigned cmd) +{ + void *data = NULL; + struct ip_set_req_max_sets req_max_sets; + struct ip_set_name_list *name_list; + struct set *set; + ip_set_id_t i; + socklen_t size, req_size; + int repeated = 0, res = 0; + + DP("%s %s", cmd == CMD_MAX_SETS ? "MAX_SETS" + : cmd == CMD_LIST_SIZE ? "LIST_SIZE" + : "SAVE_SIZE", + name); + +tryagain: + if (set_list) { + for (i = 0; i < max_sets; i++) + if (set_list[i]) + free(set_list[i]); + free(set_list); + set_list = NULL; + } + /* Get max_sets */ + req_max_sets.op = IP_SET_OP_MAX_SETS; + req_max_sets.version = IP_SET_PROTOCOL_VERSION; + strcpy(req_max_sets.set.name, name); + size = sizeof(req_max_sets); + kernel_getfrom(CMD_MAX_SETS, &req_max_sets, &size); + + DP("got MAX_SETS: sets %d, max_sets %d", + req_max_sets.sets, req_max_sets.max_sets); + + max_sets = req_max_sets.max_sets; + set_list = ipset_malloc(max_sets * sizeof(struct set *)); + memset(set_list, 0, max_sets * sizeof(struct set *)); + *idx = req_max_sets.set.index; + + if (req_max_sets.sets == 0) + /* No sets in kernel */ + return 0; + + /* Get setnames */ + size = req_size = sizeof(struct ip_set_req_setnames) + + req_max_sets.sets * sizeof(struct ip_set_name_list); + data = ipset_malloc(size); + ((struct ip_set_req_setnames *) data)->op = op; + ((struct ip_set_req_setnames *) data)->index = *idx; + + res = kernel_getfrom_handleerrno(cmd, data, &size); + + if (res != 0 || size != req_size) { + free(data); + if (repeated++ < LIST_TRIES) + goto tryagain; + exit_error(OTHER_PROBLEM, + "Tried to get sets from kernel %d times" + " and failed. Please try again when the load on" + " the sets has gone down.", LIST_TRIES); + } + + /* Load in setnames */ + size = sizeof(struct ip_set_req_setnames); + while (size + sizeof(struct ip_set_name_list) <= req_size) { + name_list = (struct ip_set_name_list *) + (data + size); + set = ipset_malloc(sizeof(struct set)); + strcpy(set->name, name_list->name); + set->index = name_list->index; + set->id = name_list->id; + set->settype = settype_load(name_list->typename); + set_list[name_list->index] = set; + DP("loaded %s, type %s, index %u", + set->name, set->settype->typename, set->index); + size += sizeof(struct ip_set_name_list); + } + /* Size to get set members, bindings */ + size = ((struct ip_set_req_setnames *)data)->size; + free(data); + + return size; +} + +/* + * Save operation + */ +static size_t save_bindings(void *data, size_t offset, size_t len) +{ + struct ip_set_hash_save *hash = + (struct ip_set_hash_save *) (data + offset); + struct set *set; + + DP("offset %u, len %u", offset, len); + if (offset + sizeof(struct ip_set_hash_save) > len) + exit_error(OTHER_PROBLEM, + "Save operation failed, try again later."); + + set = set_find_byid(hash->id); + if (!(set && set_list[hash->binding])) + exit_error(OTHER_PROBLEM, + "Save binding failed, try again later."); + printf("-B %s %s -b %s\n", + set->name, + set->settype->bindip_tostring(set, hash->ip, OPT_NUMERIC), + set_list[hash->binding]->name); + + return sizeof(struct ip_set_hash_save); +} + +static size_t save_set(void *data, int *bindings, + size_t offset, size_t len) +{ + struct ip_set_save *set_save = + (struct ip_set_save *) (data + offset); + struct set *set; + struct settype *settype; + size_t used; + + DP("offset %u (%u/%u/%u), len %u", offset, + sizeof(struct ip_set_save), + set_save->header_size, set_save->members_size, + len); + if (offset + sizeof(struct ip_set_save) > len + || offset + sizeof(struct ip_set_save) + + set_save->header_size + set_save->members_size > len) + exit_error(OTHER_PROBLEM, + "Save operation failed, try again later."); + + DP("index: %u", set_save->index); + if (set_save->index == IP_SET_INVALID_ID) { + /* Marker */ + *bindings = 1; + return sizeof(struct ip_set_save); + } + set = set_list[set_save->index]; + if (!set) + exit_error(OTHER_PROBLEM, + "Save set failed, try again later."); + settype = set->settype; + + /* Init set header */ + used = sizeof(struct ip_set_save); + settype->initheader(set, data + offset + used); + + /* Print create set */ + settype->saveheader(set, OPT_NUMERIC); + + /* Print add IPs */ + used += set_save->header_size; + settype->saveips(set, data + offset + used, + set_save->members_size, OPT_NUMERIC); + + return (used + set_save->members_size); +} + +static size_t save_default_bindings(void *data, int *bindings) +{ + struct ip_set_save *set_save = (struct ip_set_save *) data; + struct set *set; + + if (set_save->index == IP_SET_INVALID_ID) { + /* Marker */ + *bindings = 1; + return sizeof(struct ip_set_save); + } + + set = set_list[set_save->index]; + DP("%s, binding %u", set->name, set_save->binding); + if (set_save->binding != IP_SET_INVALID_ID) { + if (!set_list[set_save->binding]) + exit_error(OTHER_PROBLEM, + "Save set failed, try again later."); + + printf("-B %s %s -b %s\n", + set->name, IPSET_TOKEN_DEFAULT, + set_list[set_save->binding]->name); + } + return (sizeof(struct ip_set_save) + + set_save->header_size + + set_save->members_size); +} + +static int try_save_sets(const char name[IP_SET_MAXNAMELEN]) +{ + void *data = NULL; + socklen_t size, req_size = 0; + ip_set_id_t idx; + int res = 0, bindings = 0; + time_t now = time(NULL); + + /* Load set_list from kernel */ + size = load_set_list(name, &idx, + IP_SET_OP_SAVE_SIZE, CMD_SAVE); + + if (size) { + /* Get sets, bindings and print them */ + /* Take into account marker */ + req_size = (size += sizeof(struct ip_set_save)); + data = ipset_malloc(size); + ((struct ip_set_req_list *) data)->op = IP_SET_OP_SAVE; + ((struct ip_set_req_list *) data)->index = idx; + res = kernel_getfrom_handleerrno(CMD_SAVE, data, &size); + + if (res != 0 || size != req_size) { + free(data); + return -EAGAIN; + } + } + + printf("# Generated by ipset %s on %s", IPSET_VERSION, ctime(&now)); + size = 0; + while (size < req_size) { + DP("size: %u, req_size: %u", size, req_size); + if (bindings) + size += save_bindings(data, size, req_size); + else + size += save_set(data, &bindings, size, req_size); + } + /* Re-read data to save default bindings */ + bindings = 0; + size = 0; + while (size < req_size && bindings == 0) + size += save_default_bindings(data + size, &bindings); + + printf("COMMIT\n"); + now = time(NULL); + printf("# Completed on %s", ctime(&now)); + ipset_free(&data); + return res; +} + +/* + * Performs a save to stdout + */ +static void set_save(const char name[IP_SET_MAXNAMELEN]) +{ + int i; + + DP("%s", name); + for (i = 0; i < LIST_TRIES; i++) + if (try_save_sets(name) == 0) + return; + + if (errno == EAGAIN) + exit_error(OTHER_PROBLEM, + "Tried to save sets from kernel %d times" + " and failed. Please try again when the load on" + " the sets has gone down.", LIST_TRIES); + else + kernel_error(CMD_SAVE, errno); +} + +/* + * Restore operation + */ + +/* global new argv and argc */ +static char *newargv[255]; +static int newargc = 0; + +/* Build faked argv from parsed line */ +static void build_argv(int line, char *buffer) { + char *ptr; + int i; + + /* Reset */ + for (i = 1; i < newargc; i++) + free(newargv[i]); + newargc = 1; + + ptr = strtok(buffer, " \t\n"); + newargv[newargc++] = ipset_strdup(ptr); + while ((ptr = strtok(NULL, " \t\n")) != NULL) { + if ((newargc + 1) < sizeof(newargv)/sizeof(char *)) + newargv[newargc++] = ipset_strdup(ptr); + else + exit_error(PARAMETER_PROBLEM, + "Line %d is too long to restore\n", line); + } +} + +static FILE *create_tempfile(void) +{ + char buffer[1024]; + char *tmpdir = NULL; + char *filename; + int fd; + FILE *file; + + if (!(tmpdir = getenv("TMPDIR")) && !(tmpdir = getenv("TMP"))) + tmpdir = "/tmp"; + filename = ipset_malloc(strlen(tmpdir) + strlen(TEMPFILE_PATTERN) + 1); + strcpy(filename, tmpdir); + strcat(filename, TEMPFILE_PATTERN); + + (void) umask(077); /* Create with restrictive permissions */ + fd = mkstemp(filename); + if (fd == -1) + exit_error(OTHER_PROBLEM, "Could not create temporary file."); + if (!(file = fdopen(fd, "r+"))) + exit_error(OTHER_PROBLEM, "Could not open temporary file."); + if (unlink(filename) == -1) + exit_error(OTHER_PROBLEM, "Could not unlink temporary file."); + free(filename); + + while (fgets(buffer, sizeof(buffer), stdin)) { + fputs(buffer, file); + } + fseek(file, 0L, SEEK_SET); + + return file; +} + +/* + * Performs a restore from a file + */ +static void set_restore(char *argv0) +{ + char buffer[1024]; + char *ptr, *name = NULL; + char cmd = ' '; + int line = 0, first_pass, i, bindings = 0; + struct settype *settype = NULL; + struct ip_set_req_setnames *header; + ip_set_id_t idx; + FILE *in; + int res; + + /* Create and store stdin in temporary file */ + in = create_tempfile(); + + /* Load existing sets from kernel */ + load_set_list(IPSET_TOKEN_ALL, &idx, + IP_SET_OP_LIST_SIZE, CMD_RESTORE); + + restore_size = sizeof(struct ip_set_req_setnames)/* header */ + + sizeof(struct ip_set_restore); /* marker */ + DP("restore_size: %u", restore_size); + /* First pass: calculate required amount of data */ + while (fgets(buffer, sizeof(buffer), in)) { + line++; + + if (buffer[0] == '\n') + continue; + else if (buffer[0] == '#') + continue; + else if (strcmp(buffer, "COMMIT\n") == 0) { + /* Enable restore mode */ + restore = 1; + break; + } + + /* -N, -A or -B */ + ptr = strtok(buffer, " \t\n"); + DP("ptr: %s", ptr); + if (ptr == NULL + || ptr[0] != '-' + || !(ptr[1] == 'N' + || ptr[1] == 'A' + || ptr[1] == 'B') + || ptr[2] != '\0') { + exit_error(PARAMETER_PROBLEM, + "Line %u does not start as a valid restore command\n", + line); + } + cmd = ptr[1]; + /* setname */ + ptr = strtok(NULL, " \t\n"); + DP("setname: %s", ptr); + if (ptr == NULL) + exit_error(PARAMETER_PROBLEM, + "Missing set name in line %u\n", + line); + DP("cmd %c", cmd); + switch (cmd) { + case 'N': { + name = check_set_name(ptr); + /* settype */ + ptr = strtok(NULL, " \t\n"); + if (ptr == NULL) + exit_error(PARAMETER_PROBLEM, + "Missing settype in line %u\n", + line); + if (bindings) + exit_error(PARAMETER_PROBLEM, + "Invalid line %u: create must precede bindings\n", + line); + settype = check_set_typename(ptr); + restore_size += sizeof(struct ip_set_restore) + + settype->create_size; + DP("restore_size (N): %u", restore_size); + break; + } + case 'A': { + if (name == NULL + || strncmp(name, ptr, sizeof(name)) != 0) + exit_error(PARAMETER_PROBLEM, + "Add IP to set %s in line %u without " + "preceding corresponding create set line\n", + ptr, line); + if (bindings) + exit_error(PARAMETER_PROBLEM, + "Invalid line %u: adding entries must precede bindings\n", + line); + restore_size += settype->adt_size; + DP("restore_size (A): %u", restore_size); + break; + } + case 'B': { + bindings = 1; + restore_size += sizeof(struct ip_set_hash_save); + DP("restore_size (B): %u", restore_size); + break; + } + default: { + exit_error(PARAMETER_PROBLEM, + "Unrecognized restore command in line %u\n", + line); + } + } /* end of switch */ + } + /* Sanity checking */ + if (!restore) + exit_error(PARAMETER_PROBLEM, + "Missing COMMIT line\n"); + DP("restore_size: %u", restore_size); + restore_data = ipset_malloc(restore_size); + header = (struct ip_set_req_setnames *) restore_data; + header->op = IP_SET_OP_RESTORE; + header->size = restore_size; + restore_offset = sizeof(struct ip_set_req_setnames); + + /* Rewind to scan the file again */ + fseek(in, 0L, SEEK_SET); + first_pass = line; + line = 0; + + /* Initialize newargv/newargc */ + newargv[newargc++] = ipset_strdup(argv0); + + /* Second pass: build up restore request */ + while (fgets(buffer, sizeof(buffer), in)) { + line++; + + if (buffer[0] == '\n') + continue; + else if (buffer[0] == '#') + continue; + else if (strcmp(buffer, "COMMIT\n") == 0) + goto do_restore; + DP("restoring: %s", buffer); + /* Build faked argv, argc */ + build_argv(line, buffer); + for (i = 0; i < newargc; i++) + DP("argv[%u]: %s", i, newargv[i]); + + /* Parse line */ + parse_commandline(newargc, newargv); + } + exit_error(PARAMETER_PROBLEM, + "Broken restore file\n"); + do_restore: + if (bindings == 0 + && restore_size == + (restore_offset + sizeof(struct ip_set_restore))) { + /* No bindings */ + struct ip_set_restore *marker = + (struct ip_set_restore *) (restore_data + restore_offset); + + DP("restore