From e88e514e1fde119b567ff15b215d58b93722697c Mon Sep 17 00:00:00 2001 From: Patrick McHardy Date: Sat, 29 Mar 2014 10:43:01 +0000 Subject: [PATCH 01/25] netfilter: nft_ct: add missing ifdef for NFT_MARK setting The set operation for ct mark is only valid if CONFIG_NF_CONNTRACK_MARK is enabled. Signed-off-by: Patrick McHardy Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nft_ct.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/net/netfilter/nft_ct.c b/net/netfilter/nft_ct.c index bd0d41e69341..a2c45bd8691c 100644 --- a/net/netfilter/nft_ct.c +++ b/net/netfilter/nft_ct.c @@ -268,8 +268,10 @@ static int nft_ct_init_validate_get(const struct nft_expr *expr, static int nft_ct_init_validate_set(uint32_t key) { switch (key) { +#ifdef CONFIG_NF_CONNTRACK_MARK case NFT_CT_MARK: break; +#endif default: return -EOPNOTSUPP; } From d2caa696addd70e686a05b3360c40cb40b106286 Mon Sep 17 00:00:00 2001 From: Patrick McHardy Date: Sat, 29 Mar 2014 10:43:02 +0000 Subject: [PATCH 02/25] netfilter: nft_meta: split nft_meta_init() into two functions for get/set For value spanning multiple registers, we need to validate the length of data loads. In order to add this to nft_meta, we need the length from key validation. Split the nft_meta_init() function into two functions for the get and set operations as preparation for that. Signed-off-by: Patrick McHardy Signed-off-by: Pablo Neira Ayuso Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nft_meta.c | 65 +++++++++++++++++++--------------------- 1 file changed, 30 insertions(+), 35 deletions(-) diff --git a/net/netfilter/nft_meta.c b/net/netfilter/nft_meta.c index 425cf39af890..6d0b8cc27f2a 100644 --- a/net/netfilter/nft_meta.c +++ b/net/netfilter/nft_meta.c @@ -170,21 +170,15 @@ static const struct nla_policy nft_meta_policy[NFTA_META_MAX + 1] = { [NFTA_META_SREG] = { .type = NLA_U32 }, }; -static int nft_meta_init_validate_set(uint32_t key) +static int nft_meta_get_init(const struct nft_ctx *ctx, + const struct nft_expr *expr, + const struct nlattr * const tb[]) { - switch (key) { - case NFT_META_MARK: - case NFT_META_PRIORITY: - case NFT_META_NFTRACE: - return 0; - default: - return -EOPNOTSUPP; - } -} + struct nft_meta *priv = nft_expr_priv(expr); + int err; -static int nft_meta_init_validate_get(uint32_t key) -{ - switch (key) { + priv->key = ntohl(nla_get_be32(tb[NFTA_META_KEY])); + switch (priv->key) { case NFT_META_LEN: case NFT_META_PROTOCOL: case NFT_META_NFPROTO: @@ -205,39 +199,40 @@ static int nft_meta_init_validate_get(uint32_t key) #ifdef CONFIG_NETWORK_SECMARK case NFT_META_SECMARK: #endif - return 0; + break; default: return -EOPNOTSUPP; } + priv->dreg = ntohl(nla_get_be32(tb[NFTA_META_DREG])); + err = nft_validate_output_register(priv->dreg); + if (err < 0) + return err; + + err = nft_validate_data_load(ctx, priv->dreg, NULL, NFT_DATA_VALUE); + if (err < 0) + return err; + + return 0; } -static int nft_meta_init(const struct nft_ctx *ctx, const struct nft_expr *expr, - const struct nlattr * const tb[]) +static int nft_meta_set_init(const struct nft_ctx *ctx, + const struct nft_expr *expr, + const struct nlattr * const tb[]) { struct nft_meta *priv = nft_expr_priv(expr); int err; priv->key = ntohl(nla_get_be32(tb[NFTA_META_KEY])); - - if (tb[NFTA_META_DREG]) { - err = nft_meta_init_validate_get(priv->key); - if (err < 0) - return err; - - priv->dreg = ntohl(nla_get_be32(tb[NFTA_META_DREG])); - err = nft_validate_output_register(priv->dreg); - if (err < 0) - return err; - - return nft_validate_data_load(ctx, priv->dreg, NULL, - NFT_DATA_VALUE); + switch (priv->key) { + case NFT_META_MARK: + case NFT_META_PRIORITY: + case NFT_META_NFTRACE: + break; + default: + return -EOPNOTSUPP; } - err = nft_meta_init_validate_set(priv->key); - if (err < 0) - return err; - priv->sreg = ntohl(nla_get_be32(tb[NFTA_META_SREG])); err = nft_validate_input_register(priv->sreg); if (err < 0) @@ -282,7 +277,7 @@ static const struct nft_expr_ops nft_meta_get_ops = { .type = &nft_meta_type, .size = NFT_EXPR_SIZE(sizeof(struct nft_meta)), .eval = nft_meta_get_eval, - .init = nft_meta_init, + .init = nft_meta_get_init, .dump = nft_meta_get_dump, }; @@ -290,7 +285,7 @@ static const struct nft_expr_ops nft_meta_set_ops = { .type = &nft_meta_type, .size = NFT_EXPR_SIZE(sizeof(struct nft_meta)), .eval = nft_meta_set_eval, - .init = nft_meta_init, + .init = nft_meta_set_init, .dump = nft_meta_set_dump, }; From fe92ca45a170cb8d09c163db23d46634110b3c2f Mon Sep 17 00:00:00 2001 From: Patrick McHardy Date: Sat, 29 Mar 2014 10:43:03 +0000 Subject: [PATCH 03/25] netfilter: nft_ct: split nft_ct_init() into two functions for get/set For value spanning multiple registers, we need to validate the length of data loads. In order to add this to nft_ct, we need the length from key validation. Split the nft_ct_init() function into two functions for the get and set operations as preparation for that. Signed-off-by: Patrick McHardy Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nft_ct.c | 96 +++++++++++++++++++----------------------- 1 file changed, 43 insertions(+), 53 deletions(-) diff --git a/net/netfilter/nft_ct.c b/net/netfilter/nft_ct.c index a2c45bd8691c..cc5603016242 100644 --- a/net/netfilter/nft_ct.c +++ b/net/netfilter/nft_ct.c @@ -215,22 +215,14 @@ static void nft_ct_l3proto_module_put(uint8_t family) nf_ct_l3proto_module_put(family); } -static int nft_ct_init_validate_get(const struct nft_expr *expr, - const struct nlattr * const tb[]) +static int nft_ct_get_init(const struct nft_ctx *ctx, + const struct nft_expr *expr, + const struct nlattr * const tb[]) { struct nft_ct *priv = nft_expr_priv(expr); + int err; - if (tb[NFTA_CT_DIRECTION] != NULL) { - priv->dir = nla_get_u8(tb[NFTA_CT_DIRECTION]); - switch (priv->dir) { - case IP_CT_DIR_ORIGINAL: - case IP_CT_DIR_REPLY: - break; - default: - return -EINVAL; - } - } - + priv->key = ntohl(nla_get_be32(tb[NFTA_CT_KEY])); switch (priv->key) { case NFT_CT_STATE: case NFT_CT_DIRECTION: @@ -262,12 +254,42 @@ static int nft_ct_init_validate_get(const struct nft_expr *expr, return -EOPNOTSUPP; } + if (tb[NFTA_CT_DIRECTION] != NULL) { + priv->dir = nla_get_u8(tb[NFTA_CT_DIRECTION]); + switch (priv->dir) { + case IP_CT_DIR_ORIGINAL: + case IP_CT_DIR_REPLY: + break; + default: + return -EINVAL; + } + } + + priv->dreg = ntohl(nla_get_be32(tb[NFTA_CT_DREG])); + err = nft_validate_output_register(priv->dreg); + if (err < 0) + return err; + + err = nft_validate_data_load(ctx, priv->dreg, NULL, NFT_DATA_VALUE); + if (err < 0) + return err; + + err = nft_ct_l3proto_try_module_get(ctx->afi->family); + if (err < 0) + return err; + return 0; } -static int nft_ct_init_validate_set(uint32_t key) +static int nft_ct_set_init(const struct nft_ctx *ctx, + const struct nft_expr *expr, + const struct nlattr * const tb[]) { - switch (key) { + struct nft_ct *priv = nft_expr_priv(expr); + int err; + + priv->key = ntohl(nla_get_be32(tb[NFTA_CT_KEY])); + switch (priv->key) { #ifdef CONFIG_NF_CONNTRACK_MARK case NFT_CT_MARK: break; @@ -276,42 +298,10 @@ static int nft_ct_init_validate_set(uint32_t key) return -EOPNOTSUPP; } - return 0; -} - -static int nft_ct_init(const struct nft_ctx *ctx, - const struct nft_expr *expr, - const struct nlattr * const tb[]) -{ - struct nft_ct *priv = nft_expr_priv(expr); - int err; - - priv->key = ntohl(nla_get_be32(tb[NFTA_CT_KEY])); - - if (tb[NFTA_CT_DREG]) { - err = nft_ct_init_validate_get(expr, tb); - if (err < 0) - return err; - - priv->dreg = ntohl(nla_get_be32(tb[NFTA_CT_DREG])); - err = nft_validate_output_register(priv->dreg); - if (err < 0) - return err; - - err = nft_validate_data_load(ctx, priv->dreg, NULL, - NFT_DATA_VALUE); - if (err < 0) - return err; - } else { - err = nft_ct_init_validate_set(priv->key); - if (err < 0) - return err; - - priv->sreg = ntohl(nla_get_be32(tb[NFTA_CT_SREG])); - err = nft_validate_input_register(priv->sreg); - if (err < 0) - return err; - } + priv->sreg = ntohl(nla_get_be32(tb[NFTA_CT_SREG])); + err = nft_validate_input_register(priv->sreg); + if (err < 0) + return err; err = nft_ct_l3proto_try_module_get(ctx->afi->family); if (err < 0) @@ -372,7 +362,7 @@ static const struct nft_expr_ops nft_ct_get_ops = { .type = &nft_ct_type, .size = NFT_EXPR_SIZE(sizeof(struct nft_ct)), .eval = nft_ct_get_eval, - .init = nft_ct_init, + .init = nft_ct_get_init, .destroy = nft_ct_destroy, .dump = nft_ct_get_dump, }; @@ -381,7 +371,7 @@ static const struct nft_expr_ops nft_ct_set_ops = { .type = &nft_ct_type, .size = NFT_EXPR_SIZE(sizeof(struct nft_ct)), .eval = nft_ct_set_eval, - .init = nft_ct_init, + .init = nft_ct_set_init, .destroy = nft_ct_destroy, .dump = nft_ct_set_dump, }; From c50b960ccc5981627628302701e93e6aceccdb1c Mon Sep 17 00:00:00 2001 From: Patrick McHardy Date: Fri, 28 Mar 2014 10:19:47 +0000 Subject: [PATCH 04/25] netfilter: nf_tables: implement proper set selection The current set selection simply choses the first set type that provides the requested features, which always results in the rbtree being chosen by virtue of being the first set in the list. What we actually want to do is choose the implementation that can provide the requested features and is optimal from either a performance or memory perspective depending on the characteristics of the elements and the preferences specified by the user. The elements are not known when creating a set. Even if we would provide them for anonymous (literal) sets, we'd still have standalone sets where the elements are not known in advance. We therefore need an abstract description of the data charcteristics. The kernel already knows the size of the key, this patch starts by introducing a nested set description which so far contains only the maximum amount of elements. Based on this the set implementations are changed to provide an estimate of the required amount of memory and the lookup complexity class. The set ops have a new callback ->estimate() that is invoked during set selection. It receives a structure containing the attributes known to the kernel and is supposed to populate a struct nft_set_estimate with the complexity class and, in case the size is known, the complete amount of memory required, or the amount of memory required per element otherwise. Based on the policy specified by the user (performance/memory, defaulting to performance) the kernel will then select the best suited implementation. Even if the set implementation would allow to add more than the specified maximum amount of elements, they are enforced since new implementations might not be able to add more than maximum based on which they were selected. Signed-off-by: Patrick McHardy Signed-off-by: Pablo Neira Ayuso --- include/net/netfilter/nf_tables.h | 46 +++++++++ include/uapi/linux/netfilter/nf_tables.h | 27 +++++ net/netfilter/nf_tables_api.c | 121 ++++++++++++++++++++--- net/netfilter/nft_hash.c | 45 ++++++++- net/netfilter/nft_rbtree.c | 21 ++++ 5 files changed, 242 insertions(+), 18 deletions(-) diff --git a/include/net/netfilter/nf_tables.h b/include/net/netfilter/nf_tables.h index e6bc14d8fa9a..29ff1dc41ef3 100644 --- a/include/net/netfilter/nf_tables.h +++ b/include/net/netfilter/nf_tables.h @@ -145,6 +145,44 @@ struct nft_set_iter { const struct nft_set_elem *elem); }; +/** + * struct nft_set_desc - description of set elements + * + * @klen: key length + * @dlen: data length + * @size: number of set elements + */ +struct nft_set_desc { + unsigned int klen; + unsigned int dlen; + unsigned int size; +}; + +/** + * enum nft_set_class - performance class + * + * @NFT_LOOKUP_O_1: constant, O(1) + * @NFT_LOOKUP_O_LOG_N: logarithmic, O(log N) + * @NFT_LOOKUP_O_N: linear, O(N) + */ +enum nft_set_class { + NFT_SET_CLASS_O_1, + NFT_SET_CLASS_O_LOG_N, + NFT_SET_CLASS_O_N, +}; + +/** + * struct nft_set_estimate - estimation of memory and performance + * characteristics + * + * @size: required memory + * @class: lookup performance class + */ +struct nft_set_estimate { + unsigned int size; + enum nft_set_class class; +}; + /** * struct nft_set_ops - nf_tables set operations * @@ -174,7 +212,11 @@ struct nft_set_ops { struct nft_set_iter *iter); unsigned int (*privsize)(const struct nlattr * const nla[]); + bool (*estimate)(const struct nft_set_desc *desc, + u32 features, + struct nft_set_estimate *est); int (*init)(const struct nft_set *set, + const struct nft_set_desc *desc, const struct nlattr * const nla[]); void (*destroy)(const struct nft_set *set); @@ -194,6 +236,8 @@ void nft_unregister_set(struct nft_set_ops *ops); * @name: name of the set * @ktype: key type (numeric type defined by userspace, not used in the kernel) * @dtype: data type (verdict or numeric type defined by userspace) + * @size: maximum set size + * @nelems: number of elements * @ops: set ops * @flags: set flags * @klen: key length @@ -206,6 +250,8 @@ struct nft_set { char name[IFNAMSIZ]; u32 ktype; u32 dtype; + u32 size; + u32 nelems; /* runtime data below here */ const struct nft_set_ops *ops ____cacheline_aligned; u16 flags; diff --git a/include/uapi/linux/netfilter/nf_tables.h b/include/uapi/linux/netfilter/nf_tables.h index c88ccbfda5f1..160159274cab 100644 --- a/include/uapi/linux/netfilter/nf_tables.h +++ b/include/uapi/linux/netfilter/nf_tables.h @@ -211,6 +211,29 @@ enum nft_set_flags { NFT_SET_MAP = 0x8, }; +/** + * enum nft_set_policies - set selection policy + * + * @NFT_SET_POL_PERFORMANCE: prefer high performance over low memory use + * @NFT_SET_POL_MEMORY: prefer low memory use over high performance + */ +enum nft_set_policies { + NFT_SET_POL_PERFORMANCE, + NFT_SET_POL_MEMORY, +}; + +/** + * enum nft_set_desc_attributes - set element description + * + * @NFTA_SET_DESC_SIZE: number of elements in set (NLA_U32) + */ +enum nft_set_desc_attributes { + NFTA_SET_DESC_UNSPEC, + NFTA_SET_DESC_SIZE, + __NFTA_SET_DESC_MAX +}; +#define NFTA_SET_DESC_MAX (__NFTA_SET_DESC_MAX - 1) + /** * enum nft_set_attributes - nf_tables set netlink attributes * @@ -221,6 +244,8 @@ enum nft_set_flags { * @NFTA_SET_KEY_LEN: key data length (NLA_U32) * @NFTA_SET_DATA_TYPE: mapping data type (NLA_U32) * @NFTA_SET_DATA_LEN: mapping data length (NLA_U32) + * @NFTA_SET_POLICY: selection policy (NLA_U32) + * @NFTA_SET_DESC: set description (NLA_NESTED) */ enum nft_set_attributes { NFTA_SET_UNSPEC, @@ -231,6 +256,8 @@ enum nft_set_attributes { NFTA_SET_KEY_LEN, NFTA_SET_DATA_TYPE, NFTA_SET_DATA_LEN, + NFTA_SET_POLICY, + NFTA_SET_DESC, __NFTA_SET_MAX }; #define NFTA_SET_MAX (__NFTA_SET_MAX - 1) diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 33045a562297..bd3381e16084 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -1912,9 +1912,18 @@ void nft_unregister_set(struct nft_set_ops *ops) } EXPORT_SYMBOL_GPL(nft_unregister_set); -static const struct nft_set_ops *nft_select_set_ops(const struct nlattr * const nla[]) +/* + * Select a set implementation based on the data characteristics and the + * given policy. The total memory use might not be known if no size is + * given, in that case the amount of memory per element is used. + */ +static const struct nft_set_ops * +nft_select_set_ops(const struct nlattr * const nla[], + const struct nft_set_desc *desc, + enum nft_set_policies policy) { - const struct nft_set_ops *ops; + const struct nft_set_ops *ops, *bops; + struct nft_set_estimate est, best; u32 features; #ifdef CONFIG_MODULES @@ -1932,15 +1941,45 @@ static const struct nft_set_ops *nft_select_set_ops(const struct nlattr * const features &= NFT_SET_INTERVAL | NFT_SET_MAP; } - // FIXME: implement selection properly + bops = NULL; + best.size = ~0; + best.class = ~0; + list_for_each_entry(ops, &nf_tables_set_ops, list) { if ((ops->features & features) != features) continue; + if (!ops->estimate(desc, features, &est)) + continue; + + switch (policy) { + case NFT_SET_POL_PERFORMANCE: + if (est.class < best.class) + break; + if (est.class == best.class && est.size < best.size) + break; + continue; + case NFT_SET_POL_MEMORY: + if (est.size < best.size) + break; + if (est.size == best.size && est.class < best.class) + break; + continue; + default: + break; + } + if (!try_module_get(ops->owner)) continue; - return ops; + if (bops != NULL) + module_put(bops->owner); + + bops = ops; + best = est; } + if (bops != NULL) + return bops; + return ERR_PTR(-EOPNOTSUPP); } @@ -1952,6 +1991,12 @@ static const struct nla_policy nft_set_policy[NFTA_SET_MAX + 1] = { [NFTA_SET_KEY_LEN] = { .type = NLA_U32 }, [NFTA_SET_DATA_TYPE] = { .type = NLA_U32 }, [NFTA_SET_DATA_LEN] = { .type = NLA_U32 }, + [NFTA_SET_POLICY] = { .type = NLA_U32 }, + [NFTA_SET_DESC] = { .type = NLA_NESTED }, +}; + +static const struct nla_policy nft_set_desc_policy[NFTA_SET_DESC_MAX + 1] = { + [NFTA_SET_DESC_SIZE] = { .type = NLA_U32 }, }; static int nft_ctx_init_from_setattr(struct nft_ctx *ctx, @@ -2043,6 +2088,7 @@ static int nf_tables_fill_set(struct sk_buff *skb, const struct nft_ctx *ctx, { struct nfgenmsg *nfmsg; struct nlmsghdr *nlh; + struct nlattr *desc; u32 portid = NETLINK_CB(ctx->skb).portid; u32 seq = ctx->nlh->nlmsg_seq; @@ -2076,6 +2122,14 @@ static int nf_tables_fill_set(struct sk_buff *skb, const struct nft_ctx *ctx, goto nla_put_failure; } + desc = nla_nest_start(skb, NFTA_SET_DESC); + if (desc == NULL) + goto nla_put_failure; + if (set->size && + nla_put_be32(skb, NFTA_SET_DESC_SIZE, htonl(set->size))) + goto nla_put_failure; + nla_nest_end(skb, desc); + return nlmsg_end(skb, nlh); nla_put_failure: @@ -2304,6 +2358,23 @@ err: return err; } +static int nf_tables_set_desc_parse(const struct nft_ctx *ctx, + struct nft_set_desc *desc, + const struct nlattr *nla) +{ + struct nlattr *da[NFTA_SET_DESC_MAX + 1]; + int err; + + err = nla_parse_nested(da, NFTA_SET_DESC_MAX, nla, nft_set_desc_policy); + if (err < 0) + return err; + + if (da[NFTA_SET_DESC_SIZE] != NULL) + desc->size = ntohl(nla_get_be32(da[NFTA_SET_DESC_SIZE])); + + return 0; +} + static int nf_tables_newset(struct sock *nlsk, struct sk_buff *skb, const struct nlmsghdr *nlh, const struct nlattr * const nla[]) @@ -2318,7 +2389,8 @@ static int nf_tables_newset(struct sock *nlsk, struct sk_buff *skb, char name[IFNAMSIZ]; unsigned int size; bool create; - u32 ktype, klen, dlen, dtype, flags; + u32 ktype, dtype, flags, policy; + struct nft_set_desc desc; int err; if (nla[NFTA_SET_TABLE] == NULL || @@ -2326,6 +2398,8 @@ static int nf_tables_newset(struct sock *nlsk, struct sk_buff *skb, nla[NFTA_SET_KEY_LEN] == NULL) return -EINVAL; + memset(&desc, 0, sizeof(desc)); + ktype = NFT_DATA_VALUE; if (nla[NFTA_SET_KEY_TYPE] != NULL) { ktype = ntohl(nla_get_be32(nla[NFTA_SET_KEY_TYPE])); @@ -2333,8 +2407,8 @@ static int nf_tables_newset(struct sock *nlsk, struct sk_buff *skb, return -EINVAL; } - klen = ntohl(nla_get_be32(nla[NFTA_SET_KEY_LEN])); - if (klen == 0 || klen > FIELD_SIZEOF(struct nft_data, data)) + desc.klen = ntohl(nla_get_be32(nla[NFTA_SET_KEY_LEN])); + if (desc.klen == 0 || desc.klen > FIELD_SIZEOF(struct nft_data, data)) return -EINVAL; flags = 0; @@ -2346,7 +2420,6 @@ static int nf_tables_newset(struct sock *nlsk, struct sk_buff *skb, } dtype = 0; - dlen = 0; if (nla[NFTA_SET_DATA_TYPE] != NULL) { if (!