marker"); + marker->index = IP_SET_INVALID_ID; + marker->header_size = marker->members_size = 0; + restore_offset += sizeof(struct ip_set_restore); + } + if (restore_size != restore_offset) + exit_error(PARAMETER_PROBLEM, + "Giving up, restore file is screwed up!"); + res = kernel_getfrom_handleerrno(CMD_RESTORE, restore_data, &restore_size); + + if (res != 0) { + if (restore_size != sizeof(struct ip_set_req_setnames)) + exit_error(PARAMETER_PROBLEM, + "Communication with kernel failed (%u %u)!", + restore_size, sizeof(struct ip_set_req_setnames)); + /* Check errors */ + header = (struct ip_set_req_setnames *) restore_data; + if (header->size != 0) + exit_error(PARAMETER_PROBLEM, + "Committing restoring failed at line %u!", + header->size); + } +} + +/* + * Send ADT_GET order to kernel for a set + */ +static struct set *set_adt_get(const char *name) +{ + struct ip_set_req_adt_get req_adt_get; + struct set *set; + socklen_t size; + + DP("%s", name); + + req_adt_get.op = IP_SET_OP_ADT_GET; + req_adt_get.version = IP_SET_PROTOCOL_VERSION; + strcpy(req_adt_get.set.name, name); + size = sizeof(struct ip_set_req_adt_get); + + kernel_getfrom(CMD_ADT_GET, (void *) &req_adt_get, &size); + + set = ipset_malloc(sizeof(struct set)); + strcpy(set->name, name); + set->index = req_adt_get.set.index; + set->settype = settype_load(req_adt_get.typename); + + return set; +} + +/* + * Send add/del/test order to kernel for a set + */ +static int set_adtip(struct set *set, const char *adt, + unsigned op, unsigned cmd) +{ + struct ip_set_req_adt *req_adt; + size_t size; + void *data; + int res = 0; + + DP("%s -> %s", set->name, adt); + + /* Alloc memory for the data to send */ + size = sizeof(struct ip_set_req_adt) + set->settype->adt_size ; + DP("alloc size %i", size); + data = ipset_malloc(size); + + /* Fill out the request */ + req_adt = (struct ip_set_req_adt *) data; + req_adt->op = op; + req_adt->index = set->index; + memcpy(data + sizeof(struct ip_set_req_adt), + set->settype->data, set->settype->adt_size); + + if (kernel_sendto_handleerrno(cmd, op, data, size) == -1) + switch (op) { + case IP_SET_OP_ADD_IP: + exit_error(OTHER_PROBLEM, "%s is already in set %s.", + adt, set->name); + break; + case IP_SET_OP_DEL_IP: + exit_error(OTHER_PROBLEM, "%s is not in set %s.", + adt, set->name); + break; + case IP_SET_OP_TEST_IP: + ipset_printf("%s is in set %s.", adt, set->name); + res = 0; + break; + default: + break; + } + else + switch (op) { + case IP_SET_OP_TEST_IP: + ipset_printf("%s is NOT in set %s.", adt, set->name); + res = 1; + break; + default: + break; + } + free(data); + + return res; +} + +static void set_restore_add(struct set *set, const char *adt) +{ + DP("%s %s", set->name, adt); + /* Sanity checking */ + if (restore_offset + set->settype->adt_size > restore_size) + exit_error(PARAMETER_PROBLEM, + "Giving up, restore file is screwed up!"); + + memcpy(restore_data + restore_offset, + set->settype->data, set->settype->adt_size); + restore_set->members_size += set->settype->adt_size; + restore_offset += set->settype->adt_size; +} + +/* + * Send bind/unbind/test binding order to kernel for a set + */ +static int set_bind(struct set *set, const char *adt, + const char *binding, + unsigned op, unsigned cmd) +{ + struct ip_set_req_bind *req_bind; + size_t size; + void *data; + int res = 0; + + /* set may be null: '-U :all: :all:|:default:' */ + DP("(%s, %s) -> %s", set ? set->name : IPSET_TOKEN_ALL, adt, binding); + + /* Ugly */ + if (set && strcmp(set->settype->typename, "iptreemap") == 0) + exit_error(PARAMETER_PROBLEM, + "iptreemap type of sets cannot be used at binding operations\n"); + /* Alloc memory for the data to send */ + size = sizeof(struct ip_set_req_bind); + if (op != IP_SET_OP_UNBIND_SET && adt[0] == ':') + /* Set default binding */ + size += IP_SET_MAXNAMELEN; + else if (!(op == IP_SET_OP_UNBIND_SET && set == NULL)) + size += set->settype->adt_size; + DP("alloc size %i", size); + data = ipset_malloc(size); + + /* Fill out the request */ + req_bind = (struct ip_set_req_bind *) data; + req_bind->op = op; + req_bind->index = set ? set->index : IP_SET_INVALID_ID; + if (adt[0] == ':') { + /* ':default:' and ':all:' */ + strncpy(req_bind->binding, adt, IP_SET_MAXNAMELEN); + if (op != IP_SET_OP_UNBIND_SET && adt[0] == ':') + strncpy(data + sizeof(struct ip_set_req_bind), + binding, IP_SET_MAXNAMELEN); + } else { + strncpy(req_bind->binding, binding, IP_SET_MAXNAMELEN); + memcpy(data + sizeof(struct ip_set_req_bind), + set->settype->data, set->settype->adt_size); + } + + if (op == IP_SET_OP_TEST_BIND_SET) { + if (kernel_sendto_handleerrno(cmd, op, data, size) == -1) { + ipset_printf("%s in set %s is bound to %s.", + adt, set->name, binding); + res = 0; + } else { + ipset_printf("%s in set %s is NOT bound to %s.", + adt, set->name, binding); + res = 1; + } + } else + kernel_sendto(cmd, data, size); + free(data); + + return res; +} + +static void set_restore_bind(struct set *set, + const char *adt, + const char *binding) +{ + struct ip_set_hash_save *hash_restore; + + if (restore == 1) { + /* Marker */ + struct ip_set_restore *marker = + (struct ip_set_restore *) (restore_data + restore_offset); + + DP("restore marker"); + if (restore_offset + sizeof(struct ip_set_restore) + > restore_size) + exit_error(PARAMETER_PROBLEM, + "Giving up, restore file is screwed up!"); + marker->index = IP_SET_INVALID_ID; + marker->header_size = marker->members_size = 0; + restore_offset += sizeof(struct ip_set_restore); + restore = 2; + } + /* Sanity checking */ + if (restore_offset + sizeof(struct ip_set_hash_save) > restore_size) + exit_error(PARAMETER_PROBLEM, + "Giving up, restore file is screwed up!"); + + hash_restore = (struct ip_set_hash_save *) (restore_data + restore_offset); + DP("%s -> %s", adt, binding); + if (strcmp(adt, IPSET_TOKEN_DEFAULT) == 0) + hash_restore->ip = 0; + else + set->settype->bindip_parse(adt, &hash_restore->ip); + hash_restore->id = set->index; + hash_restore->binding = (set_find_byname(binding))->index; + DP("id %u, ip %u, binding %u", + hash_restore->id, hash_restore->ip, hash_restore->binding); + restore_offset += sizeof(struct ip_set_hash_save); +} + +/* + * Print operation + */ + +static void print_bindings(struct set *set, + void *data, size_t size, unsigned options, + char * (*printip)(struct set *set, + ip_set_ip_t ip, unsigned options)) +{ + size_t offset = 0; + struct ip_set_hash_list *hash; + + while (offset < size) { + hash = (struct ip_set_hash_list *) (data + offset); + printf("%s -> %s\n", + printip(set, hash->ip, options), + set_list[hash->binding]->name); + offset += sizeof(struct ip_set_hash_list); + } +} + +/* Help function to set_list() */ +static size_t print_set(void *data, unsigned options) +{ + struct ip_set_list *setlist = (struct ip_set_list *) data; + struct set *set = set_list[setlist->index]; + struct settype *settype = set->settype; + size_t offset; + + /* Pretty print the set */ + printf("Name: %s\n", set->name); + printf("Type: %s\n", settype->typename); + printf("References: %d\n", setlist->ref); + printf("Default binding: %s\n", + setlist->binding == IP_SET_INVALID_ID ? "" + : set_list[setlist->binding]->name); + + /* Init header */ + offset = sizeof(struct ip_set_list); + settype->initheader(set, data + offset); + + /* Pretty print the type header */ + printf("Header:"); + settype->printheader(set, options); + + /* Pretty print all IPs */ + printf("Members:\n"); + offset += setlist->header_size; + if (options & OPT_SORTED) + settype->printips_sorted(set, data + offset, + setlist->members_size, options); + else + settype->printips(set, data + offset, + setlist->members_size, options); + + /* Print bindings */ + printf("Bindings:\n"); + offset += setlist->members_size; + print_bindings(set, + data + offset, setlist->bindings_size, options, + settype->bindip_tostring); + + printf("\n"); /* One newline between sets */ + + return (offset + setlist->bindings_size); +} + +static int try_list_sets(const char name[IP_SET_MAXNAMELEN], + unsigned options) +{ + void *data = NULL; + ip_set_id_t idx; + socklen_t size, req_size; + int res = 0; + + DP("%s", name); + /* Load set_list from kernel */ + size = req_size = load_set_list(name, &idx, + IP_SET_OP_LIST_SIZE, CMD_LIST); + + if (size) { + /* Get sets and print them */ + data = ipset_malloc(size); + ((struct ip_set_req_list *) data)->op = IP_SET_OP_LIST; + ((struct ip_set_req_list *) data)->index = idx; + res = kernel_getfrom_handleerrno(CMD_LIST, data, &size); + DP("get_lists getsockopt() res=%d errno=%d", res, errno); + + if (res != 0 || size != req_size) { + free(data); + return -EAGAIN; + } + size = 0; + } + while (size != req_size) + size += print_set(data + size, options); + + ipset_free(&data); + return res; +} + +/* Print a set or all sets + * All sets: name = NULL + */ +static void list_sets(const char name[IP_SET_MAXNAMELEN], unsigned options) +{ + int i; + + DP("%s", name); + for (i = 0; i < LIST_TRIES; i++) + if (try_list_sets(name, options) == 0) + return; + + if (errno == EAGAIN) + exit_error(OTHER_PROBLEM, + "Tried to list sets from kernel %d times" + " and failed. Please try again when the load on" + " the sets has gone down.", LIST_TRIES); + else + kernel_error(CMD_LIST, errno); +} + +/* Prints help + * If settype is non null help for that type is printed as well + */ +static void set_help(const struct settype *settype) +{ +#ifdef IPSET_DEBUG + char debughelp[] = + " --debug -z Enable debugging\n\n"; +#else + char debughelp[] = "\n"; +#endif + + printf("%s v%s\n\n" + "Usage: %s -N new-set settype [options]\n" + " %s -[XFLSH] [set] [options]\n" + " %s -[EW] from-set to-set\n" + " %s -[ADTU] set IP\n" + " %s -B set IP option\n" + " %s -R\n" + " %s -h (print this help information)\n\n", + program_name, program_version, + program_name, program_name, program_name, + program_name, program_name, program_name, + program_name); + + printf("Commands:\n" + "Either long or short options are allowed.\n" + " --create -N setname settype \n" + " Create a new set\n" + " --destroy -X [setname]\n" + " Destroy a set or all sets\n" + " --flush -F [setname]\n" + " Flush a set or all sets\n" + " --rename -E from-set to-set\n" + " Rename from-set to to-set\n" + " --swap -W from-set to-set\n" + " Swap the content of two existing sets\n" + " --list -L [setname] [options]\n" + " List the IPs in a set or all sets\n" + " --save -S [setname]\n" + " Save the set or all sets to stdout\n" + " --restore -R [option]\n" + " Restores a saved state\n" + " --add -A setname IP\n" + " Add an IP to a set\n" + " --del -D setname IP\n" + " Deletes an IP from a set\n" + " --test -T setname IP \n" + " Tests if an IP exists in a set.\n" + " --bind -B setname IP|:default: -b bind-setname\n" + " Bind the IP in setname to bind-setname.\n" + " --unbind -U setname IP|:all:|:default:\n" + " Delete binding belonging to IP,\n" + " all bindings or default binding of setname.\n" + " --unbind -U :all: :all:|:default:\n" + " Delete all bindings or all default bindings.\n" + " --help -H [settype]\n" + " Prints this help, and settype specific help\n" + " --version -V\n" + " Prints version information\n\n" + "Options:\n" + " --sorted -s Numeric sort of the IPs in -L\n" + " --numeric -n Numeric output of addresses in a -L\n" + " --quiet -q Suppress any output to stdout and stderr.