(flags & NFT_SET_MAP)) return -EINVAL; @@ -2359,15 +2432,25 @@ static int nf_tables_newset(struct sock *nlsk, struct sk_buff *skb, if (dtype != NFT_DATA_VERDICT) { if (nla[NFTA_SET_DATA_LEN] == NULL) return -EINVAL; - dlen = ntohl(nla_get_be32(nla[NFTA_SET_DATA_LEN])); - if (dlen == 0 || - dlen > FIELD_SIZEOF(struct nft_data, data)) + desc.dlen = ntohl(nla_get_be32(nla[NFTA_SET_DATA_LEN])); + if (desc.dlen == 0 || + desc.dlen > FIELD_SIZEOF(struct nft_data, data)) return -EINVAL; } else - dlen = sizeof(struct nft_data); + desc.dlen = sizeof(struct nft_data); } else if (flags & NFT_SET_MAP) return -EINVAL; + policy = NFT_SET_POL_PERFORMANCE; + if (nla[NFTA_SET_POLICY] != NULL) + policy = ntohl(nla_get_be32(nla[NFTA_SET_POLICY])); + + if (nla[NFTA_SET_DESC] != NULL) { + err = nf_tables_set_desc_parse(&ctx, &desc, nla[NFTA_SET_DESC]); + if (err < 0) + return err; + } + create = nlh->nlmsg_flags & NLM_F_CREATE ? true : false; afi = nf_tables_afinfo_lookup(net, nfmsg->nfgen_family, create); @@ -2398,7 +2481,7 @@ static int nf_tables_newset(struct sock *nlsk, struct sk_buff *skb, if (!(nlh->nlmsg_flags & NLM_F_CREATE)) return -ENOENT; - ops = nft_select_set_ops(nla); + ops = nft_select_set_ops(nla, &desc, policy); if (IS_ERR(ops)) return PTR_ERR(ops); @@ -2419,12 +2502,13 @@ static int nf_tables_newset(struct sock *nlsk, struct sk_buff *skb, INIT_LIST_HEAD(&set->bindings); set->ops = ops; set->ktype = ktype; - set->klen = klen; + set->klen = desc.klen; set->dtype = dtype; - set->dlen = dlen; + set->dlen = desc.dlen; set->flags = flags; + set->size = desc.size; - err = ops->init(set, nla); + err = ops->init(set, &desc, nla); if (err < 0) goto err2; @@ -2733,6 +2817,9 @@ static int nft_add_set_elem(const struct nft_ctx *ctx, struct nft_set *set, enum nft_registers dreg; int err; + if (set->size && set->nelems == set->size) + return -ENFILE; + err = nla_parse_nested(nla, NFTA_SET_ELEM_MAX, attr, nft_set_elem_policy); if (err < 0) @@ -2798,6 +2885,7 @@ static int nft_add_set_elem(const struct nft_ctx *ctx, struct nft_set *set, err = set->ops->insert(set, &elem); if (err < 0) goto err3; + set->nelems++; return 0; @@ -2867,6 +2955,7 @@ static int nft_del_setelem(const struct nft_ctx *ctx, struct nft_set *set, goto err2; set->ops->remove(set, &elem); + set->nelems--; nft_data_uninit(&elem.key, NFT_DATA_VALUE); if (set->flags & NFT_SET_MAP) diff --git a/net/netfilter/nft_hash.c b/net/netfilter/nft_hash.c index 3b1ad876d6b0..01884b315c4a 100644 --- a/net/netfilter/nft_hash.c +++ b/net/netfilter/nft_hash.c @@ -12,6 +12,7 @@ #include #include #include +#include #include #include #include @@ -19,7 +20,7 @@ #include #include -#define NFT_HASH_MIN_SIZE 4 +#define NFT_HASH_MIN_SIZE 4UL struct nft_hash { struct nft_hash_table __rcu *tbl; @@ -82,6 +83,11 @@ static void nft_hash_tbl_free(const struct nft_hash_table *tbl) kfree(tbl); } +static unsigned int nft_hash_tbl_size(unsigned int nelem) +{ + return max(roundup_pow_of_two(nelem * 4 / 3), NFT_HASH_MIN_SIZE); +} + static struct nft_hash_table *nft_hash_tbl_alloc(unsigned int nbuckets) { struct nft_hash_table *tbl; @@ -335,17 +341,23 @@ static unsigned int nft_hash_privsize(const struct nlattr * const nla[]) } static int nft_hash_init(const struct nft_set *set, + const struct nft_set_desc *desc, const struct nlattr * const tb[]) { struct nft_hash *priv = nft_set_priv(set); struct nft_hash_table *tbl; + unsigned int size; if (unlikely(!nft_hash_rnd_initted)) { get_random_bytes(&nft_hash_rnd, 4); nft_hash_rnd_initted = true; } - tbl = nft_hash_tbl_alloc(NFT_HASH_MIN_SIZE); + size = NFT_HASH_MIN_SIZE; + if (desc->size) + size = nft_hash_tbl_size(desc->size); + + tbl = nft_hash_tbl_alloc(size); if (tbl == NULL) return -ENOMEM; RCU_INIT_POINTER(priv->tbl, tbl); @@ -369,8 +381,37 @@ static void nft_hash_destroy(const struct nft_set *set) kfree(tbl); } +static bool nft_hash_estimate(const struct nft_set_desc *desc, u32 features, + struct nft_set_estimate *est) +{ + unsigned int esize; + + esize = sizeof(struct nft_hash_elem); + if (features & NFT_SET_MAP) + esize += FIELD_SIZEOF(struct nft_hash_elem, data[0]); + + if (desc->size) { + est->size = sizeof(struct nft_hash) + + nft_hash_tbl_size(desc->size) * + sizeof(struct nft_hash_elem *) + + desc->size * esize; + } else { + /* Resizing happens when the load drops below 30% or goes + * above 75%. The average of 52.5% load (approximated by 50%) + * is used for the size estimation of the hash buckets, + * meaning we calculate two buckets per element. + */ + est->size = esize + 2 * sizeof(struct nft_hash_elem *); + } + + est->class = NFT_SET_CLASS_O_1; + + return true; +} + static struct nft_set_ops nft_hash_ops __read_mostly = { .privsize = nft_hash_privsize, + .estimate = nft_hash_estimate, .init = nft_hash_init, .destroy = nft_hash_destroy, .get = nft_hash_get, diff --git a/net/netfilter/nft_rbtree.c b/net/netfilter/nft_rbtree.c index e21d69d13506..072e611e9f71 100644 --- a/net/netfilter/nft_rbtree.c +++ b/net/netfilter/nft_rbtree.c @@ -201,6 +201,7 @@ static unsigned int nft_rbtree_privsize(const struct nlattr * const nla[]) } static int nft_rbtree_init(const struct nft_set *set, + const struct nft_set_desc *desc, const struct nlattr * const nla[]) { struct nft_rbtree *priv = nft_set_priv(set); @@ -222,8 +223,28 @@ static void nft_rbtree_destroy(const struct nft_set *set) } } +static bool nft_rbtree_estimate(const struct nft_set_desc *desc, u32 features, + struct nft_set_estimate *est) +{ + unsigned int nsize; + + nsize = sizeof(struct nft_rbtree_elem); + if (features & NFT_SET_MAP) + nsize += FIELD_SIZEOF(struct nft_rbtree_elem, data[0]); + + if (desc->size) + est->size = sizeof(struct nft_rbtree) + desc->size * nsize; + else + est->size = nsize; + + est->class = NFT_SET_CLASS_O_LOG_N; + + return true; +} + static struct nft_set_ops nft_rbtree_ops __read_mostly = { .privsize = nft_rbtree_privsize, + .estimate = nft_rbtree_estimate, .init = nft_rbtree_init, .destroy = nft_rbtree_destroy, .insert = nft_rbtree_insert, From 2c96c25d114083c7661936ad4d27e6a2c46c527d Mon Sep 17 00:00:00 2001 From: Patrick McHardy Date: Fri, 28 Mar 2014 10:19:48 +0000 Subject: [PATCH 05/25] netfilter: nft_hash: use set global element counter instead of private one Now that nf_tables performs global accounting of set elements, it is not needed in the hash type anymore. Signed-off-by: Patrick McHardy Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nft_hash.c | 9 ++------- 1 file changed, 2 insertions(+), 7 deletions(-) diff --git a/net/netfilter/nft_hash.c b/net/netfilter/nft_hash.c index 01884b315c4a..1dfeb6786832 100644 --- a/net/netfilter/nft_hash.c +++ b/net/netfilter/nft_hash.c @@ -28,7 +28,6 @@ struct nft_hash { struct nft_hash_table { unsigned int size; - unsigned int elements; struct nft_hash_elem __rcu *buckets[]; }; @@ -167,7 +166,6 @@ static int nft_hash_tbl_expand(const struct nft_set *set, struct nft_hash *priv) break; } } - ntbl->elements = tbl->elements; /* Publish new table */ rcu_assign_pointer(priv->tbl, ntbl); @@ -207,7 +205,6 @@ static int nft_hash_tbl_shrink(const struct nft_set *set, struct nft_hash *priv) ; RCU_INIT_POINTER(*pprev, tbl->buckets[i + ntbl->size]); } - ntbl->elements = tbl->elements; /* Publish new table */ rcu_assign_pointer(priv->tbl, ntbl); @@ -243,10 +240,9 @@ static int nft_hash_insert(const struct nft_set *set, h = nft_hash_data(&he->key, tbl->size, set->klen); RCU_INIT_POINTER(he->next, tbl->buckets[h]); rcu_assign_pointer(tbl->buckets[h], he); - tbl->elements++; /* Expand table when exceeding 75% load */ - if (tbl->elements > tbl->size / 4 * 3) + if (set->nelems + 1 > tbl->size / 4 * 3) nft_hash_tbl_expand(set, priv); return 0; @@ -274,10 +270,9 @@ static void nft_hash_remove(const struct nft_set *set, RCU_INIT_POINTER(*pprev, he->next); synchronize_rcu(); kfree(he); - tbl->elements--; /* Shrink table beneath 30% load */ - if (tbl->elements < tbl->size * 3 / 10 && + if (set->nelems - 1 < tbl->size * 3 / 10 && tbl->size > NFT_HASH_MIN_SIZE) nft_hash_tbl_shrink(set, priv); } From d60ce62fb5949ea6bf09323462c15d05c9e76200 Mon Sep 17 00:00:00 2001 From: Arturo Borrero Date: Tue, 1 Apr 2014 14:06:07 +0200 Subject: [PATCH 06/25] netfilter: nf_tables: add set_elem notifications This patch adds set_elems notifications. When a set_elem is added/deleted, all listening peers in userspace will receive the corresponding notification. Signed-off-by: Arturo Borrero Gonzalez Acked-by: Patrick McHardy Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_tables_api.c | 82 +++++++++++++++++++++++++++++++++++ 