\n" + " --binding -b Specifies the binding for -B\n"); + printf(debughelp); + + if (settype != NULL) { + printf("Type '%s' specific:\n", settype->typename); + settype->usage(); + } +} + +static int find_cmd(const char option) +{ + int i; + + for (i = 1; i <= NUMBER_OF_CMD; i++) + if (cmdflags[i] == option) + return i; + + return CMD_NONE; +} + +static int parse_adt_cmdline(unsigned command, + const char *name, + char *adt, + struct set **set, + struct settype **settype) +{ + int res = 0; + + /* -U :all: :all:|:default: */ + if (command == CMD_UNBIND) { + if (strcmp(name, IPSET_TOKEN_ALL) == 0) { + if (strcmp(adt, IPSET_TOKEN_DEFAULT) == 0 + || strcmp(adt, IPSET_TOKEN_ALL) == 0) { + *set = NULL; + *settype = NULL; + return 1; + } else + exit_error(PARAMETER_PROBLEM, + "-U %s requires %s or %s as binding name", + IPSET_TOKEN_ALL, + IPSET_TOKEN_DEFAULT, + IPSET_TOKEN_ALL); + } + } + *set = restore ? set_find_byname(name) + : set_adt_get(name); + + /* Reset space for adt data */ + *settype = (*set)->settype; + memset((*settype)->data, 0, (*settype)->adt_size); + + if ((command == CMD_TEST + || command == CMD_BIND + || command == CMD_UNBIND) + && (strcmp(adt, IPSET_TOKEN_DEFAULT) == 0 + || strcmp(adt, IPSET_TOKEN_ALL) == 0)) + res = 1; + else + res = (*settype)->adt_parser( + command, + adt, + (*settype)->data); + + return res; +} + +/* Main worker function */ +int parse_commandline(int argc, char *argv[]) +{ + int res = 0; + unsigned command = CMD_NONE; + unsigned options = 0; + int c; + + char *name = NULL; /* All except -H, -R */ + char *newname = NULL; /* -E, -W */ + char *adt = NULL; /* -A, -D, -T, -B, -U */ + char *binding = NULL; /* -B */ + struct set *set = NULL; /* -A, -D, -T, -B, -U */ + struct settype *settype = NULL; /* -N, -H */ + char all_sets[] = IPSET_TOKEN_ALL; + + struct option *opts = opts_long; + + /* Suppress error messages: we may add new options if we + demand-load a protocol. */ + opterr = 0; + /* Reset optind to 0 for restore */ + optind = 0; + + while ((c = getopt_long(argc, argv, opts_short, opts, NULL)) != -1) { + + DP("commandline parsed: opt %c (%s)", c, argv[optind]); + + switch (c) { + /* + * Command selection + */ + case 'h': + case 'H':{ /* Help: -H [typename [options]] */ + check_protocolversion(); + set_command(&command, CMD_HELP); + + if (optarg) + settype = check_set_typename(optarg); + else if (optind < argc + && argv[optind][0] != '-') + settype = check_set_typename(argv[optind++]); + + break; + } + + case 'V':{ /* Version */ + printf("%s v%s Protocol version %u.\n", + program_name, program_version, + IP_SET_PROTOCOL_VERSION); + check_protocolversion(); + exit(0); + } + + case 'N':{ /* Create: -N name typename options */ + set_command(&command, CMD_CREATE); + + name = check_set_name(optarg); + + /* Protect reserved names (binding) */ + if (name[0] == ':') + exit_error(PARAMETER_PROBLEM, + "setname might not start with colon", + cmd2char(CMD_CREATE)); + + if (optind < argc + && argv[optind][0] != '-') + settype = check_set_typename(argv[optind++]); + else + exit_error(PARAMETER_PROBLEM, + "-%c requires setname and settype", + cmd2char(CMD_CREATE)); + + DP("merge options"); + /* Merge the create options */ + opts = merge_options(opts, + settype->create_opts, + &settype->option_offset); + + /* Reset space for create data */ + memset(settype->data, 0, settype->create_size); + + /* Zero the flags */ + settype->flags = 0; + + DP("call create_init"); + /* Call the settype create_init */ + settype->create_init(settype->data); + + break; + } + + case 'X': /* Destroy */ + case 'F': /* Flush */ + case 'L': /* List */ + case 'S':{ /* Save */ + set_command(&command, find_cmd(c)); + + if (optarg) + name = check_set_name(optarg); + else if (optind < argc + && argv[optind][0] != '-') + name = check_set_name(argv[optind++]); + else + name = all_sets; + + break; + } + + case 'R':{ /* Restore */ + set_command(&command, find_cmd(c)); + + break; + } + + case 'E': /* Rename */ + case 'W':{ /* Swap */ + set_command(&command, find_cmd(c)); + name = check_set_name(optarg); + + if (optind < argc + && argv[optind][0] != '-') + newname = check_set_name(argv[optind++]); + else + exit_error(PARAMETER_PROBLEM, + "-%c requires a setname " + "and the new name for that set", + cmd2char(CMD_RENAME)); + + break; + } + + case 'A': /* Add IP */ + case 'D': /* Del IP */ + case 'T': /* Test IP */ + case 'B': /* Bind IP */ + case 'U':{ /* Unbind IP */ + set_command(&command, find_cmd(c)); + + name = check_set_name(optarg); + + /* IP */ + if (optind < argc + && argv[optind][0] != '-') + adt = argv[optind++]; + else + exit_error(PARAMETER_PROBLEM, + "-%c requires setname and IP", + c); + + res = parse_adt_cmdline(command, name, adt, + &set, &settype); + + if (!res) + exit_error(PARAMETER_PROBLEM, + "Unknown arg `%s'", + argv[optind - 1]); + + res = 0; + break; + } + + /* options */ + + case 'n': + add_option(&options, OPT_NUMERIC); + break; + + case 's': + add_option(&options, OPT_SORTED); + break; + + case 'q': + add_option(&options, OPT_QUIET); + option_quiet = 1; + break; + +#ifdef IPSET_DEBUG + case 'z': /* debug */ + add_option(&options, OPT_DEBUG); + option_debug = 1; + break; +#endif + + case 'b': + add_option(&options, OPT_BINDING); + binding = check_set_name(optarg); + break; + + case 1: /* non option */ + printf("Bad argument `%s'\n", optarg); + exit_tryhelp(2); + break; /*always good */ + + default:{ + DP("default"); + + switch (command) { + case CMD_CREATE: + res = settype->create_parse( + c - settype->option_offset, + argv, + settype->data, + &settype->flags); + break; + + default: + res = 0; /* failed */ + } /* switch (command) */ + + + if (!res) + exit_error(PARAMETER_PROBLEM, + "Unknown arg `%s'", + argv[optind - 1]); + + res = 0; + } + + DP("next arg"); + } /* switch */ + + } /* while( getopt_long() ) */ + + + if (optind < argc) + exit_error(PARAMETER_PROBLEM, + "unknown arguments found on commandline"); + if (command == CMD_NONE) + exit_error(PARAMETER_PROBLEM, "no command specified"); + + /* Check options */ + generic_opt_check(command, options); + + DP("cmd: %c", cmd2char(command)); + + switch (command) { + case CMD_CREATE: + DP("CMD_CREATE"); + if (restore) + set_restore_create(name, settype); + else + set_create(name, settype); + break; + + case CMD_DESTROY: + set_destroy(name, IP_SET_OP_DESTROY, CMD_DESTROY); + break; + + case CMD_FLUSH: + set_destroy(name, IP_SET_OP_FLUSH, CMD_FLUSH); + break; + + case CMD_RENAME: + set_rename(name, newname, IP_SET_OP_RENAME, CMD_RENAME); + break; + + case CMD_SWAP: + set_rename(name, newname, IP_SET_OP_SWAP, CMD_SWAP); + break; + + case CMD_LIST: + list_sets(name, options); + break; + + case CMD_SAVE: + set_save(name); + break; + + case CMD_RESTORE: + set_restore(argv[0]); + break; + + case CMD_ADD: + if (restore) + set_restore_add(set, adt); + else + set_adtip(set, adt, IP_SET_OP_ADD_IP, CMD_ADD); + break; + + case CMD_DEL: + set_adtip(set, adt, IP_SET_OP_DEL_IP, CMD_DEL); + break; + + case CMD_TEST: + if (binding) + res = set_bind(set, adt, binding, + IP_SET_OP_TEST_BIND_SET, CMD_TEST); + else + res = set_adtip(set, adt, + IP_SET_OP_TEST_IP, CMD_TEST); + break; + + case CMD_BIND: + if (restore) + set_restore_bind(set, adt, binding); + else + set_bind(set, adt, binding, + IP_SET_OP_BIND_SET, CMD_BIND); + break; + + case CMD_UNBIND: + set_bind(set, adt, "", IP_SET_OP_UNBIND_SET, CMD_UNBIND); + break; + + case CMD_HELP: + set_help(settype); + break; + + default: + /* Will never happen */ + break; /* Keep the compiler happy */ + + } /* switch( command ) */ + + return res; +} + + +int main(int argc, char *argv[]) +{ + const char *p = getenv("XTABLES_LIBDIR"); + + if (p != NULL) + xtables_libdir = p; + + return parse_commandline(argc, argv); + +} diff --git a/extensions/ipset/ipset.h b/extensions/ipset/ipset.h new file mode 100644 index 0000000..35153af --- /dev/null +++ b/extensions/ipset/ipset.h @@ -0,0 +1,191 @@ +#ifndef __IPSET_H +#define __IPSET_H + +/* Copyright 2000-2004 Joakim Axelsson (gozem@linux.nu) + * Patrick Schaaf (bof@bof.de) + * Jozsef Kadlecsik (kadlec@blackhole.kfki.hu) + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include +#include +#include + +#include "ip_set.h" + +#define PROC_SYS_MODPROBE "/proc/sys/kernel/modprobe" + +#define LIST_TRIES 5 + +#ifdef IPSET_DEBUG +extern int option_debug; +#define DP(format, args...) if (option_debug) \ + do { \ + fprintf(stderr, "%s: %s (DBG): ", __FILE__, __FUNCTION__);\ + fprintf(stderr, format "\n" , ## args); \ + } while (0) +#else +#define DP(format, args...) +#endif + +/* Commands */ +enum set_commands { + CMD_NONE, + CMD_CREATE, /* -N */ + CMD_DESTROY, /* -X */ + CMD_FLUSH, /* -F */ + CMD_RENAME, /* -E */ + CMD_SWAP, /* -W */ + CMD_LIST, /* -L */ + CMD_SAVE, /* -S */ + CMD_RESTORE, /* -R */ + CMD_ADD, /* -A */ + CMD_DEL, /* -D */ + CMD_TEST, /* -T */ + CMD_BIND, /* -B */ + CMD_UNBIND, /* -U */ + CMD_HELP, /* -H */ + CMD_VERSION, /* -V */ + NUMBER_OF_CMD = CMD_VERSION, + /* Internal commands */ + CMD_MAX_SETS, + CMD_LIST_SIZE, + CMD_SAVE_SIZE, + CMD_ADT_GET, +}; + +enum exittype { + OTHER_PROBLEM = 1, + PARAMETER_PROBLEM, + VERSION_PROBLEM +}; + +/* The view of an ipset in userspace */ +struct set { + char name[IP_SET_MAXNAMELEN]; /* Name of the set */ + ip_set_id_t id; /* Unique set id */ + ip_set_id_t index; /* Array index */ + unsigned ref; /* References in kernel */ + struct settype *settype; /* Pointer to set type functions */ +}; + +struct settype { + struct settype *next; + + char typename[IP_SET_MAXNAMELEN]; + + int protocol_version; + + /* + * Create set + */ + + /* Size of create data. Will be sent to kernel */ + size_t create_size; + + /* Initialize the create. */ + void (*create_init) (void *data); + + /* Function which parses command options; returns true if it ate an option */ + int (*create_parse) (int c, char *argv[], void *data, + unsigned *flags); + + /* Final check; exit if not ok. */ + void (*create_final) (void *data, unsigned int flags); + + /* Pointer to list of extra command-line options for create */ + const struct option *create_opts; + + /* + * Add/del/test IP + */ + + /* Size of data. Will be sent to kernel */ + size_t adt_size; + + /* Function which parses command options */ + ip_set_ip_t (*adt_parser) (unsigned cmd, const char *optarg, void *data); + + /* + * Printing + */ + + /* Size of header. */ + size_t header_size; + + /* Initialize the type-header */ + void (*initheader) (struct set *set, const void *data); + + /* Pretty print the type-header */ + void (*printheader) (struct set *set, unsigned options); + + /* Pretty print all IPs */ + void (*printips) (struct set *set, void *data, size_t len, unsigned options); + + /* Pretty print all IPs sorted */ + void (*printips_sorted) (struct set *set, void *data, size_t len, unsigned options); + + /* Print save arguments for creating the set */ + void (*saveheader) (struct set *set, unsigned options); + + /* Print save for all IPs */ + void (*saveips) (struct set *set, void *data, size_t len, unsigned options); + + /* Conver a single IP (binding) to string */ + char * (*bindip_tostring)(struct set *set, ip_set_ip_t ip, unsigned options); + + /* Parse an IP at restoring bindings. FIXME */ + void (*bindip_parse) (const char *str, ip_set_ip_t * ip); + + /* Print usage */ + void (*usage) (void); + + /* Internal data */ + void *header; + void *data; + unsigned int option_offset; + unsigned int flags; +}; + +extern void settype_register(struct settype *settype); + +/* extern void unregister_settype(set_type_t *set_type); */ + +extern void exit_error(enum exittype status, const char *msg, ...); + +extern char *binding_ip_tostring(struct set *set, + ip_set_ip_t ip, unsigned options); +extern char *ip_tostring(ip_set_ip_t ip, unsigned options); +extern char *ip_tostring_numeric(ip_set_ip_t ip); +extern void parse_ip(const char *str, ip_set_ip_t * ip); +extern void parse_mask(const char *str, ip_set_ip_t * mask); +extern void parse_ipandmask(const char *str, ip_set_ip_t * ip, + ip_set_ip_t * mask); +extern char *port_tostring(ip_set_ip_t port, unsigned options); +extern void parse_port(const char *str, ip_set_ip_t * port); +extern int string_to_number(const char *str, unsigned int min, unsigned int max, + ip_set_ip_t *port); + +extern void *ipset_malloc(size_t size); +extern char *ipset_strdup(const char *); +extern void ipset_free(void **data); + +#define BITSPERBYTE (8*sizeof(char)) +#define ID2BYTE(id) ((id)/BITSPERBYTE) +#define ID2MASK(id) (1 << ((id)%BITSPERBYTE)) +#define test_bit(id, heap) ((((char *)(heap))[ID2BYTE(id)] & ID2MASK(id)) != 0) + +#endif /* __IPSET_H */ diff --git a/extensions/ipset/ipset_iphash.c b/extensions/ipset/ipset_iphash.c new file mode 100644 index 0000000..9426180 --- /dev/null +++ b/extensions/ipset/ipset_iphash.c @@ -0,0 +1,299 @@ +/* Copyright 2004 Jozsef Kadlecsik (kadlec@blackhole.kfki.hu) + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "ip_set_iphash.h" +#include "ip_set_jhash.h" + +#include "ipset.h" + +#define BUFLEN 30; + +#define OPT_CREATE_HASHSIZE 0x01U +#define OPT_CREATE_PROBES 0x02U +#define OPT_CREATE_RESIZE 0x04U +#define OPT_CREATE_NETMASK 0x08U + +/* Initialize the create. */ +static void create_init(void *data) +{ + struct ip_set_req_iphash_create *mydata = + (struct ip_set_req_iphash_create *) data; + + DP("create INIT"); + + /* Default create parameters */ + mydata->hashsize = 1024; + mydata->probes = 8; + mydata->resize = 50; + + mydata->netmask = 0xFFFFFFFF; +} + +/* Function which parses command options; returns true if it ate an option */ +static int create_parse(int c, char *argv[], void *data, unsigned int *flags) +{ + struct ip_set_req_iphash_create *mydata = + (struct ip_set_req_iphash_create *) data; + unsigned int bits; + ip_set_ip_t value; + + DP("create_parse"); + + switch (c) { + case '1': + + if (string_to_number(optarg, 1, UINT_MAX - 1, &mydata->hashsize)) + exit_error(PARAMETER_PROBLEM, "Invalid hashsize `%s' specified", optarg); + + *flags |= OPT_CREATE_HASHSIZE; + + DP("--hashsize %u", mydata->hashsize); + + break; + + case '2': + + if (string_to_number(optarg, 1, 65535, &value)) + exit_error(PARAMETER_PROBLEM, "Invalid probes `%s' specified", optarg); + + mydata->probes = value; + *flags |= OPT_CREATE_PROBES; + + DP("--probes %u", mydata->probes); + + break; + + case '3': + + if (string_to_number(optarg, 0, 65535, &value)) + exit_error(PARAMETER_PROBLEM, "Invalid resize `%s' specified", optarg); + + mydata->resize = value; + *flags |= OPT_CREATE_RESIZE; + + DP("--resize %u", mydata->resize); + + break; + + case '4': + + if (string_to_number(optarg, 0, 32, &bits)) + exit_error(PARAMETER_PROBLEM, + "Invalid netmask `%s' specified", optarg); + + if (bits != 0) + mydata->netmask = 0xFFFFFFFF << (32 - bits); + + *flags |= OPT_CREATE_NETMASK; + + DP("--netmask %x", mydata->netmask); + + break; + + default: + return 0; + } + + return 1; +} + +/* Final check; exit if not ok. */ +static void create_final(void *data, unsigned int flags) +{ +#ifdef IPSET_DEBUG + struct ip_set_req_iphash_create *mydata = + (struct ip_set_req_iphash_create *) data; + + DP("hashsize %u probes %u resize %u", + mydata->hashsize, mydata->probes, mydata->resize); +#endif +} + +/* Create commandline options */ +static const struct option create_opts[] = { + {"hashsize", 1, 0, '1'}, + {"probes", 1, 0, '2'}, + {"resize", 1, 0, '3'}, + {"netmask", 1, 0, '4'}, + {NULL}, +}; + +/* Add, del, test parser */ +static ip_set_ip_t adt_parser(unsigned int cmd, const char *arg, void *data) +{ + struct ip_set_req_iphash *mydata = + (struct ip_set_req_iphash *) data; + + parse_ip(arg, &mydata->ip); + if (!mydata->ip) + exit_error(PARAMETER_PROBLEM, + "Zero valued IP address `%s' specified", arg); + + return mydata->ip; +}; + +/* + * Print and save + */ + +static void initheader(struct set *set, const void *data) +{ + struct ip_set_req_iphash_create *header = + (struct ip_set_req_iphash_create *) data; + struct ip_set_iphash *map = + (struct ip_set_iphash *) set->settype->header; + + memset(map, 0, sizeof(struct ip_set_iphash)); + map->hashsize = header->hashsize; + map->probes = header->probes; + map->resize = header->resize; + map->netmask = header->netmask; +} + +static 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 void printheader(struct set *set, unsigned int options) +{ + struct ip_set_iphash *mysetdata = + (struct ip_set_iphash *) set->settype->header; + + printf(" hashsize: %u", mysetdata->hashsize); + printf(" probes: %u", mysetdata->probes); + printf(" resize: %u", mysetdata->resize); + if (mysetdata->netmask == 0xFFFFFFFF) + printf("\n"); + else + printf(" netmask: %d\n", mask_to_bits(mysetdata->netmask)); +} + +static void printips(struct set *set, void *data, size_t len, + unsigned int options) +{ + size_t offset = 0; + ip_set_ip_t *ip; + + while (offset < len) { + ip = data + offset; + if (*ip) + printf("%s\n", ip_tostring(*ip, options)); + offset += sizeof(ip_set_ip_t); + } +} + +static void saveheader(struct set *set, unsigned int options) +{ + struct ip_set_iphash *mysetdata = + (struct ip_set_iphash *) set->settype->header; + + printf("-N %s %s --hashsize %u --probes %u --resize %u", + set->name, set->settype->typename, + mysetdata->hashsize, mysetdata->probes, mysetdata->resize); + if (mysetdata->netmask == 0xFFFFFFFF) + printf("\n"); + else + printf(" --netmask %d\n", mask_to_bits(mysetdata->netmask)); +} + +/* Print save for an IP */ +static void saveips(struct set *set, void *data, size_t len, + unsigned int options) +{ + size_t offset = 0; + ip_set_ip_t *ip; + + while (offset < len) { + ip = data + offset; + if (*ip) + printf("-A %s %s\n", set->name, + ip_tostring(*ip, options)); + offset += sizeof(ip_set_ip_t); + } +} + +static void usage(void) +{ + printf + ("-N set iphash [--hashsize hashsize] [--probes probes ]\n" + " [--resize resize] [--netmask CIDR-netmask]\n" + "-A set IP\n" + "-D set IP\n" + "-T set IP\n"); +} + +static struct settype settype_iphash = { + .typename = SETTYPE_NAME, + .protocol_version = IP_SET_PROTOCOL_VERSION, + + /* Create */ + .create_size = sizeof(struct ip_set_req_iphash_create), + .create_init = &create_init, + .create_parse = &create_parse, + .create_final = &create_final, + .create_opts = create_opts, + + /* Add/del/test */ + .adt_size = sizeof(struct ip_set_req_iphash), + .adt_parser = &adt_parser, + + /* Printing */ + .header_size = sizeof(struct ip_set_iphash), + .initheader = &initheader, + .printheader = &printheader, + .printips = &printips, /* We only have the unsorted version */ + .printips_sorted = &printips, + .saveheader = &saveheader, + .saveips = &saveips, + + /* Bindings */ + .bindip_tostring = &binding_ip_tostring, + .bindip_parse = &parse_ip, + + .usage = &usage, +}; + +static __attribute__((constructor)) void iphash_init(void) +{ + settype_register(&settype_iphash); + +} diff --git a/extensions/ipset/ipset_ipmap.c b/extensions/ipset/ipset_ipmap.c new file mode 100644 index 0000000..9a0619c --- /dev/null +++ b/extensions/ipset/ipset_ipmap.c @@ -0,0 +1,362 @@ +/* Copyright 2000-2004 Joakim Axelsson (gozem@linux.nu) + * Patrick Schaaf (bof@bof.de) + * Jozsef Kadlecsik (kadlec@blackhole.kfki.hu) + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include +#include +#include +#include +#include +/* #include */ + +#include "ip_set_ipmap.h" +#include "ipset.h" + +#define BUFLEN 30; + +#define OPT_CREATE_FROM 0x01U +#define OPT_CREATE_TO 0x02U +#define OPT_CREATE_NETWORK 0x04U +#define OPT_CREATE_NETMASK 0x08U + +#define OPT_ADDDEL_IP 0x01U + +/* Initialize the create. */ +static void create_init(void *data) +{ + struct ip_set_req_ipmap_create *mydata = + (struct ip_set_req_ipmap_create *) data; + + DP("create INIT"); + mydata->netmask = 0xFFFFFFFF; +} + +/* Function which parses command options; returns true if it ate an option */ +static int create_parse(int c, char *argv[], void *data, unsigned int *flags) +{ + struct ip_set_req_ipmap_create *mydata = + (struct ip_set_req_ipmap_create *) data; + unsigned int bits; + + DP("create_parse"); + + switch (c) { + case '1': + parse_ip(optarg, &mydata->from); + + *flags |= OPT_CREATE_FROM; + + DP("--from %x (%s)", mydata->from, + ip_tostring_numeric(mydata->from)); + + break; + + case '2': + parse_ip(optarg, &mydata->to); + + *flags |= OPT_CREATE_TO; + + DP("--to %x (%s)", mydata->to, + ip_tostring_numeric(mydata->to)); + + break; + + case '3': + parse_ipandmask(optarg, &mydata->from, &mydata->to); + + /* Make to the last of from + mask */ + if (mydata->to) + mydata->to = mydata->from | ~(mydata->to); + else { + mydata->from = 0x00000000; + mydata->to = 0xFFFFFFFF; + } + *flags |= OPT_CREATE_NETWORK; + + DP("--network from %x (%s)", + mydata->from, ip_tostring_numeric(mydata->from)); + DP("--network to %x (%s)", + mydata->to, ip_tostring_numeric(mydata->to)); + + break; + + case '4': + if (string_to_number(optarg, 0, 32, &bits)) + exit_error(PARAMETER_PROBLEM, + "Invalid netmask `%s' specified", optarg); + + if (bits != 0) + mydata->netmask = 0xFFFFFFFF << (32 - bits); + + *flags |= OPT_CREATE_NETMASK; + + DP("--netmask %x", mydata->netmask); + + break; + + default: + return 0; + } + + return 1; +} + +#define ERRSTRLEN 256 + +/* Final check; exit if not ok. */ +static void create_final(void *data, unsigned int flags) +{ + struct ip_set_req_ipmap_create *mydata = + (struct ip_set_req_ipmap_create *) data; + ip_set_ip_t range; + char errstr[ERRSTRLEN]; + + if (flags == 0) + exit_error(PARAMETER_PROBLEM, + "Need to specify --from and --to, or --network\n"); + + if (flags & OPT_CREATE_NETWORK) { + /* --network */ + if ((flags & OPT_CREATE_FROM) || (flags & OPT_CREATE_TO)) + exit_error(PARAMETER_PROBLEM, + "Can't specify --from or --to with --network\n"); + } else { + /* --from --to */ + if ((flags & OPT_CREATE_FROM) == 0 + || (flags & OPT_CREATE_TO) == 0) + exit_error(PARAMETER_PROBLEM, + "Need to specify both --from and --to\n"); + } + + DP("from : %x to: %x diff: %x", + mydata->from, mydata->to, + mydata->to - mydata->from); + + if (mydata->from > mydata->to) + exit_error(PARAMETER_PROBLEM, + "From can't be lower than to.\n"); + + if (flags & OPT_CREATE_NETMASK) { + unsigned int mask_bits, netmask_bits; + ip_set_ip_t mask; + + if ((mydata->from & mydata->netmask) != mydata->from) + exit_error(PARAMETER_PROBLEM, + "%s is not a network address according to netmask %d\n", + ip_tostring_numeric(mydata->from), + mask_to_bits(mydata->netmask)); + + mask = range_to_mask(mydata->from, mydata->to, &mask_bits); + if (!mask + && (mydata->from || mydata->to != 0xFFFFFFFF)) { + strncpy(errstr, ip_tostring_numeric(mydata->from), + ERRSTRLEN-2); + errstr[ERRSTRLEN-1] = '\0'; + exit_error(PARAMETER_PROBLEM, + "%s-%s is not a full network (%x)\n", + errstr, + ip_tostring_numeric(mydata->to), mask); + } + netmask_bits = mask_to_bits(mydata->netmask); + + if (netmask_bits <= mask_bits) { + strncpy(errstr, ip_tostring_numeric(mydata->from), + ERRSTRLEN-2); + errstr[ERRSTRLEN-1] = '\0'; + exit_error(PARAMETER_PROBLEM, + "%d netmask specifies larger or equal netblock than %s-%s (%d)\n", + netmask_bits, + errstr, + ip_tostring_numeric(mydata->to), + mask_bits); + } + range = (1<<(netmask_bits - mask_bits)) - 1; + } else { + range = mydata->to - mydata->from; + } + if (range > MAX_RANGE) + exit_error(PARAMETER_PROBLEM, + "Range too large. Max is %d IPs in range\n", + MAX_RANGE+1); +} + +/* Create commandline options */ +static const struct option create_opts[] = { + {"from", 1, 0, '1'}, + {"to", 1, 0, '2'}, + {"network", 1, 0, '3'}, + {"netmask", 1, 0, '4'}, + {NULL}, +}; + +/* Add, del, test parser */ +static ip_set_ip_t adt_parser(unsigned int cmd, const char *arg, void *data) +{ + struct ip_set_req_ipmap *mydata = + (struct ip_set_req_ipmap *) data; + + DP("ipmap: %p %p", arg, data); + + parse_ip(arg, &mydata->ip); + DP("%s", ip_tostring_numeric(mydata->ip)); + + return 1; +} + +/* + * Print and save + */ + +static void initheader(struct set *set, const void *data) +{ + struct ip_set_req_ipmap_create *header = + (struct ip_set_req_ipmap_create *) data; + struct ip_set_ipmap *map = + (struct ip_set_ipmap *) set->settype->header; + + memset(map, 0, sizeof(struct ip_set_ipmap)); + map->first_ip = header->from; + map->last_ip = header->to; + map->netmask = header->netmask; + + if (map->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; + + mask = range_to_mask(header->from, header->to, &mask_bits); + netmask_bits = mask_to_bits(header->netmask); + + DP("bits: %i %i", mask_bits, netmask_bits); + map->hosts = 2 << (32 - netmask_bits - 1); + map->sizeid = 2 << (netmask_bits - mask_bits - 1); + } + + DP("%i %i", map->hosts, map->sizeid ); +} + +static void printheader(struct set *set, unsigned int options) +{ + struct ip_set_ipmap *mysetdata = + (struct ip_set_ipmap *) set->settype->header; + + printf(" from: %s", ip_tostring(mysetdata->first_ip, options)); + printf(" to: %s", ip_tostring(mysetdata->last_ip, options)); + if (mysetdata->netmask == 0xFFFFFFFF) + printf("\n"); + else + printf(" netmask: %d\n", mask_to_bits(mysetdata->netmask)); +} + +static void printips_sorted(struct set *set, void *data, size_t len, + unsigned int options) +{ + struct ip_set_ipmap *mysetdata = + (struct ip_set_ipmap *) set->settype->header; + ip_set_ip_t