1 file changed, 82 insertions(+) diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index bd3381e16084..60feca9aa70a 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -2807,6 +2807,85 @@ static int nf_tables_getsetelem(struct sock *nlsk, struct sk_buff *skb, return -EOPNOTSUPP; } +static int nf_tables_fill_setelem_info(struct sk_buff *skb, + const struct nft_ctx *ctx, u32 seq, + u32 portid, int event, u16 flags, + const struct nft_set *set, + const struct nft_set_elem *elem) +{ + struct nfgenmsg *nfmsg; + struct nlmsghdr *nlh; + struct nlattr *nest; + int err; + + event |= NFNL_SUBSYS_NFTABLES << 8; + nlh = nlmsg_put(skb, portid, seq, event, sizeof(struct nfgenmsg), + flags); + if (nlh == NULL) + goto nla_put_failure; + + nfmsg = nlmsg_data(nlh); + nfmsg->nfgen_family = ctx->afi->family; + nfmsg->version = NFNETLINK_V0; + nfmsg->res_id = 0; + + if (nla_put_string(skb, NFTA_SET_TABLE, ctx->table->name)) + goto nla_put_failure; + if (nla_put_string(skb, NFTA_SET_NAME, set->name)) + goto nla_put_failure; + + nest = nla_nest_start(skb, NFTA_SET_ELEM_LIST_ELEMENTS); + if (nest == NULL) + goto nla_put_failure; + + err = nf_tables_fill_setelem(skb, set, elem); + if (err < 0) + goto nla_put_failure; + + nla_nest_end(skb, nest); + + return nlmsg_end(skb, nlh); + +nla_put_failure: + nlmsg_trim(skb, nlh); + return -1; +} + +static int nf_tables_setelem_notify(const struct nft_ctx *ctx, + const struct nft_set *set, + const struct nft_set_elem *elem, + int event, u16 flags) +{ + const struct sk_buff *oskb = ctx->skb; + struct net *net = sock_net(oskb->sk); + u32 portid = NETLINK_CB(oskb).portid; + bool report = nlmsg_report(ctx->nlh); + struct sk_buff *skb; + int err; + + if (!report && !nfnetlink_has_listeners(net, NFNLGRP_NFTABLES)) + return 0; + + err = -ENOBUFS; + skb = nlmsg_new(NLMSG_GOODSIZE, GFP_KERNEL); + if (skb == NULL) + goto err; + + err = nf_tables_fill_setelem_info(skb, ctx, 0, portid, event, flags, + set, elem); + if (err < 0) { + kfree_skb(skb); + goto err; + } + + err = nfnetlink_send(skb, net, portid, NFNLGRP_NFTABLES, report, + GFP_KERNEL); +err: + if (err < 0) + nfnetlink_set_err(net, portid, NFNLGRP_NFTABLES, err); + return err; +} + static int nft_add_set_elem(const struct nft_ctx *ctx, struct nft_set *set, const struct nlattr *attr) { @@ -2887,6 +2966,7 @@ static int nft_add_set_elem(const struct nft_ctx *ctx, struct nft_set *set, goto err3; set->nelems++; + nf_tables_setelem_notify(ctx, set, &elem, NFT_MSG_NEWSETELEM, 0); return 0; err3: @@ -2957,6 +3037,8 @@ static int nft_del_setelem(const struct nft_ctx *ctx, struct nft_set *set, set->ops->remove(set, &elem); set->nelems--; + nf_tables_setelem_notify(ctx, set, &elem, NFT_MSG_DELSETELEM, 0); + nft_data_uninit(&elem.key, NFT_DATA_VALUE); if (set->flags & NFT_SET_MAP) nft_data_uninit(&elem.data, set->dtype); From 60eb18943bd787ac12685e863a37ea440f70abba Mon Sep 17 00:00:00 2001 From: Patrick McHardy Date: Fri, 7 Mar 2014 12:34:05 +0100 Subject: [PATCH 07/25] netfilter: nf_tables: handle more than 8 * PAGE_SIZE set name allocations We currently have a limit of 8 * PAGE_SIZE anonymous sets. Lift that limit by continuing the scan if the entire page is exhausted. Signed-off-by: Patrick McHardy Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_tables_api.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-) diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 60feca9aa70a..1c5e6335b2e3 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -2049,7 +2049,7 @@ static int nf_tables_set_alloc_name(struct nft_ctx *ctx, struct nft_set *set, const struct nft_set *i; const char *p; unsigned long *inuse; - unsigned int n = 0; + unsigned int n = 0, min = 0; p = strnchr(name, IFNAMSIZ, '%'); if (p != NULL) { @@ -2059,23 +2059,28 @@ static int nf_tables_set_alloc_name(struct nft_ctx *ctx, struct nft_set *set, inuse = (unsigned long *)get_zeroed_page(GFP_KERNEL); if (inuse == NULL) return -ENOMEM; - +cont: list_for_each_entry(i, &ctx->table->sets, list) { int tmp; if (!sscanf(i->name, name, &tmp)) continue; - if (tmp < 0 || tmp >= BITS_PER_BYTE * PAGE_SIZE) + if (tmp < min || tmp >= min + BITS_PER_BYTE * PAGE_SIZE) continue; - set_bit(tmp, inuse); + set_bit(tmp - min, inuse); } n = find_first_zero_bit(inuse, BITS_PER_BYTE * PAGE_SIZE); + if (n >= BITS_PER_BYTE * PAGE_SIZE) { + min += BITS_PER_BYTE * PAGE_SIZE; + memset(inuse, 0, PAGE_SIZE); + goto cont; + } free_page((unsigned long)inuse); } - snprintf(set->name, sizeof(set->name), name, n); + snprintf(set->name, sizeof(set->name), name, min + n); list_for_each_entry(i, &ctx->table->sets, list) { if (!strcmp(set->name, i->name)) return -ENFILE; From 758dbcecf180a161e15971a2674fa6e65622a281 Mon Sep 17 00:00:00 2001 From: Tomasz Bursztyka Date: Mon, 14 Apr 2014 15:41:26 +0300 Subject: [PATCH 08/25] netfilter: nf_tables: Stack expression type depending on their family To ensure family tight expression gets selected in priority to family agnostic ones. Signed-off-by: Tomasz Bursztyka Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_tables_api.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 1c5e6335b2e3..1d0c174d30fc 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -1093,7 +1093,10 @@ static void nft_ctx_init(struct nft_ctx *ctx, int nft_register_expr(struct nft_expr_type *type) { nfnl_lock(NFNL_SUBSYS_NFTABLES); - list_add_tail(&type->list, &nf_tables_expressions); + if (type->family == NFPROTO_UNSPEC) + list_add_tail(&type->list, &nf_tables_expressions); + else + list_add(&type->list, &nf_tables_expressions); nfnl_unlock(NFNL_SUBSYS_NFTABLES); return 0; } From aa45660c6b59388fac3995a8c2998d710ef28fd4 Mon Sep 17 00:00:00 2001 From: Tomasz Bursztyka Date: Mon, 14 Apr 2014 15:41:27 +0300 Subject: [PATCH 09/25] netfilter: nf_tables: Make meta expression core functions public This will be useful to create network family dedicated META expression as for NFPROTO_BRIDGE for instance. Signed-off-by: Tomasz Bursztyka Signed-off-by: Pablo Neira Ayuso --- include/net/netfilter/nft_meta.h | 36 +++++++++++++++++++++++ net/netfilter/nft_meta.c | 50 ++++++++++++++++---------------- 2 files changed, 61 insertions(+), 25 deletions(-) create mode 100644 include/net/netfilter/nft_meta.h diff --git a/include/net/netfilter/nft_meta.h b/include/net/netfilter/nft_meta.h new file mode 100644 index 000000000000..0ee47c3e2e31 --- /dev/null +++ b/include/net/netfilter/nft_meta.h @@ -0,0 +1,36 @@ +#ifndef _NFT_META_H_ +#define _NFT_META_H_ + +struct nft_meta { + enum nft_meta_keys key:8; + union { + enum nft_registers dreg:8; + enum nft_registers sreg:8; + }; +}; + +extern const struct nla_policy nft_meta_policy[]; + +int nft_meta_get_init(const struct nft_ctx *ctx, + const struct nft_expr *expr, + const struct nlattr * const tb[]); + +int nft_meta_set_init(const struct nft_ctx *ctx, + const struct nft_expr *expr, + const struct nlattr * const tb[]); + +int nft_meta_get_dump(struct sk_buff *skb, + const struct nft_expr *expr); + +int nft_meta_set_dump(struct sk_buff *skb, + const struct nft_expr *expr); + +void nft_meta_get_eval(const struct nft_expr *expr, + struct nft_data data[NFT_REG_MAX + 1], + const struct nft_pktinfo *pkt); + +void nft_meta_set_eval(const struct nft_expr *expr, + struct nft_data data[NFT_REG_MAX + 1], + const struct nft_pktinfo *pkt); + +#endif diff --git a/net/netfilter/nft_meta.c b/net/netfilter/nft_meta.c index 6d0b8cc27f2a..852b178c6ae7 100644 --- a/net/netfilter/nft_meta.c +++ b/net/netfilter/nft_meta.c @@ -18,18 +18,11 @@ #include #include /* for TCP_TIME_WAIT */ #include +#include -struct nft_meta { - enum nft_meta_keys key:8; - union { - enum nft_registers dreg:8; - enum nft_registers sreg:8; - }; -}; - -static void nft_meta_get_eval(const struct nft_expr *expr, - struct nft_data data[NFT_REG_MAX + 1], - const struct nft_pktinfo *pkt) +void nft_meta_get_eval(const struct nft_expr *expr, + struct nft_data data[NFT_REG_MAX + 1], + const struct nft_pktinfo *pkt) { const struct nft_meta *priv = nft_expr_priv(expr); const struct sk_buff *skb = pkt->skb; @@ -140,10 +133,11 @@ static void nft_meta_get_eval(const struct nft_expr *expr, err: data[NFT_REG_VERDICT].verdict = NFT_BREAK; } +EXPORT_SYMBOL_GPL(nft_meta_get_eval); -static void nft_meta_set_eval(const struct nft_expr *expr, - struct nft_data data[NFT_REG_MAX + 1], - const struct nft_pktinfo *pkt) +void nft_meta_set_eval(const struct nft_expr *expr, + struct nft_data data[NFT_REG_MAX + 1], + const struct nft_pktinfo *pkt) { const struct nft_meta *meta = nft_expr_priv(expr); struct sk_buff *skb = pkt->skb; @@ -163,16 +157,18 @@ static void nft_meta_set_eval(const struct nft_expr *expr, WARN_ON(1); } } +EXPORT_SYMBOL_GPL(nft_meta_set_eval); -static const struct nla_policy nft_meta_policy[NFTA_META_MAX + 1] = { +const struct nla_policy nft_meta_policy[NFTA_META_MAX + 1] = { [NFTA_META_DREG] = { .type = NLA_U32 }, [NFTA_META_KEY] = { .type = NLA_U32 }, [NFTA_META_SREG] = { .type = NLA_U32 }, }; +EXPORT_SYMBOL_GPL(nft_meta_policy); -static int nft_meta_get_init(const struct nft_ctx *ctx, - const struct nft_expr *expr, - const struct nlattr * const tb[]) +int nft_meta_get_init(const struct nft_ctx *ctx, + const struct nft_expr *expr, + const struct nlattr * const