id; + + for (id = 0; id < mysetdata->sizeid; id++) + if (test_bit(id, data)) + printf("%s\n", + ip_tostring(mysetdata->first_ip + + id * mysetdata->hosts, + options)); +} + +static void saveheader(struct set *set, unsigned int options) +{ + struct ip_set_ipmap *mysetdata = + (struct ip_set_ipmap *) set->settype->header; + + printf("-N %s %s --from %s", + set->name, set->settype->typename, + ip_tostring(mysetdata->first_ip, options)); + printf(" --to %s", + ip_tostring(mysetdata->last_ip, options)); + if (mysetdata->netmask == 0xFFFFFFFF) + printf("\n"); + else + printf(" --netmask %d\n", + mask_to_bits(mysetdata->netmask)); +} + +static void saveips(struct set *set, void *data, size_t len, + unsigned int options) +{ + struct ip_set_ipmap *mysetdata = + (struct ip_set_ipmap *) set->settype->header; + ip_set_ip_t id; + + DP("%s", set->name); + for (id = 0; id < mysetdata->sizeid; id++) + if (test_bit(id, data)) + printf("-A %s %s\n", + set->name, + ip_tostring(mysetdata->first_ip + + id * mysetdata->hosts, + options)); +} + +static void usage(void) +{ + printf + ("-N set ipmap --from IP --to IP [--netmask CIDR-netmask]\n" + "-N set ipmap --network IP/mask [--netmask CIDR-netmask]\n" + "-A set IP\n" + "-D set IP\n" + "-T set IP\n"); +} + +static struct settype settype_ipmap = { + .typename = SETTYPE_NAME, + .protocol_version = IP_SET_PROTOCOL_VERSION, + + /* Create */ + .create_size = sizeof(struct ip_set_req_ipmap_create), + .create_init = &create_init, + .create_parse = &create_parse, + .create_final = &create_final, + .create_opts = create_opts, + + /* Add/del/test */ + .adt_size = sizeof(struct ip_set_req_ipmap), + .adt_parser = &adt_parser, + + /* Printing */ + .header_size = sizeof(struct ip_set_ipmap), + .initheader = &initheader, + .printheader = &printheader, + .printips = &printips_sorted, /* We only have sorted version */ + .printips_sorted = &printips_sorted, + .saveheader = &saveheader, + .saveips = &saveips, + + /* Bindings */ + .bindip_tostring = &binding_ip_tostring, + .bindip_parse = &parse_ip, + + .usage = &usage, +}; + +static __attribute__((constructor)) void ipmap_init(void) +{ + settype_register(&settype_ipmap); + +} diff --git a/extensions/ipset/ipset_ipporthash.c b/extensions/ipset/ipset_ipporthash.c new file mode 100644 index 0000000..eecacf5 --- /dev/null +++ b/extensions/ipset/ipset_ipporthash.c @@ -0,0 +1,375 @@ +/* Copyright 2004 Jozsef Kadlecsik (kadlec@blackhole.kfki.hu) + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "ip_set_ipporthash.h" +#include "ip_set_jhash.h" + +#include "ipset.h" + +#define OPT_CREATE_HASHSIZE 0x01U +#define OPT_CREATE_PROBES 0x02U +#define OPT_CREATE_RESIZE 0x04U +#define OPT_CREATE_NETWORK 0x08U +#define OPT_CREATE_FROM 0x10U +#define OPT_CREATE_TO 0x20U + +/* Initialize the create. */ +static void create_init(void *data) +{ + struct ip_set_req_ipporthash_create *mydata = + (struct ip_set_req_ipporthash_create *) data; + + DP("create INIT"); + + /* Default create parameters */ + mydata->hashsize = 1024; + mydata->probes = 8; + mydata->resize = 50; +} + +/* Function which parses command options; returns true if it ate an option */ +static int create_parse(int c, char *argv[], void *data, unsigned int *flags) +{ + struct ip_set_req_ipporthash_create *mydata = + (struct ip_set_req_ipporthash_create *) data; + ip_set_ip_t value; + + DP("create_parse"); + + switch (c) { + case '1': + + if (string_to_number(optarg, 1, UINT_MAX - 1, &mydata->hashsize)) + exit_error(PARAMETER_PROBLEM, "Invalid hashsize `%s' specified", optarg); + + *flags |= OPT_CREATE_HASHSIZE; + + DP("--hashsize %u", mydata->hashsize); + + break; + + case '2': + + if (string_to_number(optarg, 1, 65535, &value)) + exit_error(PARAMETER_PROBLEM, "Invalid probes `%s' specified", optarg); + + mydata->probes = value; + *flags |= OPT_CREATE_PROBES; + + DP("--probes %u", mydata->probes); + + break; + + case '3': + + if (string_to_number(optarg, 0, 65535, &value)) + exit_error(PARAMETER_PROBLEM, "Invalid resize `%s' specified", optarg); + + mydata->resize = value; + *flags |= OPT_CREATE_RESIZE; + + DP("--resize %u", mydata->resize); + + break; + + case '4': + parse_ip(optarg, &mydata->from); + + *flags |= OPT_CREATE_FROM; + + DP("--from %x (%s)", mydata->from, + ip_tostring_numeric(mydata->from)); + + break; + + case '5': + parse_ip(optarg, &mydata->to); + + *flags |= OPT_CREATE_TO; + + DP("--to %x (%s)", mydata->to, + ip_tostring_numeric(mydata->to)); + + break; + + case '6': + parse_ipandmask(optarg, &mydata->from, &mydata->to); + + /* Make to the last of from + mask */ + if (mydata->to) + mydata->to = mydata->from | ~(mydata->to); + else { + mydata->from = 0x00000000; + mydata->to = 0xFFFFFFFF; + } + *flags |= OPT_CREATE_NETWORK; + + DP("--network from %x (%s)", + mydata->from, ip_tostring_numeric(mydata->from)); + DP("--network to %x (%s)", + mydata->to, ip_tostring_numeric(mydata->to)); + + break; + + default: + return 0; + } + + return 1; +} + +/* Final check; exit if not ok. */ +static void create_final(void *data, unsigned int flags) +{ + struct ip_set_req_ipporthash_create *mydata = + (struct ip_set_req_ipporthash_create *) data; + +#ifdef IPSET_DEBUG + DP("hashsize %u probes %u resize %u", + mydata->hashsize, mydata->probes, mydata->resize); +#endif + + if (flags & OPT_CREATE_NETWORK) { + /* --network */ + if ((flags & OPT_CREATE_FROM) || (flags & OPT_CREATE_TO)) + exit_error(PARAMETER_PROBLEM, + "Can't specify --from or --to with --network\n"); + } else if (flags & (OPT_CREATE_FROM | OPT_CREATE_TO)) { + /* --from --to */ + if (!(flags & OPT_CREATE_FROM) || !(flags & OPT_CREATE_TO)) + exit_error(PARAMETER_PROBLEM, + "Need to specify both --from and --to\n"); + } else { + exit_error(PARAMETER_PROBLEM, + "Need to specify --from and --to, or --network\n"); + + } + + DP("from : %x to: %x diff: %x", + mydata->from, mydata->to, + mydata->to - mydata->from); + + if (mydata->from > mydata->to) + exit_error(PARAMETER_PROBLEM, + "From can't be higher than to.\n"); + + if (mydata->to - mydata->from > MAX_RANGE) + exit_error(PARAMETER_PROBLEM, + "Range too large. Max is %d IPs in range\n", + MAX_RANGE+1); +} + +/* Create commandline options */ +static const struct option create_opts[] = { + {"hashsize", 1, 0, '1'}, + {"probes", 1, 0, '2'}, + {"resize", 1, 0, '3'}, + {"from", 1, 0, '4'}, + {"to", 1, 0, '5'}, + {"network", 1, 0, '6'}, + {NULL}, +}; + +/* Add, del, test parser */ +static ip_set_ip_t adt_parser(unsigned int cmd, const char *arg, void *data) +{ + struct ip_set_req_ipporthash *mydata = + (struct ip_set_req_ipporthash *) data; + char *saved = ipset_strdup(arg); + char *ptr, *tmp = saved; + + DP("ipporthash: %p %p", arg, data); + + ptr = strsep(&tmp, ":%"); + parse_ip(ptr, &mydata->ip); + + if (tmp) + parse_port(tmp, &mydata->port); + else + exit_error(PARAMETER_PROBLEM, + "IP address and port must be specified: ip%%port"); + free(saved); + return 1; +}; + +/* + * Print and save + */ + +static void initheader(struct set *set, const void *data) +{ + struct ip_set_req_ipporthash_create *header = + (struct ip_set_req_ipporthash_create *) data; + struct ip_set_ipporthash *map = + (struct ip_set_ipporthash *) set->settype->header; + + memset(map, 0, sizeof(struct ip_set_ipporthash)); + map->hashsize = header->hashsize; + map->probes = header->probes; + map->resize = header->resize; + map->first_ip = header->from; + map->last_ip = header->to; +} + +static void printheader(struct set *set, unsigned int options) +{ + struct ip_set_ipporthash *mysetdata = + (struct ip_set_ipporthash *) set->settype->header; + + printf(" from: %s", ip_tostring(mysetdata->first_ip, options)); + printf(" to: %s", ip_tostring(mysetdata->last_ip, options)); + printf(" hashsize: %u", mysetdata->hashsize); + printf(" probes: %u", mysetdata->probes); + printf(" resize: %u\n", mysetdata->resize); +} + +static void printips(struct set *set, void *data, size_t len, + unsigned int options) +{ + struct ip_set_ipporthash *mysetdata = + (struct ip_set_ipporthash *) set->settype->header; + size_t offset = 0; + ip_set_ip_t *ipptr, ip; + uint16_t port; + + while (offset < len) { + ipptr = data + offset; + if (*ipptr) { + ip = (*ipptr>>16) + mysetdata->first_ip; + port = (uint16_t) *ipptr; + printf("%s:%s\n", + ip_tostring(ip, options), + port_tostring(port, options)); + } + offset += sizeof(ip_set_ip_t); + } +} + +static void saveheader(struct set *set, unsigned int options) +{ + struct ip_set_ipporthash *mysetdata = + (struct ip_set_ipporthash *) set->settype->header; + + printf("-N %s %s --from %s", + set->name, set->settype->typename, + ip_tostring(mysetdata->first_ip, options)); + printf(" --to %s", + ip_tostring(mysetdata->last_ip, options)); + printf(" --hashsize %u --probes %u --resize %u\n", + mysetdata->hashsize, mysetdata->probes, mysetdata->resize); +} + +/* Print save for an IP */ +static void saveips(struct set *set, void *data, size_t len, + unsigned int options) +{ + struct ip_set_ipporthash *mysetdata = + (struct ip_set_ipporthash *) set->settype->header; + size_t offset = 0; + ip_set_ip_t *ipptr, ip; + uint16_t port; + + while (offset < len) { + ipptr = data + offset; + if (*ipptr) { + ip = (*ipptr>>16) + mysetdata->first_ip; + port = (uint16_t) *ipptr; + printf("-A %s %s:%s\n", set->name, + ip_tostring(ip, options), + port_tostring(port, options)); + } + offset += sizeof(ip_set_ip_t); + } +} + +static char buffer[22]; + +static char * unpack_ipport_tostring(struct set *set, ip_set_ip_t bip, unsigned options) +{ + struct ip_set_ipporthash *mysetdata = + (struct ip_set_ipporthash *) set->settype->header; + ip_set_ip_t ip, port; + + ip = (bip>>16) + mysetdata->first_ip; + port = (uint16_t) bip; + sprintf(buffer, "%s:%s", + ip_tostring(ip, options), port_tostring(port, options)); + + return buffer; +} + +static void usage(void) +{ + printf + ("-N set ipporthash --from IP --to IP\n" + " [--hashsize hashsize] [--probes probes ] [--resize resize]\n" + "-N set ipporthash --network IP/mask\n" + " [--hashsize hashsize] [--probes probes ] [--resize resize]\n" + "-A set IP:port\n" + "-D set IP:port\n" + "-T set IP:port\n"); +} + +static struct settype settype_ipporthash = { + .typename = SETTYPE_NAME, + .protocol_version = IP_SET_PROTOCOL_VERSION, + + /* Create */ + .create_size = sizeof(struct ip_set_req_ipporthash_create), + .create_init = &create_init, + .create_parse = &create_parse, + .create_final = &create_final, + .create_opts = create_opts, + + /* Add/del/test */ + .adt_size = sizeof(struct ip_set_req_ipporthash), + .adt_parser = &adt_parser, + + /* Printing */ + .header_size = sizeof(struct ip_set_ipporthash), + .initheader = &initheader, + .printheader = &printheader, + .printips = &printips, /* We only have the unsorted version */ + .printips_sorted = &printips, + .saveheader = &saveheader, + .saveips = &saveips, + + /* Bindings */ + .bindip_tostring = &unpack_ipport_tostring, + .bindip_parse = &parse_ip, + + .usage = &usage, +}; + +static __attribute__((constructor)) void ipporthash_init(void) +{ + settype_register(&settype_ipporthash); + +} diff --git a/extensions/ipset/ipset_iptree.c b/extensions/ipset/ipset_iptree.c new file mode 100644 index 0000000..565c2c4 --- /dev/null +++ b/extensions/ipset/ipset_iptree.c @@ -0,0 +1,226 @@ +/* Copyright 2005 Jozsef Kadlecsik (kadlec@blackhole.kfki.hu) + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include +#include +#include +#include +#include +#include +#include + +#include "ip_set_iptree.h" +#include "ipset.h" + +#define BUFLEN 30; + +#define OPT_CREATE_TIMEOUT 0x01U + +/* Initialize the create. */ +static void create_init(void *data) +{ + struct ip_set_req_iptree_create *mydata = + (struct ip_set_req_iptree_create *) data; + + DP("create INIT"); + mydata->timeout = 0; +} + +/* Function which parses command options; returns true if it ate an option */ +static int create_parse(int c, char *argv[], void *data, unsigned int *flags) +{ + struct ip_set_req_iptree_create *mydata = + (struct ip_set_req_iptree_create *) data; + + DP("create_parse"); + + switch (c) { + case '1': + string_to_number(optarg, 0, UINT_MAX, &mydata->timeout); + + *flags |= OPT_CREATE_TIMEOUT; + + DP("--timeout %u", mydata->timeout); + + break; + default: + return 0; + } + + return 1; +} + +/* Final check; exit if not ok. */ +static void create_final(void *data, unsigned int flags) +{ +} + +/* Create commandline options */ +static const struct option create_opts[] = { + {"timeout", 1, 0, '1'}, + {NULL}, +}; + +/* Add, del, test parser */ +static ip_set_ip_t adt_parser(unsigned int cmd, const char *arg, void *data) +{ + struct ip_set_req_iptree *mydata = + (struct ip_set_req_iptree *) data; + char *saved = ipset_strdup(arg); + char *ptr, *tmp = saved; + + DP("iptree: %p %p", arg, data); + + ptr = strsep(&tmp, ":%"); + parse_ip(ptr, &mydata->ip); + + if (tmp) + string_to_number(tmp, 0, UINT_MAX, &mydata->timeout); + else + mydata->timeout = 0; + + free(saved); + return 1; +} + +/* + * Print and save + */ + +static void initheader(struct set *set, const void *data) +{ + struct ip_set_req_iptree_create *header = + (struct ip_set_req_iptree_create *) data; + struct ip_set_iptree *map = + (struct ip_set_iptree *) set->settype->header; + + map->timeout = header->timeout; +} + +static void printheader(struct set *set, unsigned int options) +{ + struct ip_set_iptree *mysetdata = + (struct ip_set_iptree *) set->settype->header; + + if (mysetdata->timeout) + printf(" timeout: %u", mysetdata->timeout); + printf("\n"); +} + +static void printips_sorted(struct set *set, void *data, size_t len, + unsigned int options) +{ + struct ip_set_iptree *mysetdata = + (struct ip_set_iptree *) set->settype->header; + struct ip_set_req_iptree *req; + size_t offset = 0; + + while (len >= offset + sizeof(struct ip_set_req_iptree)) { + req = (struct ip_set_req_iptree *)(data + offset); + if (mysetdata->timeout) + printf("%s:%u\n", ip_tostring(req->ip, options), + req->timeout); + else + printf("%s\n", ip_tostring(req->ip, options)); + offset += sizeof(struct ip_set_req_iptree); + } +} + +static void saveheader(struct set *set, unsigned int options) +{ + struct ip_set_iptree *mysetdata = + (struct ip_set_iptree *) set->settype->header; + + if (mysetdata->timeout) + printf("-N %s %s --timeout %u\n", + set->name, set->settype->typename, + mysetdata->timeout); + else + printf("-N %s %s\n", + set->name, set->settype->typename); +} + +static void saveips(struct set *set, void *data, size_t len, + unsigned int options) +{ + struct ip_set_iptree *mysetdata = + (struct ip_set_iptree *) set->settype->header; + struct ip_set_req_iptree *req; + size_t offset = 0; + + DP("%s", set->name); + + while (len >= offset + sizeof(struct ip_set_req_iptree)) { + req = (struct ip_set_req_iptree *)(data + offset); + if (mysetdata->timeout) + printf("-A %s %s:%u\n", + set->name, + ip_tostring(req->ip, options), + req->timeout); + else + printf("-A %s %s\n", + set->name, + ip_tostring(req->ip, options)); + offset += sizeof(struct ip_set_req_iptree); + } +} + +static void usage(void) +{ + printf + ("-N set iptree [--timeout value]\n" + "-A set IP[:timeout]\n" + "-D set IP\n" + "-T set IP\n"); +} + +static struct settype settype_iptree = { + .typename = SETTYPE_NAME, + .protocol_version = IP_SET_PROTOCOL_VERSION, + + /* Create */ + .create_size = sizeof(struct ip_set_req_iptree_create), + .create_init = &create_init, + .create_parse = &create_parse, + .create_final = &create_final, + .create_opts = create_opts, + + /* Add/del/test */ + .adt_size = sizeof(struct ip_set_req_iptree), + .adt_parser = &adt_parser, + + /* Printing */ + .header_size = sizeof(struct ip_set_iptree), + .initheader = &initheader, + .printheader = &printheader, + .printips = &printips_sorted, /* We only have sorted version */ + .printips_sorted = &printips_sorted, + .saveheader = &saveheader, + .saveips = &saveips, + + /* Bindings */ + .bindip_tostring = &binding_ip_tostring, + .bindip_parse = &parse_ip, + + .usage = &usage, +}; + +static __attribute__((constructor)) void iptree_init(void) +{ + settype_register(&settype_iptree); + +} diff --git a/extensions/ipset/ipset_iptreemap.c b/extensions/ipset/ipset_iptreemap.c new file mode 100644 index 0000000..9e3d072 --- /dev/null +++ b/extensions/ipset/ipset_iptreemap.c @@ -0,0 +1,206 @@ +/* Copyright 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 as published by the Free + * Software Foundation; either version 2 of the License, or (at your option) + * any later version. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., 59 Temple + * Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include +#include +#include +#include +#include +#include + +#include "ip_set_iptreemap.h" + +#include "ipset.h" + +#define OPT_CREATE_GC 0x1 + +static void +create_init(void *data) +{ + struct ip_set_req_iptreemap_create *mydata = data; + + mydata->gc_interval = 0; +} + +static int +create_parse(int c, char *argv[], void *data, unsigned int *flags) +{ + struct ip_set_req_iptreemap_create *mydata = data; + + switch (c) { + case 'g': + string_to_number(optarg, 0, UINT_MAX, &mydata->gc_interval); + + *flags |= OPT_CREATE_GC; + break; + default: + return 0; + break; + } + + return 1; +} + +static void +create_final(void *data, unsigned int flags) +{ +} + +static const struct option create_opts[] = { + {"gc", 1, 0, 'g'}, + {NULL}, +}; + +static ip_set_ip_t +adt_parser(unsigned int cmd, const char *arg, void *data) +{ + struct ip_set_req_iptreemap *mydata = data; + ip_set_ip_t mask; + + char *saved = ipset_strdup(arg); + char *ptr, *tmp = saved; + + if (strchr(tmp, '/')) { + parse_ipandmask(tmp, &mydata->start, &mask); + mydata->end = mydata->start | ~mask; + } else { + ptr = strsep(&tmp, ":"); + parse_ip(ptr, &mydata->start); + + if (tmp) { + parse_ip(tmp, &mydata->end); + } else { + mydata->end = mydata->start; + } + } + + return 1; +} + +static void +initheader(struct set *set, const void *data) +{ + const struct ip_set_req_iptreemap_create *header = data; + struct ip_set_iptreemap *map = set->settype->header; + + map->gc_interval = header->gc_interval; +} + +static void +printheader(struct set *set, unsigned int options) +{ + struct ip_set_iptreemap *mysetdata = set->settype->header; + + if (mysetdata->gc_interval) + printf(" gc: %u", mysetdata->gc_interval); + + printf("\n"); +} + +static void +printips_sorted(struct set *set, void *data, size_t len, unsigned int options) +{ + struct ip_set_req_iptreemap *req; + size_t offset = 0; + + while (len >= offset + sizeof(struct ip_set_req_iptreemap)) { + req = data + offset; + + printf("%s", ip_tostring(req->start, options)); + if (req->start != req->end) + printf(":%s", ip_tostring(req->end, options)); + printf("\n"); + + offset += sizeof(struct ip_set_req_iptreemap); + } +} + +static void +saveheader(struct set *set, unsigned int options) +{ + struct ip_set_iptreemap *mysetdata = set->settype->header; + + printf("-N %s %s", set->name, set->settype->typename); + + if (mysetdata->gc_interval) + printf(" --gc %u", mysetdata->gc_interval); + + printf("\n"); +} + +static void +saveips(struct set *set, void *data, size_t len, unsigned int options) +{ + struct ip_set_req_iptreemap *req; + size_t offset = 0; + + while (len >= offset + sizeof(struct ip_set_req_iptreemap)) { + req = data + offset; + + printf("-A %s %s", set->name, ip_tostring(req->start, options)); + + if (req->start != req->end) + printf(":%s", ip_tostring(req->end, options)); + + printf("\n"); + + offset += sizeof(struct ip_set_req_iptreemap); + } +} + +static void +usage(void) +{ + printf( + "-N set iptreemap --gc interval\n" + "-A set IP\n" + "-D set IP\n" + "-T set IP\n" + ); +} + +static struct settype settype_iptreemap = { + .typename = SETTYPE_NAME, + .protocol_version = IP_SET_PROTOCOL_VERSION, + + .create_size = sizeof(struct ip_set_req_iptreemap_create), + .create_init = &create_init, + .create_parse = &create_parse, + .create_final = &create_final, + .create_opts = create_opts, + + .adt_size = sizeof(struct ip_set_req_iptreemap), + .adt_parser = &adt_parser, + + .header_size = sizeof(struct ip_set_iptreemap), + .initheader = &initheader, + .printheader = &printheader, + .printips = &printips_sorted, + .printips_sorted = &printips_sorted, + .saveheader = &saveheader, + .saveips = &saveips, + + .bindip_tostring = &binding_ip_tostring, + .bindip_parse = &parse_ip, + + .usage = &usage, +}; + +static __attribute__((constructor)) void iptreemap_init(void) +{ + settype_register(&settype_iptreemap); +} diff --git a/extensions/ipset/ipset_macipmap.c b/extensions/ipset/ipset_macipmap.c new file mode 100644 index 0000000..3c2f316 --- /dev/null +++ b/extensions/ipset/ipset_macipmap.c @@ -0,0 +1,341 @@ +/* Copyright 2000, 2001, 2002 Joakim Axelsson (gozem@linux.nu) + * Patrick Schaaf (bof@bof.de) + * Martin Josefsson (gandalf@wlug.westbo.se) + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + + +#include +#include +#include +#include +#include +#include +#include + +#include "ip_set_macipmap.h" +#include "ipset.h" + +#define BUFLEN 30; + +#define OPT_CREATE_FROM 0x01U +#define OPT_CREATE_TO 0x02U +#define OPT_CREATE_NETWORK 0x04U +#define OPT_CREATE_MATCHUNSET 0x08U + +#define OPT_ADDDEL_IP 0x01U +#define OPT_ADDDEL_MAC 0x02U + +/* Initialize the create. */ +static void create_init(void *data) +{ + DP("create INIT"); + /* Nothing */ +} + +/* Function which parses command options; returns true if it ate an option */ +static int create_parse(int c, char *argv[], void *data, unsigned int *flags) +{ + struct ip_set_req_macipmap_create *mydata = + (struct ip_set_req_macipmap_create *) data; + + DP("create_parse"); + + switch (c) { + case '1': + parse_ip(optarg, &mydata->from); + + *flags |= OPT_CREATE_FROM; + + DP("--from %x (%s)", mydata->from, + ip_tostring_numeric(mydata->from)); + + break; + + case '2': + parse_ip(optarg, &mydata->to); + + *flags |= OPT_CREATE_TO; + + DP("--to %x (%s)", mydata->to, + ip_tostring_numeric(mydata->to)); + + break; + + case '3': + parse_ipandmask(optarg, &mydata->from, &mydata->to); + + /* Make to the last of from + mask */ + mydata->to = mydata->from | (~mydata->to); + + *flags |= OPT_CREATE_NETWORK; + + DP("--network from %x (%s)", + mydata->from, ip_tostring_numeric(mydata->from)); + DP("--network to %x (%s)", + mydata->to, ip_tostring_numeric(mydata->to)); + + break; + + case '4': + mydata->flags |= IPSET_MACIP_MATCHUNSET; + + *flags |= OPT_CREATE_MATCHUNSET; + + DP("--matchunset"); + + break; + + default: + return 0; + } + + return 1; +} + +/* Final check; exit if not ok. */ +static void create_final(void *data, unsigned int flags) +{ + struct ip_set_req_macipmap_create *mydata = + (struct ip_set_req_macipmap_create *) data; + + if (flags == 0) + exit_error(PARAMETER_PROBLEM, + "Need to specify --from and --to, or --network\n"); + + if (flags & OPT_CREATE_NETWORK) { + /* --network */ + if ((flags & OPT_CREATE_FROM) || (flags & OPT_CREATE_TO)) + exit_error(PARAMETER_PROBLEM, + "Can't specify --from or --to with --network\n"); + } else { + /* --from --to */ + if ((flags & OPT_CREATE_FROM) == 0 + || (flags & OPT_CREATE_TO) == 0) + exit_error(PARAMETER_PROBLEM, + "Need to specify both --from and --to\n"); + } + + + DP("from : %x to: %x diff: %d match unset: %d", mydata->from, + mydata->to, mydata->to - mydata->from, + flags & OPT_CREATE_MATCHUNSET); + + if (mydata->from > mydata->to) + exit_error(PARAMETER_PROBLEM, + "From can't be lower than to.