tb[]) { struct nft_meta *priv = nft_expr_priv(expr); int err; @@ -215,10 +211,11 @@ static int nft_meta_get_init(const struct nft_ctx *ctx, return 0; } +EXPORT_SYMBOL_GPL(nft_meta_get_init); -static int nft_meta_set_init(const struct nft_ctx *ctx, - const struct nft_expr *expr, - const struct nlattr * const tb[]) +int nft_meta_set_init(const struct nft_ctx *ctx, + const struct nft_expr *expr, + const struct nlattr * const tb[]) { struct nft_meta *priv = nft_expr_priv(expr); int err; @@ -240,9 +237,10 @@ static int nft_meta_set_init(const struct nft_ctx *ctx, return 0; } +EXPORT_SYMBOL_GPL(nft_meta_set_init); -static int nft_meta_get_dump(struct sk_buff *skb, - const struct nft_expr *expr) +int nft_meta_get_dump(struct sk_buff *skb, + const struct nft_expr *expr) { const struct nft_meta *priv = nft_expr_priv(expr); @@ -255,9 +253,10 @@ static int nft_meta_get_dump(struct sk_buff *skb, nla_put_failure: return -1; } +EXPORT_SYMBOL_GPL(nft_meta_get_dump); -static int nft_meta_set_dump(struct sk_buff *skb, - const struct nft_expr *expr) +int nft_meta_set_dump(struct sk_buff *skb, + const struct nft_expr *expr) { const struct nft_meta *priv = nft_expr_priv(expr); @@ -271,6 +270,7 @@ static int nft_meta_set_dump(struct sk_buff *skb, nla_put_failure: return -1; } +EXPORT_SYMBOL_GPL(nft_meta_set_dump); static struct nft_expr_type nft_meta_type; static const struct nft_expr_ops nft_meta_get_ops = { From f5efc696cc711021cc73e7543cc3038e58459707 Mon Sep 17 00:00:00 2001 From: Tomasz Bursztyka Date: Mon, 14 Apr 2014 15:41:28 +0300 Subject: [PATCH 10/25] netfilter: nf_tables: Add meta expression key for bridge interface name NFT_META_BRI_IIFNAME to get packet input bridge interface name NFT_META_BRI_OIFNAME to get packet output bridge interface name Such meta key are accessible only through NFPROTO_BRIDGE family, on a dedicated nft meta module: nft_meta_bridge. Suggested-by: Pablo Neira Ayuso Signed-off-by: Tomasz Bursztyka Signed-off-by: Pablo Neira Ayuso --- include/uapi/linux/netfilter/nf_tables.h | 4 + net/bridge/Makefile | 2 +- net/bridge/netfilter/Kconfig | 14 ++- net/bridge/netfilter/Makefile | 1 + net/bridge/netfilter/nft_meta_bridge.c | 139 +++++++++++++++++++++++ 5 files changed, 158 insertions(+), 2 deletions(-) create mode 100644 net/bridge/netfilter/nft_meta_bridge.c diff --git a/include/uapi/linux/netfilter/nf_tables.h b/include/uapi/linux/netfilter/nf_tables.h index 160159274cab..7d6433f66bf8 100644 --- a/include/uapi/linux/netfilter/nf_tables.h +++ b/include/uapi/linux/netfilter/nf_tables.h @@ -563,6 +563,8 @@ enum nft_exthdr_attributes { * @NFT_META_SECMARK: packet secmark (skb->secmark) * @NFT_META_NFPROTO: netfilter protocol * @NFT_META_L4PROTO: layer 4 protocol number + * @NFT_META_BRI_IIFNAME: packet input bridge interface name + * @NFT_META_BRI_OIFNAME: packet output bridge interface name */ enum nft_meta_keys { NFT_META_LEN, @@ -582,6 +584,8 @@ enum nft_meta_keys { NFT_META_SECMARK, NFT_META_NFPROTO, NFT_META_L4PROTO, + NFT_META_BRI_IIFNAME, + NFT_META_BRI_OIFNAME, }; /** diff --git a/net/bridge/Makefile b/net/bridge/Makefile index e85498b2f166..906a18b4e74a 100644 --- a/net/bridge/Makefile +++ b/net/bridge/Makefile @@ -16,4 +16,4 @@ bridge-$(CONFIG_BRIDGE_IGMP_SNOOPING) += br_multicast.o br_mdb.o bridge-$(CONFIG_BRIDGE_VLAN_FILTERING) += br_vlan.o -obj-$(CONFIG_BRIDGE_NF_EBTABLES) += netfilter/ +obj-$(CONFIG_BRIDGE_NETFILTER) += netfilter/ diff --git a/net/bridge/netfilter/Kconfig b/net/bridge/netfilter/Kconfig index 5ca74a0e595f..3baf29d34e62 100644 --- a/net/bridge/netfilter/Kconfig +++ b/net/bridge/netfilter/Kconfig @@ -2,13 +2,25 @@ # Bridge netfilter configuration # # -config NF_TABLES_BRIDGE +menuconfig NF_TABLES_BRIDGE depends on NF_TABLES + select BRIDGE_NETFILTER tristate "Ethernet Bridge nf_tables support" +if NF_TABLES_BRIDGE + +config NFT_BRIDGE_META + tristate "Netfilter nf_table bridge meta support" + depends on NFT_META + help + Add support for bridge dedicated meta key. + +endif # NF_TABLES_BRIDGE + menuconfig BRIDGE_NF_EBTABLES tristate "Ethernet Bridge tables (ebtables) support" depends on BRIDGE && NETFILTER + select BRIDGE_NETFILTER select NETFILTER_XTABLES help ebtables is a general, extensible frame/packet identification diff --git a/net/bridge/netfilter/Makefile b/net/bridge/netfilter/Makefile index ea7629f58b3d..6f2f3943d66f 100644 --- a/net/bridge/netfilter/Makefile +++ b/net/bridge/netfilter/Makefile @@ -3,6 +3,7 @@ # obj-$(CONFIG_NF_TABLES_BRIDGE) += nf_tables_bridge.o +obj-$(CONFIG_NFT_BRIDGE_META) += nft_meta_bridge.o obj-$(CONFIG_BRIDGE_NF_EBTABLES) += ebtables.o diff --git a/net/bridge/netfilter/nft_meta_bridge.c b/net/bridge/netfilter/nft_meta_bridge.c new file mode 100644 index 000000000000..4f02109d708f --- /dev/null +++ b/net/bridge/netfilter/nft_meta_bridge.c @@ -0,0 +1,139 @@ +/* + * Copyright (c) 2014 Intel Corporation + * + * 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 "../br_private.h" + +static void nft_meta_bridge_get_eval(const struct nft_expr *expr, + struct nft_data data[NFT_REG_MAX + 1], + const struct nft_pktinfo *pkt) +{ + const struct nft_meta *priv = nft_expr_priv(expr); + const struct net_device *in = pkt->in, *out = pkt->out; + struct nft_data *dest = &data[priv->dreg]; + const struct net_bridge_port *p; + + switch (priv->key) { + case NFT_META_BRI_IIFNAME: + if (in == NULL || (p = br_port_get_rcu(in)) == NULL) + goto err; + break; + case NFT_META_BRI_OIFNAME: + if (out == NULL || (p = br_port_get_rcu(out)) == NULL) + goto err; + break; + default: + goto out; + } + + strncpy((char *)dest->data, p->br->dev->name, sizeof(dest->data)); + return; +out: + return nft_meta_get_eval(expr, data, pkt); +err: + data[NFT_REG_VERDICT].verdict = NFT_BREAK; +} + +static int nft_meta_bridge_get_init(const struct nft_ctx *ctx, + const struct nft_expr *expr, + const struct nlattr * const tb[]) +{ + struct nft_meta *priv = nft_expr_priv(expr); + int err; + + priv->key = ntohl(nla_get_be32(tb[NFTA_META_KEY])); + switch (priv->key) { + case NFT_META_BRI_IIFNAME: + case NFT_META_BRI_OIFNAME: + break; + default: + return nft_meta_get_init(ctx, expr, tb); + } + + priv->dreg = ntohl(nla_get_be32(tb[NFTA_META_DREG])); + err = nft_validate_output_register(priv->dreg); + if (err < 0) + return err; + + err = nft_validate_data_load(ctx, priv->dreg, NULL, NFT_DATA_VALUE); + if (err < 0) + return err; + + return 0; +} + +static struct nft_expr_type nft_meta_bridge_type; +static const struct nft_expr_ops nft_meta_bridge_get_ops = { + .type = &nft_meta_bridge_type, + .size = NFT_EXPR_SIZE(sizeof(struct nft_meta)), + .eval = nft_meta_bridge_get_eval, + .init = nft_meta_bridge_get_init, + .dump = nft_meta_get_dump, +}; + +static const struct nft_expr_ops nft_meta_bridge_set_ops = { + .type = &nft_meta_bridge_type, + .size = NFT_EXPR_SIZE(sizeof(struct nft_meta)), + .eval = nft_meta_set_eval, + .init = nft_meta_set_init, + .dump = nft_meta_set_dump, +}; + +static const struct nft_expr_ops * +nft_meta_bridge_select_ops(const struct nft_ctx *ctx, + const struct nlattr * const tb[]) +{ + if (tb[NFTA_META_KEY] == NULL) + return ERR_PTR(-EINVAL); + + if (tb[NFTA_META_DREG] && tb[NFTA_META_SREG]) + return ERR_PTR(-EINVAL); + + if (tb[NFTA_META_DREG]) + return &nft_meta_bridge_get_ops; + + if (tb[NFTA_META_SREG]) + return &nft_meta_bridge_set_ops; + + return ERR_PTR(-EINVAL); +} + +static struct nft_expr_type nft_meta_bridge_type __read_mostly = { + .family = NFPROTO_BRIDGE, + .name = "meta", + .select_ops = &nft_meta_bridge_select_ops, + .policy = nft_meta_policy, + .maxattr = NFTA_META_MAX, + .owner = THIS_MODULE, +}; + +static int __init nft_meta_bridge_module_init(void) +{ + return nft_register_expr(&nft_meta_bridge_type); +} + +static void __exit nft_meta_bridge_module_exit(void) +{ + nft_unregister_expr(&nft_meta_bridge_type); +} + +module_init(nft_meta_bridge_module_init); +module_exit(nft_meta_bridge_module_exit); + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Tomasz Bursztyka "); +MODULE_ALIAS_NFT_AF_EXPR(AF_BRIDGE, "meta"); From 4c1f7818e4002ceb0498ef48cb7efa7a9597d210 Mon Sep 17 00:00:00 2001 From: Pablo Neira Date: Mon, 31 Mar 2014 17:43:47 +0200 Subject: [PATCH 11/25] netfilter: nf_tables: relax string validation of NFTA_CHAIN_TYPE Use NLA_STRING for consistency with other string attributes in nf_tables. Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_tables_api.