\n"); + + if (mydata->to - mydata->from > MAX_RANGE) + exit_error(PARAMETER_PROBLEM, + "Range too large. Max is %d IPs in range\n", + MAX_RANGE+1); +} + +/* Create commandline options */ +static const struct option create_opts[] = { + {"from", 1, 0, '1'}, + {"to", 1, 0, '2'}, + {"network", 1, 0, '3'}, + {"matchunset", 0, 0, '4'}, + {NULL}, +}; + +static void parse_mac(const char *mac, unsigned char *ethernet) +{ + unsigned int i = 0; + + if (strlen(mac) != ETH_ALEN * 3 - 1) + exit_error(PARAMETER_PROBLEM, "Bad mac address `%s'", mac); + + for (i = 0; i < ETH_ALEN; i++) { + long number; + char *end; + + number = strtol(mac + i * 3, &end, 16); + + if (end == mac + i * 3 + 2 && number >= 0 && number <= 255) + ethernet[i] = number; + else + exit_error(PARAMETER_PROBLEM, + "Bad mac address `%s'", mac); + } +} + +/* Add, del, test parser */ +static ip_set_ip_t adt_parser(unsigned int cmd, const char *arg, void *data) +{ + struct ip_set_req_macipmap *mydata = + (struct ip_set_req_macipmap *) data; + char *saved = ipset_strdup(arg); + char *ptr, *tmp = saved; + + DP("macipmap: %p %p", arg, data); + + ptr = strsep(&tmp, ":%"); + parse_ip(ptr, &mydata->ip); + + if (tmp) + parse_mac(tmp, mydata->ethernet); + else + memset(mydata->ethernet, 0, ETH_ALEN); + + free(saved); + return 1; +} + +/* + * Print and save + */ + +static void initheader(struct set *set, const void *data) +{ + struct ip_set_req_macipmap_create *header = + (struct ip_set_req_macipmap_create *) data; + struct ip_set_macipmap *map = + (struct ip_set_macipmap *) set->settype->header; + + memset(map, 0, sizeof(struct ip_set_macipmap)); + map->first_ip = header->from; + map->last_ip = header->to; + map->flags = header->flags; +} + +static void printheader(struct set *set, unsigned int options) +{ + struct ip_set_macipmap *mysetdata = + (struct ip_set_macipmap *) set->settype->header; + + printf(" from: %s", ip_tostring(mysetdata->first_ip, options)); + printf(" to: %s", ip_tostring(mysetdata->last_ip, options)); + + if (mysetdata->flags & IPSET_MACIP_MATCHUNSET) + printf(" matchunset"); + printf("\n"); +} + +static void print_mac(unsigned char macaddress[ETH_ALEN]) +{ + unsigned int i; + + printf("%02X", macaddress[0]); + for (i = 1; i < ETH_ALEN; i++) + printf(":%02X", macaddress[i]); +} + +static void printips_sorted(struct set *set, void *data, size_t len, + unsigned int options) +{ + struct ip_set_macipmap *mysetdata = + (struct ip_set_macipmap *) set->settype->header; + struct ip_set_macip *table = + (struct ip_set_macip *) data; + u_int32_t addr = mysetdata->first_ip; + + while (addr <= mysetdata->last_ip) { + if (test_bit(IPSET_MACIP_ISSET, + (void *)&table[addr - mysetdata->first_ip].flags)) { + printf("%s:", ip_tostring(addr, options)); + print_mac(table[addr - mysetdata->first_ip]. + ethernet); + printf("\n"); + } + addr++; + } +} + +static void saveheader(struct set *set, unsigned int options) +{ + struct ip_set_macipmap *mysetdata = + (struct ip_set_macipmap *) set->settype->header; + + printf("-N %s %s --from %s", + set->name, set->settype->typename, + ip_tostring(mysetdata->first_ip, options)); + printf(" --to %s", ip_tostring(mysetdata->last_ip, options)); + + if (mysetdata->flags & IPSET_MACIP_MATCHUNSET) + printf(" --matchunset"); + printf("\n"); +} + +static void saveips(struct set *set, void *data, size_t len, + unsigned int options) +{ + struct ip_set_macipmap *mysetdata = + (struct ip_set_macipmap *) set->settype->header; + struct ip_set_macip *table = + (struct ip_set_macip *) data; + u_int32_t addr = mysetdata->first_ip; + + while (addr <= mysetdata->last_ip) { + if (test_bit(IPSET_MACIP_ISSET, + (void *)&table[addr - mysetdata->first_ip].flags)) { + printf("-A %s %s:", + set->name, ip_tostring(addr, options)); + print_mac(table[addr - mysetdata->first_ip]. + ethernet); + printf("\n"); + } + addr++; + } +} + +static void usage(void) +{ + printf + ("-N set macipmap --from IP --to IP [--matchunset]\n" + "-N set macipmap --network IP/mask [--matchunset]\n" + "-A set IP:MAC\n" + "-D set IP[:MAC]\n" + "-T set IP[:MAC]\n"); +} + +static struct settype settype_macipmap = { + .typename = SETTYPE_NAME, + .protocol_version = IP_SET_PROTOCOL_VERSION, + + /* Create */ + .create_size = sizeof(struct ip_set_req_macipmap_create), + .create_init = &create_init, + .create_parse = &create_parse, + .create_final = &create_final, + .create_opts = create_opts, + + /* Add/del/test */ + .adt_size = sizeof(struct ip_set_req_macipmap), + .adt_parser = &adt_parser, + + /* Printing */ + .header_size = sizeof(struct ip_set_macipmap), + .initheader = &initheader, + .printheader = &printheader, + .printips = &printips_sorted, /* We only have sorted version */ + .printips_sorted = &printips_sorted, + .saveheader = &saveheader, + .saveips = &saveips, + + /* Bindings */ + .bindip_tostring = &binding_ip_tostring, + .bindip_parse = &parse_ip, + + .usage = &usage, +}; + +static __attribute__((constructor)) void macipmap_init(void) +{ + settype_register(&settype_macipmap); + +} diff --git a/extensions/ipset/ipset_nethash.c b/extensions/ipset/ipset_nethash.c new file mode 100644 index 0000000..efa7508 --- /dev/null +++ b/extensions/ipset/ipset_nethash.c @@ -0,0 +1,352 @@ +/* Copyright 2004 Jozsef Kadlecsik (kadlec@blackhole.kfki.hu) + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include + +#include "ip_set_nethash.h" +#include "ip_set_jhash.h" + +#include "ipset.h" + +#define BUFLEN 30; + +#define OPT_CREATE_HASHSIZE 0x01U +#define OPT_CREATE_PROBES 0x02U +#define OPT_CREATE_RESIZE 0x04U + +/* Initialize the create. */ +static void create_init(void *data) +{ + struct ip_set_req_nethash_create *mydata = + (struct ip_set_req_nethash_create *) data; + + DP("create INIT"); + + /* Default create parameters */ + mydata->hashsize = 1024; + mydata->probes = 4; + mydata->resize = 50; +} + +/* Function which parses command options; returns true if it ate an option */ +static int create_parse(int c, char *argv[], void *data, unsigned int *flags) +{ + struct ip_set_req_nethash_create *mydata = + (struct ip_set_req_nethash_create *) data; + ip_set_ip_t value; + + DP("create_parse"); + + switch (c) { + case '1': + + if (string_to_number(optarg, 1, UINT_MAX - 1, &mydata->hashsize)) + exit_error(PARAMETER_PROBLEM, "Invalid hashsize `%s' specified", optarg); + + *flags |= OPT_CREATE_HASHSIZE; + + DP("--hashsize %u", mydata->hashsize); + + break; + + case '2': + + if (string_to_number(optarg, 1, 65535, &value)) + exit_error(PARAMETER_PROBLEM, "Invalid probes `%s' specified", optarg); + + mydata->probes = value; + *flags |= OPT_CREATE_PROBES; + + DP("--probes %u", mydata->probes); + + break; + + case '3': + + if (string_to_number(optarg, 0, 65535, &value)) + exit_error(PARAMETER_PROBLEM, "Invalid resize `%s' specified", optarg); + + mydata->resize = value; + *flags |= OPT_CREATE_RESIZE; + + DP("--resize %u", mydata->resize); + + break; + + default: + return 0; + } + + return 1; +} + +/* Final check; exit if not ok. */ +static void create_final(void *data, unsigned int flags) +{ +#ifdef IPSET_DEBUG + struct ip_set_req_nethash_create *mydata = + (struct ip_set_req_nethash_create *) data; + + DP("hashsize %u probes %u resize %u", + mydata->hashsize, mydata->probes, mydata->resize); +#endif +} + +/* Create commandline options */ +static const struct option create_opts[] = { + {"hashsize", 1, 0, '1'}, + {"probes", 1, 0, '2'}, + {"resize", 1, 0, '3'}, + {NULL}, +}; + +/* Add, del, test parser */ +static ip_set_ip_t adt_parser(unsigned int cmd, const char *arg, void *data) +{ + struct ip_set_req_nethash *mydata = + (struct ip_set_req_nethash *) data; + char *saved = ipset_strdup(arg); + char *ptr, *tmp = saved; + ip_set_ip_t cidr; + + ptr = strsep(&tmp, "/"); + + if (tmp == NULL) { + if (cmd == CMD_TEST) + cidr = 32; + else + exit_error(PARAMETER_PROBLEM, + "Missing cidr from `%s'", arg); + } else + if (string_to_number(tmp, 1, 31, &cidr)) + exit_error(PARAMETER_PROBLEM, + "Out of range cidr `%s' specified", arg); + + mydata->cidr = cidr; + parse_ip(ptr, &mydata->ip); + if (!mydata->ip) + exit_error(PARAMETER_PROBLEM, + "Zero valued IP address `%s' specified", ptr); + free(saved); + + return mydata->ip; +}; + +/* + * Print and save + */ + +static void initheader(struct set *set, const void *data) +{ + struct ip_set_req_nethash_create *header = + (struct ip_set_req_nethash_create *) data; + struct ip_set_nethash *map = + (struct ip_set_nethash *) set->settype->header; + + memset(map, 0, sizeof(struct ip_set_nethash)); + map->hashsize = header->hashsize; + map->probes = header->probes; + map->resize = header->resize; +} + +static void printheader(struct set *set, unsigned int options) +{ + struct ip_set_nethash *mysetdata = + (struct ip_set_nethash *) set->settype->header; + + printf(" hashsize: %u", mysetdata->hashsize); + printf(" probes: %u", mysetdata->probes); + printf(" resize: %u\n", mysetdata->resize); +} + +static char buf[20]; + +static char * unpack_ip_tostring(ip_set_ip_t ip, unsigned options) +{ + int i, j = 3; + unsigned char a, b; + + ip = htonl(ip); + for (i = 3; i >= 0; i--) + if (((unsigned char *)&ip)[i] != 0) { + j = i; + break; + } + + a = ((unsigned char *)&ip)[j]; + if (a <= 128) { + a = (a - 1) * 2; + b = 7; + } else if (a <= 192) { + a = (a - 129) * 4; + b = 6; + } else if (a <= 224) { + a = (a - 193) * 8; + b = 5; + } else if (a <= 240) { + a = (a - 225) * 16; + b = 4; + } else if (a <= 248) { + a = (a - 241) * 32; + b = 3; + } else if (a <= 252) { + a = (a - 249) * 64; + b = 2; + } else if (a <= 254) { + a = (a - 253) * 128; + b = 1; + } else { + a = b = 0; + } + ((unsigned char *)&ip)[j] = a; + b += j * 8; + + sprintf(buf, "%u.%u.%u.%u/%u", + ((unsigned char *)&ip)[0], + ((unsigned char *)&ip)[1], + ((unsigned char *)&ip)[2], + ((unsigned char *)&ip)[3], + b); + + DP("%s %s", ip_tostring(ntohl(ip), options), buf); + return buf; +} + +static void printips(struct set *set, void *data, size_t len, + unsigned int options) +{ + size_t offset = 0; + ip_set_ip_t *ip; + + while (offset < len) { + ip = data + offset; + if (*ip) + printf("%s\n", unpack_ip_tostring(*ip, options)); + offset += sizeof(ip_set_ip_t); + } +} + +static void saveheader(struct set *set, unsigned int options) +{ + struct ip_set_nethash *mysetdata = + (struct ip_set_nethash *) set->settype->header; + + printf("-N %s %s --hashsize %u --probes %u --resize %u\n", + set->name, set->settype->typename, + mysetdata->hashsize, mysetdata->probes, mysetdata->resize); +} + +/* Print save for an IP */ +static void saveips(struct set *set, void *data, size_t len, + unsigned int options) +{ + size_t offset = 0; + ip_set_ip_t *ip; + + while (offset < len) { + ip = data + offset; + if (*ip) + printf("-A %s %s\n", set->name, + unpack_ip_tostring(*ip, options)); + offset += sizeof(ip_set_ip_t); + } +} + +static char * net_tostring(struct set *set, ip_set_ip_t ip, unsigned options) +{ + return unpack_ip_tostring(ip, options); +} + +static void parse_net(const char *str, ip_set_ip_t *ip) +{ + char *saved = strdup(str); + char *ptr, *tmp = saved; + ip_set_ip_t cidr; + + ptr = strsep(&tmp, "/"); + + if (tmp == NULL) + exit_error(PARAMETER_PROBLEM, + "Missing cidr from `%s'", str); + + if (string_to_number(tmp, 1, 31, &cidr)) + exit_error(PARAMETER_PROBLEM, + "Out of range cidr `%s' specified", str); + + parse_ip(ptr, ip); + free(saved); + + *ip = pack(*ip, cidr); +} + +static void usage(void) +{ + printf + ("-N set nethash [--hashsize hashsize] [--probes probes ]\n" + " [--resize resize]\n" + "-A set IP/cidr\n" + "-D set IP/cidr\n" + "-T set IP/cidr\n"); +} + +static struct settype settype_nethash = { + .typename = SETTYPE_NAME, + .protocol_version = IP_SET_PROTOCOL_VERSION, + + /* Create */ + .create_size = sizeof(struct ip_set_req_nethash_create), + .create_init = &create_init, + .create_parse = &create_parse, + .create_final = &create_final, + .create_opts = create_opts, + + /* Add/del/test */ + .adt_size = sizeof(struct ip_set_req_nethash), + .adt_parser = &adt_parser, + + /* Printing */ + .header_size = sizeof(struct ip_set_nethash), + .initheader = &initheader, + .printheader = &printheader, + .printips = &printips, /* We only have the unsorted version */ + .printips_sorted = &printips, + .saveheader = &saveheader, + .saveips = &saveips, + + /* Bindings */ + .bindip_tostring = &net_tostring, + .bindip_parse = &parse_net, + + .usage = &usage, +}; + +static __attribute__((constructor)) void nethash_init(void) +{ + settype_register(&settype_nethash); + +} diff --git a/extensions/ipset/ipset_portmap.c b/extensions/ipset/ipset_portmap.c new file mode 100644 index 0000000..5928666 --- /dev/null +++ b/extensions/ipset/ipset_portmap.c @@ -0,0 +1,247 @@ +/* Copyright 2004 Jozsef Kadlecsik (kadlec@blackhole.kfki.hu) + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + + +#include +#include +#include +#include +#include + +#include "ip_set_portmap.h" +#include "ipset.h" + + +#define BUFLEN 30; + +#define OPT_CREATE_FROM 0x01U +#define OPT_CREATE_TO 0x02U + +#define OPT_ADDDEL_PORT 0x01U + +/* Initialize the create. */ +static void create_init(void *data) +{ + DP("create INIT"); + /* Nothing */ +} + +/* Function which parses command options; returns true if it ate an option */ +static int create_parse(int c, char *argv[], void *data, unsigned int *flags) +{ + struct ip_set_req_portmap_create *mydata = + (struct ip_set_req_portmap_create *) data; + + DP("create_parse"); + + switch (c) { + case '1': + parse_port(optarg, &mydata->from); + + *flags |= OPT_CREATE_FROM; + + DP("--from %x (%s)", mydata->from, + port_tostring(mydata->from, 0)); + + break; + + case '2': + parse_port(optarg, &mydata->to); + + *flags |= OPT_CREATE_TO; + + DP("--to %x (%s)", mydata->to, + port_tostring(mydata->to, 0)); + + break; + + default: + return 0; + } + + return 1; +} + +/* Final check; exit if not ok. */ +static void create_final(void *data, unsigned int flags) +{ + struct ip_set_req_portmap_create *mydata = + (struct ip_set_req_portmap_create *) data; + + if (flags == 0) { + exit_error(PARAMETER_PROBLEM, + "Need to specify --from and --to\n"); + } else { + /* --from --to */ + if ((flags & OPT_CREATE_FROM) == 0 + || (flags & OPT_CREATE_TO) == 0) + exit_error(PARAMETER_PROBLEM, + "Need to specify both --from and --to\n"); + } + + DP("from : %x to: %x diff: %d", mydata->from, mydata->to, + mydata->to - mydata->from); + + if (mydata->from > mydata->to) + exit_error(PARAMETER_PROBLEM, + "From can't be lower than to.\n"); + + if (mydata->to - mydata->from > MAX_RANGE) + exit_error(PARAMETER_PROBLEM, + "Range too large. Max is %d ports in range\n", + MAX_RANGE+1); +} + +/* Create commandline options */ +static const struct option create_opts[] = { + {"from", 1, 0, '1'}, + {"to", 1, 0, '2'}, + {NULL}, +}; + +/* Add, del, test parser */ +static ip_set_ip_t adt_parser(unsigned int cmd, const char *arg, void *data) +{ + struct ip_set_req_portmap *mydata = + (struct ip_set_req_portmap *) data; + + parse_port(arg, &mydata->port); + DP("%s", port_tostring(mydata->port, 0)); + + return 1; +} + +/* + * Print and save + */ + +static void initheader(struct set *set, const void *data) +{ + struct ip_set_req_portmap_create *header = + (struct ip_set_req_portmap_create *) data; + struct ip_set_portmap *map = + (struct ip_set_portmap *) set->settype->header; + + memset(map, 0, sizeof(struct ip_set_portmap)); + map->first_port = header->from; + map->last_port = header->to; +} + +static void printheader(struct set *set, unsigned int options) +{ + struct ip_set_portmap *mysetdata = + (struct ip_set_portmap *) set->settype->header; + + printf(" from: %s", port_tostring(mysetdata->first_port, options)); + printf(" to: %s\n", port_tostring(mysetdata->last_port, options)); +} + +static void printports_sorted(struct set *set, void *data, size_t len, + unsigned int options) +{ + struct ip_set_portmap *mysetdata = + (struct ip_set_portmap *) set->settype->header; + u_int32_t addr = mysetdata->first_port; + + DP("%u -- %u", mysetdata->first_port, mysetdata->last_port); + while (addr <= mysetdata->last_port) { + if (test_bit(addr - mysetdata->first_port, data)) + printf("%s\n", port_tostring(addr, options)); + addr++; + } +} + +static char *binding_port_tostring(struct set *set, ip_set_ip_t ip, + unsigned int options) +{ + return port_tostring(ip, options); +} + +static void saveheader(struct set *set, unsigned int options) +{ + struct ip_set_portmap *mysetdata = + (struct ip_set_portmap *) set->settype->header; + + printf("-N %s %s --from %s", + set->name, + set->settype->typename, + port_tostring(mysetdata->first_port, options)); + printf(" --to %s\n", + port_tostring(mysetdata->last_port, options)); +} + +static void saveports(struct set *set, void *data, size_t len, + unsigned int options) +{ + struct ip_set_portmap *mysetdata = + (struct ip_set_portmap *) set->settype->header; + u_int32_t addr = mysetdata->first_port; + + while (addr <= mysetdata->last_port) { + if (test_bit(addr - mysetdata->first_port, data)) + printf("-A %s %s\n", + set->name, + port_tostring(addr, options)); + addr++; + } +} + +static void usage(void) +{ + printf + ("-N set portmap --from PORT --to PORT\n" + "-A set PORT\n" + "-D set PORT\n" + "-T set PORT\n"); +} + +static struct settype settype_portmap = { + .typename = SETTYPE_NAME, + .protocol_version = IP_SET_PROTOCOL_VERSION, + + /* Create */ + .create_size = sizeof(struct ip_set_req_portmap_create), + .create_init = &create_init, + .create_parse = &create_parse, + .create_final = &create_final, + .create_opts = create_opts, + + /* Add/del/test */ + .adt_size = sizeof(struct ip_set_req_portmap), + .adt_parser = &adt_parser, + + /* Printing */ + .header_size = sizeof(struct ip_set_portmap), + .initheader = &initheader, + .printheader = &printheader, + .printips = &printports_sorted, /* We only have sorted version */ + .printips_sorted = &printports_sorted, + .saveheader = &saveheader, + .saveips = &saveports, + + /* Bindings */ + .bindip_tostring = &binding_port_tostring, + .bindip_parse = &parse_port, + + .usage = &usage, +}; + +static __attribute__((constructor)) void portmap_init(void) +{ + settype_register(&settype_portmap); + +} diff --git a/extensions/ipset/ipt_SET.c b/extensions/ipset/ipt_SET.c new file mode 100644 index 0000000..637d32f --- /dev/null +++ b/extensions/ipset/ipt_SET.c @@ -0,0 +1,179 @@ +/* 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 +#include +#include +#include +#include +#include +#include +#include +#include +#include "ipt_set.h" + +static unsigned int +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,24) +target(struct sk_buff *skb, +#else +target(struct sk_buff **pskb, +#endif + const struct net_device *in, + const struct net_device *out, + unsigned int hooknum, +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) + const struct xt_target *target, +#endif +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) + const void *targinfo, + void *userinfo) +#else + const void *targinfo) +#endif +{ + const struct ipt_set_info_target *info = targinfo; +#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 IPT_CONTINUE; +} + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) +static bool +#else +static int +#endif +checkentry(const char *tablename, +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) + const void *e, +#else + const struct ipt_entry *e, +#endif +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) + const struct xt_target *target, +#endif + void *targinfo, +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) + unsigned int targinfosize, +#endif + unsigned int hook_mask) +{ + struct ipt_set_info_target *info = targinfo; + ip_set_id_t index; + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) + if (targinfosize != IPT_ALIGN(sizeof(*info))) { + DP("bad target info size %u", targinfosize); + return 0; + } +#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 0; /* 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 0; /* 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 0; /* error */ + } + + return 1; +} + +static void destroy( +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) + const struct xt_target *target, +#endif +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) + void *targetinfo, unsigned int targetsize) +#else + void *targetinfo) +#endif +{ + struct ipt_set_info_target *info = targetinfo; + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) + 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(info->add_set.index); + if (info->del_set.index != IP_SET_INVALID_ID) + ip_set_put(info->del_set.index); +} + +static struct ipt_target SET_target = { + .name = "SET", +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,21) + .family = AF_INET, +#endif + .target = target, +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) + .targetsize = sizeof(struct ipt_set_info_target), +#endif + .checkentry = checkentry, + .destroy = destroy, + .me = THIS_MODULE +}; + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("iptables IP set target module"); + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,21) +#define ipt_register_target xt_register_target +#define ipt_unregister_target xt_unregister_target +#endif + +static int __init ipt_SET_init(void) +{ + return ipt_register_target(&SET_target); +} + +static void __exit ipt_SET_fini(void) +{ + ipt_unregister_target(&SET_target); +} + +module_init(ipt_SET_init); +module_exit(ipt_SET_fini); diff --git a/extensions/ipset/ipt_set.c b/extensions/ipset/ipt_set.c new file mode 100644 index 0000000..35dfb59 --- /dev/null +++ b/extensions/ipset/ipt_set.c @@ -0,0 +1,159 @@ +/* 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 + +#include +#include "ip_set.h" +#include "ipt_set.h" + +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(2,6,23) +static bool +#else +static int +#endif +match(const struct sk_buff *skb, + const struct net_device *in, + const struct net_device *out, +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) + const struct xt_match *match, +#endif + const void *matchinfo, +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) + int offset, unsigned int protoff, bool *hotdrop) +#elif LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) + int offset, unsigned int protoff, int *hotdrop) +#else + int offset, int *hotdrop) +#endif +{ + const struct ipt_set_info_match *info = matchinfo; + + return match_set(&info->match_set, + skb, + info->match_set.flags[0] & IPSET_MATCH_INV); +} + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,23) +static bool +#else +static int +#endif +checkentry(const char *tablename, +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,16) + const void *inf, +#else + const struct ipt_ip *ip, +#endif +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) + const struct xt_match *match, +#endif + void *matchinfo, +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) + unsigned int matchsize, +#endif + unsigned int hook_mask) +{ + struct ipt_set_info_match *info = matchinfo; + ip_set_id_t index; + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) + if (matchsize != IPT_ALIGN(sizeof(struct ipt_set_info_match))) { + ip_set_printk("invalid matchsize %d", matchsize); + return 0; + } +#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 0; /* error */ + } + if (info->match_set.flags[IP_SET_MAX_BINDINGS] != 0) { + ip_set_printk("That's nasty!"); + return 0; /* error */ + } + + return 1; +} + +static void destroy( +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) + const struct xt_match *match, +#endif +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) + void *matchinfo, unsigned int matchsize) +#else + void *matchinfo) +#endif +{ + struct ipt_set_info_match *info = matchinfo; + +#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19) + if (matchsize != IPT_ALIGN(sizeof(struct ipt_set_info_match))) { + ip_set_printk("invalid matchsize %d", matchsize); + return; + } +#endif + ip_set_put(info->match_set.index); +} + +static struct ipt_match set_match = { + .name = "set", +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,21) + .family = AF_INET, +#endif + .match = &match, +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,17) + .matchsize = sizeof(struct ipt_set_info_match), +#endif + .checkentry = &checkentry, + .destroy = &destroy, + .me = THIS_MODULE +}; + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Jozsef Kadlecsik "); +MODULE_DESCRIPTION("iptables IP set match module"); + +#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,6,21) +#define ipt_register_match xt_register_match +#define ipt_unregister_match xt_unregister_match +#endif + +static int __init ipt_ipset_init(void) +{ + return ipt_register_match(&set_match); +} + +static void __exit ipt_ipset_fini(void) +{ + ipt_unregister_match(&set_match); +} + +module_init(ipt_ipset_init); +module_exit(ipt_ipset_fini); diff --git a/extensions/ipset/ipt_set.h b/extensions/ipset/ipt_set.h new file mode 100644 index 0000000..a905791 --- /dev/null +++ b/extensions/ipset/ipt_set.h @@ -0,0 +1,21 @@ +#ifndef _IPT_SET_H +#define _IPT_SET_H + +#include "ip_set.h" + +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/mconfig b/mconfig index a68d4c4..8d8fcdc 100644 --- a/mconfig +++ b/mconfig @@ -10,5 +10,6 @@ build_TEE=m build_condition=m build_geoip=m build_ipp2p=m +build_ipset=m build_portscan=m build_quota2=m