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 1d0c174d30fc..a5ca900912e1 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -541,7 +541,7 @@ static const struct nla_policy nft_chain_policy[NFTA_CHAIN_MAX + 1] = { .len = NFT_CHAIN_MAXNAMELEN - 1 }, [NFTA_CHAIN_HOOK] = { .type = NLA_NESTED }, [NFTA_CHAIN_POLICY] = { .type = NLA_U32 }, - [NFTA_CHAIN_TYPE] = { .type = NLA_NUL_STRING }, + [NFTA_CHAIN_TYPE] = { .type = NLA_STRING }, [NFTA_CHAIN_COUNTERS] = { .type = NLA_NESTED }, }; From 7c95f6d866d861268a217003c5202009fa76f252 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Fri, 4 Apr 2014 01:22:45 +0200 Subject: [PATCH 12/25] netfilter: nf_tables: deconstify table and chain in context structure The new transaction infrastructure updates the family, table and chain objects in the context structure, so let's deconstify them. While at it, move the context structure initialization routine to the top of the source file as it will be also used from the table and chain routines. Signed-off-by: Pablo Neira Ayuso --- include/net/netfilter/nf_tables.h | 6 ++-- net/netfilter/nf_tables_api.c | 58 +++++++++++++++---------------- 2 files changed, 32 insertions(+), 32 deletions(-) diff --git a/include/net/netfilter/nf_tables.h b/include/net/netfilter/nf_tables.h index 29ff1dc41ef3..91505231a105 100644 --- a/include/net/netfilter/nf_tables.h +++ b/include/net/netfilter/nf_tables.h @@ -83,9 +83,9 @@ struct nft_ctx { struct net *net; const struct sk_buff *skb; const struct nlmsghdr *nlh; - const struct nft_af_info *afi; - const struct nft_table *table; - const struct nft_chain *chain; + struct nft_af_info *afi; + struct nft_table *table; + struct nft_chain *chain; const struct nlattr * const *nla; }; diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index a5ca900912e1..3643bbc720bc 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -88,6 +88,23 @@ nf_tables_afinfo_lookup(struct net *net, int family, bool autoload) return ERR_PTR(-EAFNOSUPPORT); } +static void nft_ctx_init(struct nft_ctx *ctx, + const struct sk_buff *skb, + const struct nlmsghdr *nlh, + struct nft_af_info *afi, + struct nft_table *table, + struct nft_chain *chain, + const struct nlattr * const *nla) +{ + ctx->net = sock_net(skb->sk); + ctx->skb = skb; + ctx->nlh = nlh; + ctx->afi = afi; + ctx->table = table; + ctx->chain = chain; + ctx->nla = nla; +} + /* * Tables */ @@ -812,7 +829,7 @@ static int nf_tables_newchain(struct sock *nlsk, struct sk_buff *skb, { const struct nfgenmsg *nfmsg = nlmsg_data(nlh); const struct nlattr * uninitialized_var(name); - const struct nft_af_info *afi; + struct nft_af_info *afi; struct nft_table *table; struct nft_chain *chain; struct nft_base_chain *basechain = NULL; @@ -1024,7 +1041,7 @@ static int nf_tables_delchain(struct sock *nlsk, struct sk_buff *skb, const struct nlattr * const nla[]) { const struct nfgenmsg *nfmsg = nlmsg_data(nlh); - const struct nft_af_info *afi; + struct nft_af_info *afi; struct nft_table *table; struct nft_chain *chain; struct net *net = sock_net(skb->sk); @@ -1062,23 +1079,6 @@ static int nf_tables_delchain(struct sock *nlsk, struct sk_buff *skb, return 0; } -static void nft_ctx_init(struct nft_ctx *ctx, - const struct sk_buff *skb, - const struct nlmsghdr *nlh, - const struct nft_af_info *afi, - const struct nft_table *table, - const struct nft_chain *chain, - const struct nlattr * const *nla) -{ - ctx->net = sock_net(skb->sk); - ctx->skb = skb; - ctx->nlh = nlh; - ctx->afi = afi; - ctx->table = table; - ctx->chain = chain; - ctx->nla = nla; -} - /* * Expressions */ @@ -1582,7 +1582,7 @@ static int nf_tables_newrule(struct sock *nlsk, struct sk_buff *skb, const struct nlattr * const nla[]) { const struct nfgenmsg *nfmsg = nlmsg_data(nlh); - const struct nft_af_info *afi; + struct nft_af_info *afi; struct net *net = sock_net(skb->sk); struct nft_table *table; struct nft_chain *chain; @@ -1763,9 +1763,9 @@ static int nf_tables_delrule(struct sock *nlsk, struct sk_buff *skb, const struct nlattr * const nla[]) { const struct nfgenmsg *nfmsg = nlmsg_data(nlh); - const struct nft_af_info *afi; + struct nft_af_info *afi; struct net *net = sock_net(skb->sk); - const struct nft_table *table; + struct nft_table *table; struct nft_chain *chain = NULL; struct nft_rule *rule; int family = nfmsg->nfgen_family, err = 0; @@ -2009,8 +2009,8 @@ static int nft_ctx_init_from_setattr(struct nft_ctx *ctx, { struct net *net = sock_net(skb->sk); const struct nfgenmsg *nfmsg = nlmsg_data(nlh); - const struct nft_af_info *afi = NULL; - const struct nft_table *table = NULL; + struct nft_af_info *afi = NULL; + struct nft_table *table = NULL; if (nfmsg->nfgen_family != NFPROTO_UNSPEC) { afi = nf_tables_afinfo_lookup(net, nfmsg->nfgen_family, false); @@ -2244,7 +2244,7 @@ static int nf_tables_dump_sets_all(struct nft_ctx *ctx, struct sk_buff *skb, { const struct nft_set *set; unsigned int idx, s_idx = cb->args[0]; - const struct nft_af_info *afi; + struct nft_af_info *afi; struct nft_table *table, *cur_table = (struct nft_table *)cb->args[2]; struct net *net = sock_net(skb->sk); int cur_family = cb->args[3]; @@ -2389,7 +2389,7 @@ static int nf_tables_newset(struct sock *nlsk, struct sk_buff *skb, { const struct nfgenmsg *nfmsg = nlmsg_data(nlh); const struct nft_set_ops *ops; - const struct nft_af_info *afi; + struct nft_af_info *afi; struct net *net = sock_net(skb->sk); struct nft_table *table; struct nft_set *set; @@ -2651,8 +2651,8 @@ static int nft_ctx_init_from_elemattr(struct nft_ctx *ctx, const struct nlattr * const nla[]) { const struct nfgenmsg *nfmsg = nlmsg_data(nlh); - const struct nft_af_info *afi; - const struct nft_table *table; + struct nft_af_info *afi; + struct nft_table *table; struct net *net = sock_net(skb->sk); afi = nf_tables_afinfo_lookup(net, nfmsg->nfgen_family, false); @@ -2959,7 +2959,7 @@ static int nft_add_set_elem(const struct nft_ctx *ctx, struct nft_set *set, struct nft_ctx bind_ctx = { .afi = ctx->afi, .table = ctx->table, - .chain = binding->chain, + .chain = (struct nft_chain *)binding->chain, }; err = nft_validate_data_load(&bind_ctx, dreg, From 1081d11b086afb73e1d8f52f9047d661d8770b82 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Fri, 4 Apr 2014 01:24:07 +0200 Subject: [PATCH 13/25] netfilter: nf_tables: generalise transaction infrastructure This patch generalises the existing rule transaction infrastructure so it can be used to handle set, table and chain object transactions as well. The transaction provides a data area that stores private information depending on the transaction type. Signed-off-by: Pablo Neira Ayuso --- include/net/netfilter/nf_tables.h | 15 +++- net/netfilter/nf_tables_api.c | 127 +++++++++++++++++------------- 2 files changed, 82 insertions(+), 60 deletions(-) diff --git a/include/net/netfilter/nf_tables.h b/include/net/netfilter/nf_tables.h index 91505231a105..246dbd48825f 100644 --- a/include/net/netfilter/nf_tables.h +++ b/include/net/netfilter/nf_tables.h @@ -387,18 +387,25 @@ struct nft_rule { }; /** - * struct nft_rule_trans - nf_tables rule update in transaction + * struct nft_trans - nf_tables object update in transaction * * @list: used internally - * @ctx: rule context - * @rule: rule that needs to be updated + * @ctx: transaction context + * @data: internal information related to the transaction */ -struct nft_rule_trans { +struct nft_trans { struct list_head list; struct nft_ctx ctx; + char data[0]; +}; + +struct nft_trans_rule { struct nft_rule *rule; }; +#define nft_trans_rule(trans) \ + (((struct nft_trans_rule *)trans->data)->rule) + static inline struct nft_expr *nft_expr_first(const struct nft_rule *rule) { return (struct nft_expr *)&rule->data[0]; diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 3643bbc720bc..424a6f3cb6c5 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -105,6 +105,25 @@ static void nft_ctx_init(struct nft_ctx *ctx, ctx->nla = nla; } +static struct nft_trans *nft_trans_alloc(struct nft_ctx *ctx, u32 size) +{ + struct nft_trans *trans; + + trans = kzalloc(sizeof(struct nft_trans) + size, GFP_KERNEL); + if (trans == NULL) + return NULL; + + trans->ctx = *ctx; + + return trans; +} + +static void nft_trans_destroy(struct nft_trans *trans) +{ + list_del(&trans->list); + kfree(trans); +} + /* * Tables */ @@ -1557,26 +1576,25 @@ static void nf_tables_rule_destroy(const struct nft_ctx *ctx, kfree(rule); } +static struct nft_trans *nft_trans_rule_add(struct nft_ctx *ctx, + struct nft_rule *rule) +{ + struct nft_trans *trans; + + trans = nft_trans_alloc(ctx, sizeof(struct nft_trans_rule)); + if (trans == NULL) + return NULL; + + nft_trans_rule(trans) = rule; + list_add_tail(&trans->list, &ctx->net->nft.commit_list); + + return trans; +} + #define NFT_RULE_MAXEXPRS 128 static struct nft_expr_info *info; -static struct nft_rule_trans * -nf_tables_trans_add(struct nft_ctx *ctx, struct nft_rule *rule) -{ - struct nft_rule_trans *rupd; - - rupd = kmalloc(sizeof(struct nft_rule_trans), GFP_KERNEL); - if (rupd == NULL) - return NULL; - - rupd->ctx = *ctx; - rupd->rule = rule; - list_add_tail(&rupd->list, &ctx->net->nft.commit_list); - - return rupd; -} - static int nf_tables_newrule(struct sock *nlsk, struct sk_buff *skb, const struct nlmsghdr *nlh, const struct nlattr * const nla[]) @@ -1587,7 +1605,7 @@ static int nf_tables_newrule(struct sock *nlsk, struct sk_buff *skb, struct nft_table *table; struct nft_chain *chain; struct nft_rule *rule, *old_rule = NULL; - struct nft_rule_trans *repl = NULL; + struct nft_trans *trans = NULL; struct nft_expr *expr; struct nft_ctx ctx; struct nlattr *tmp; @@ -1685,8 +1703,8 @@ static int nf_tables_newrule(struct sock *nlsk, struct sk_buff *skb, if (nlh->nlmsg_flags & NLM_F_REPLACE) { if (nft_rule_is_active_next(net, old_rule)) { - repl = nf_tables_trans_add(&ctx, old_rule); - if (repl == NULL) { + trans = nft_trans_rule_add(&ctx, old_rule); + if (trans == NULL) { err = -ENOMEM; goto err2; } @@ -1708,7 +1726,7 @@ static int nf_tables_newrule(struct sock *nlsk, struct sk_buff *skb, list_add_rcu(&rule->list, &chain->rules); } - if (nf_tables_trans_add(&ctx, rule) == NULL) { + if (nft_trans_rule_add(&ctx, rule) == NULL) { err = -ENOMEM; goto err3; } @@ -1716,11 +1734,10 @@ static int nf_tables_newrule(struct sock *nlsk, struct sk_buff *skb, err3: list_del_rcu(&rule->list); - if (repl) { - list_del_rcu(&repl->rule->list); - list_del(&repl->list); - nft_rule_clear(net, repl->rule); - kfree(repl); + if (trans) { + list_del_rcu(&nft_trans_rule(trans)->list); + nft_rule_clear(net, nft_trans_rule(trans)); + nft_trans_destroy(trans); } err2: nf_tables_rule_destroy(&ctx, rule); @@ -1737,7 +1754,7 @@ nf_tables_delrule_one(struct nft_ctx *ctx, struct nft_rule *rule) { /* You cannot delete the same rule twice */ if (nft_rule_is_active_next(ctx->net, rule)) { - if (nf_tables_trans_add(ctx, rule) == NULL) + if (nft_trans_rule_add(ctx, rule) == NULL) return -ENOMEM; nft_rule_disactivate_next(ctx->net, rule); return 0; @@ -1813,7 +1830,7 @@ static int nf_tables_delrule(struct sock *nlsk, struct sk_buff *skb, static int nf_tables_commit(struct sk_buff *skb) { struct net *net = sock_net(skb->sk); - struct nft_rule_trans *rupd, *tmp; + struct nft_trans *trans, *next; /* Bump generation counter, invalidate any dump in progress */ net->nft.genctr++; @@ -1826,38 +1843,38 @@ static int nf_tables_commit(struct sk_buff *skb) */ synchronize_rcu(); - list_for_each_entry_safe(rupd, tmp, &net->nft.commit_list, list) { + list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { /* This rule was inactive in the past and just became active. * Clear the next bit of the genmask since its meaning has * changed, now it is the future. */ - if (nft_rule_is_active(net, rupd->rule)) { - nft_rule_clear(net, rupd->rule); - nf_tables_rule_notify(skb, rupd->ctx.nlh, - rupd->ctx.table, rupd->ctx.chain, - rupd->rule, NFT_MSG_NEWRULE, 0, - rupd->ctx.afi->family); - list_del(&rupd->list); - kfree(rupd); + if (nft_rule_is_active(net, nft_trans_rule(trans))) { + nft_rule_clear(net, nft_trans_rule(trans)); + nf_tables_rule_notify(skb, trans->ctx.nlh, + trans->ctx.table, + trans->ctx.chain, + nft_trans_rule(trans), + NFT_MSG_NEWRULE, 0, + trans->ctx.afi->family); + nft_trans_destroy(trans); continue; } /* This rule is in the past, get rid of it */ - list_del_rcu(&rupd->rule->list); - nf_tables_rule_notify(skb, rupd->ctx.nlh, - rupd->ctx.table, rupd->ctx.chain, - rupd->rule, NFT_MSG_DELRULE, 0, - rupd->ctx.afi->family); + list_del_rcu(&nft_trans_rule(trans)->list); + nf_tables_rule_notify(skb, trans->ctx.nlh, + trans->ctx.table, trans->ctx.chain, + nft_trans_rule(trans), NFT_MSG_DELRULE, + 0, trans->ctx.afi->family); } /* Make sure we don't see any packet traversing old rules */ synchronize_rcu(); /* Now we can safely release unused old rules */ - list_for_each_entry_safe(rupd, tmp, &net->nft.commit_list, list) { - nf_tables_rule_destroy(&rupd->ctx, rupd->rule); - list_del(&rupd->list); - kfree(rupd); + list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { + nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); + nft_trans_destroy(trans); } return 0; @@ -1866,27 +1883,25 @@ static int nf_tables_commit(struct sk_buff *skb) static int nf_tables_abort(struct sk_buff *skb) { struct net *net = sock_net(skb->sk); - struct nft_rule_trans *rupd, *tmp; + struct nft_trans *trans, *next; - list_for_each_entry_safe(rupd, tmp, &net->nft.commit_list, list) { - if (!nft_rule_is_active_next(net, rupd->rule)) { - nft_rule_clear(net, rupd->rule); - list_del(&rupd->list); - kfree(rupd); + list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { + if (!nft_rule_is_active_next(net, nft_trans_rule(trans))) { + nft_rule_clear(net, nft_trans_rule(trans)); + nft_trans_destroy(trans); continue; } /* This rule is inactive, get rid of it */ - list_del_rcu(&rupd->rule->list); + list_del_rcu(&nft_trans_rule(trans)->list); } /* Make sure we don't see any packet accessing aborted rules */ synchronize_rcu(); - list_for_each_entry_safe(rupd, tmp, &net->nft.commit_list, list) { - nf_tables_rule_destroy(&rupd->ctx, rupd->rule); - list_del(&rupd->list); - kfree(rupd); + list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { + nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); + nft_trans_destroy(trans); } return 0; From 37082f930bb59ef6fd45e7bae6c45858af2cd972 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Thu, 3 Apr 2014 11:56:37 +0200 Subject: [PATCH 14/25] netfilter: nf_tables: relocate commit and abort routines in the source file Move the commit and abort routines to the bottom of the source code file. This change is required by the follow up patches that add the set, chain and table transaction support. This patch is just a cleanup to access several functions without having to declare their prototypes. Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_tables_api.c | 160 +++++++++++++++++----------------- 1 file changed, 80 insertions(+), 80 deletions(-) diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 424a6f3cb6c5..17df4b807f84 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -1827,86 +1827,6 @@ static int nf_tables_delrule(struct sock *nlsk, struct sk_buff *skb, return err; } -static int nf_tables_commit(struct sk_buff *skb) -{ - struct net *net = sock_net(skb->sk); - struct nft_trans *trans, *next; - - /* Bump generation counter, invalidate any dump in progress */ - net->nft.genctr++; - - /* A new generation has just started */ - net->nft.gencursor = gencursor_next(net); - - /* Make sure all packets have left the previous generation before - * purging old rules. - */ - synchronize_rcu(); - - list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { - /* This rule was inactive in the past and just became active. - * Clear the next bit of the genmask since its meaning has - * changed, now it is the future. - */ - if (nft_rule_is_active(net, nft_trans_rule(trans))) { - nft_rule_clear(net, nft_trans_rule(trans)); - nf_tables_rule_notify(skb, trans->ctx.nlh, - trans->ctx.table, - trans->ctx.chain, - nft_trans_rule(trans), - NFT_MSG_NEWRULE, 0, - trans->ctx.afi->family); - nft_trans_destroy(trans); - continue; - } - - /* This rule is in the past, get rid of it */ - list_del_rcu(&nft_trans_rule(trans)->list); - nf_tables_rule_notify(skb, trans->ctx.nlh, - trans->ctx.table, trans->ctx.chain, - nft_trans_rule(trans), NFT_MSG_DELRULE, - 0, trans->ctx.afi->family); - } - - /* Make sure we don't see any packet traversing old rules */ - synchronize_rcu(); - - /* Now we can safely release unused old rules */ - list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { - nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); - nft_trans_destroy(trans); - } - - return 0; -} - -static int nf_tables_abort(struct sk_buff *skb) -{ - struct net *net = sock_net(skb->sk); - struct nft_trans *trans, *next; - - list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { - if (!nft_rule_is_active_next(net, nft_trans_rule(trans))) { - nft_rule_clear(net, nft_trans_rule(trans)); - nft_trans_destroy(trans); - continue; - } - - /* This rule is inactive, get rid of it */ - list_del_rcu(&nft_trans_rule(trans)->list); - } - - /* Make sure we don't see any packet accessing aborted rules */ - synchronize_rcu(); - - list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { - nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); - nft_trans_destroy(trans); - } - - return 0; -} - /* * Sets */ @@ -3177,6 +3097,86 @@ static const struct nfnl_callback nf_tables_cb[NFT_MSG_MAX] = { }, }; +static int nf_tables_commit(struct sk_buff *skb) +{ + struct net *net = sock_net(skb->sk); + struct nft_trans *trans, *next; + + /* Bump generation counter, invalidate any dump in progress */ + net->nft.genctr++; + + /* A new generation has just started */ + net->nft.gencursor = gencursor_next(net); + + /* Make sure all packets have left the previous generation before + * purging old rules. + */ + synchronize_rcu(); + + list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { + /* This rule was inactive in the past and just became active. + * Clear the next bit of the genmask since its meaning has + * changed, now it is the future. + */ + if (nft_rule_is_active(net, nft_trans_rule(trans))) { + nft_rule_clear(net, nft_trans_rule(trans)); + nf_tables_rule_notify(skb, trans->ctx.nlh, + trans->ctx.table, + trans->ctx.chain, + nft_trans_rule(trans), + NFT_MSG_NEWRULE, 0, + trans->ctx.afi->family); + nft_trans_destroy(trans); + continue; + } + + /* This rule is in the past, get rid of it */ + list_del_rcu(&nft_trans_rule(trans)->list); + nf_tables_rule_notify(skb, trans->ctx.nlh, + trans->ctx.table, trans->ctx.chain, + nft_trans_rule(trans), NFT_MSG_DELRULE, + 0, trans->ctx.afi->family); + } + + /* Make sure we don't see any packet traversing old rules */ + synchronize_rcu(); + + /* Now we can safely release unused old rules */ + list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { + nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); + nft_trans_destroy(trans); + } + + return 0; +} + +static int nf_tables_abort(struct sk_buff *skb) +{ + struct net *net = sock_net(skb->sk); + struct nft_trans *trans, *next; + + list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { + if (!nft_rule_is_active_next(net, nft_trans_rule(trans))) { + nft_rule_clear(net, nft_trans_rule(trans)); + nft_trans_destroy(trans); + continue; + } + + /* This rule is inactive, get rid of it */ + list_del_rcu(&nft_trans_rule(trans)->list); + } + + /* Make sure we don't see any packet accessing aborted rules */ + synchronize_rcu(); + + list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { + nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); + nft_trans_destroy(trans); + } + + return 0; +} + static const struct nfnetlink_subsystem nf_tables_subsys = { .name = "nf_tables", .subsys_id = NFNL_SUBSYS_NFTABLES, From b380e5c733b9f18a6a3ebb97963b6dd037339bc0 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Fri, 4 Apr 2014 01:38:51 +0200 Subject: [PATCH 15/25] netfilter: nf_tables: add message type to transactions The patch adds message type to the transaction to simplify the commit the and abort routines. Yet another step forward in the generalisation of the transaction infrastructure. Signed-off-by: Pablo Neira Ayuso --- include/net/netfilter/nf_tables.h | 2 + net/netfilter/nf_tables_api.c | 74 ++++++++++++++++++------------- 2 files changed, 45 insertions(+), 31 deletions(-) diff --git a/include/net/netfilter/nf_tables.h b/include/net/netfilter/nf_tables.h index 246dbd48825f..d8dfb2695e0f 100644 --- a/include/net/netfilter/nf_tables.h +++ b/include/net/netfilter/nf_tables.h @@ -390,11 +390,13 @@ struct nft_rule { * struct nft_trans - nf_tables object update in transaction * * @list: used internally + * @msg_type: message type * @ctx: transaction context * @data: internal information related to the transaction */ struct nft_trans { struct list_head list; + int msg_type; struct nft_ctx ctx; char data[0]; }; diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 17df4b807f84..34aa3d507542 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -105,7 +105,8 @@ static void nft_ctx_init(struct nft_ctx *ctx, ctx->nla = nla; } -static struct nft_trans *nft_trans_alloc(struct nft_ctx *ctx, u32 size) +static struct nft_trans *nft_trans_alloc(struct nft_ctx *ctx, int msg_type, + u32 size) { struct nft_trans *trans; @@ -113,6 +114,7 @@ static struct nft_trans *nft_trans_alloc(struct nft_ctx *ctx, u32 size) if (trans == NULL) return NULL; + trans->msg_type = msg_type; trans->ctx = *ctx; return trans; @@ -1576,12 +1578,12 @@ static void nf_tables_rule_destroy(const struct nft_ctx *ctx, kfree(rule); } -static struct nft_trans *nft_trans_rule_add(struct nft_ctx *ctx, +static struct nft_trans *nft_trans_rule_add(struct nft_ctx *ctx, int msg_type, struct nft_rule *rule) { struct nft_trans *trans; - trans = nft_trans_alloc(ctx, sizeof(struct nft_trans_rule)); + trans = nft_trans_alloc(ctx, msg_type, sizeof(struct nft_trans_rule)); if (trans == NULL) return NULL; @@ -1703,7 +1705,8 @@ static int nf_tables_newrule(struct sock *nlsk, struct sk_buff *skb, if (nlh->nlmsg_flags & NLM_F_REPLACE) { if (nft_rule_is_active_next(net, old_rule)) { - trans = nft_trans_rule_add(&ctx, old_rule); + trans = nft_trans_rule_add(&ctx, NFT_MSG_NEWRULE, + old_rule); if (trans == NULL) { err = -ENOMEM; goto err2; @@ -1726,7 +1729,7 @@ static int nf_tables_newrule(struct sock *nlsk, struct sk_buff *skb, list_add_rcu(&rule->list, &chain->rules); } - if (nft_trans_rule_add(&ctx, rule) == NULL) { + if (nft_trans_rule_add(&ctx, NFT_MSG_NEWRULE, rule) == NULL) { err = -ENOMEM; goto err3; } @@ -1754,7 +1757,7 @@ nf_tables_delrule_one(struct nft_ctx *ctx, struct nft_rule *rule) { /* You cannot delete the same rule twice */ if (nft_rule_is_active_next(ctx->net, rule)) { - if (nft_trans_rule_add(ctx, rule) == NULL) + if (nft_trans_rule_add(ctx, NFT_MSG_DELRULE, rule) == NULL) return -ENOMEM; nft_rule_disactivate_next(ctx->net, rule); return 0; @@ -3114,28 +3117,26 @@ static int nf_tables_commit(struct sk_buff *skb) synchronize_rcu(); list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { - /* This rule was inactive in the past and just became active. - * Clear the next bit of the genmask since its meaning has - * changed, now it is the future. - */ - if (nft_rule_is_active(net, nft_trans_rule(trans))) { - nft_rule_clear(net, nft_trans_rule(trans)); - nf_tables_rule_notify(skb, trans->ctx.nlh, + switch (trans->msg_type) { + case NFT_MSG_NEWRULE: + nft_rule_clear(trans->ctx.net, nft_trans_rule(trans)); + nf_tables_rule_notify(trans->ctx.skb, trans->ctx.nlh, trans->ctx.table, trans->ctx.chain, nft_trans_rule(trans), NFT_MSG_NEWRULE, 0, trans->ctx.afi->family); nft_trans_destroy(trans); - continue; + break; + case NFT_MSG_DELRULE: + list_del_rcu(&nft_trans_rule(trans)->list); + nf_tables_rule_notify(trans->ctx.skb, trans->ctx.nlh, + trans->ctx.table, + trans->ctx.chain, + nft_trans_rule(trans), NFT_MSG_DELRULE, 0, + trans->ctx.afi->family); + break; } - - /* This rule is in the past, get rid of it */ - list_del_rcu(&nft_trans_rule(trans)->list); - nf_tables_rule_notify(skb, trans->ctx.nlh, - trans->ctx.table, trans->ctx.chain, - nft_trans_rule(trans), NFT_MSG_DELRULE, - 0, trans->ctx.afi->family); } /* Make sure we don't see any packet traversing old rules */ @@ -3143,8 +3144,13 @@ static int nf_tables_commit(struct sk_buff *skb) /* Now we can safely release unused old rules */ list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { - nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); - nft_trans_destroy(trans); + switch (trans->msg_type) { + case NFT_MSG_DELRULE: + nf_tables_rule_destroy(&trans->ctx, + nft_trans_rule(trans)); + nft_trans_destroy(trans); + break; + } } return 0; @@ -3156,22 +3162,28 @@ static int nf_tables_abort(struct sk_buff *skb) struct nft_trans *trans, *next; list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { - if (!nft_rule_is_active_next(net, nft_trans_rule(trans))) { - nft_rule_clear(net, nft_trans_rule(trans)); + switch (trans->msg_type) { + case NFT_MSG_NEWRULE: + list_del_rcu(&nft_trans_rule(trans)->list); + break; + case NFT_MSG_DELRULE: + nft_rule_clear(trans->ctx.net, nft_trans_rule(trans)); nft_trans_destroy(trans); - continue; + break; } - - /* This rule is inactive, get rid of it */ - list_del_rcu(&nft_trans_rule(trans)->list); } /* Make sure we don't see any packet accessing aborted rules */ synchronize_rcu(); list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { - nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); - nft_trans_destroy(trans); + switch (trans->msg_type) { + case NFT_MSG_NEWRULE: + nf_tables_rule_destroy(&trans->ctx, + nft_trans_rule(trans)); + nft_trans_destroy(trans); + break; + } } return 0; From 958bee14d0718ca7a5002c0f48a099d1d345812a Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Thu, 3 Apr 2014 11:48:44 +0200 Subject: [PATCH 16/25] netfilter: nf_tables: use new transaction infrastructure to handle sets This patch reworks the nf_tables API so set updates are included in the same batch that contains rule updates. This speeds up rule-set updates since we skip a dialog of four messages between kernel and user-space (two on each direction), from: 1) create the set and send netlink message to the kernel 2) process the response from the kernel that contains the allocated name. 3) add the set elements and send netlink message to the kernel. 4) process the response from the kernel (to check for errors). To: 1) add the set to the batch. 2) add the set elements to the batch. 3) add the rule that points to the set. 4) send batch to the kernel. This also introduces an internal set ID (NFTA_SET_ID) that is unique in the batch so set elements and rules can refer to new sets. Backward compatibility has been only retained in userspace, this means that new nft versions can talk to the kernel both in the new and the old fashion. Signed-off-by: Pablo Neira Ayuso --- include/net/netfilter/nf_tables.h | 12 +++ include/uapi/linux/netfilter/nf_tables.h | 6 ++ net/netfilter/nf_tables_api.c | 123 ++++++++++++++++++++--- net/netfilter/nft_lookup.c | 10 +- 4 files changed, 133 insertions(+), 18 deletions(-) diff --git a/include/net/netfilter/nf_tables.h b/include/net/netfilter/nf_tables.h index d8dfb2695e0f..0f472d668cbe 100644 --- a/include/net/netfilter/nf_tables.h +++ b/include/net/netfilter/nf_tables.h @@ -268,6 +268,8 @@ static inline void *nft_set_priv(const struct nft_set *set) struct nft_set *nf_tables_set_lookup(const struct nft_table *table, const struct nlattr *nla); +struct nft_set *nf_tables_set_lookup_byid(const struct net *net, + const struct nlattr *nla); /** * struct nft_set_binding - nf_tables set binding @@ -408,6 +410,16 @@ struct nft_trans_rule { #define nft_trans_rule(trans) \ (((struct nft_trans_rule *)trans->data)->rule) +struct nft_trans_set { + struct nft_set *set; + u32 set_id; +}; + +#define nft_trans_set(trans) \ + (((struct nft_trans_set *)trans->data)->set) +#define nft_trans_set_id(trans) \ + (((struct nft_trans_set *)trans->data)->set_id) + static inline struct nft_expr *nft_expr_first(const struct nft_rule *rule) { return (struct nft_expr *)&rule->data[0]; diff --git a/include/uapi/linux/netfilter/nf_tables.h b/include/uapi/linux/netfilter/nf_tables.h index 7d6433f66bf8..2a88f645a5d8 100644 --- a/include/uapi/linux/netfilter/nf_tables.h +++ b/include/uapi/linux/netfilter/nf_tables.h @@ -246,6 +246,7 @@ enum nft_set_desc_attributes { * @NFTA_SET_DATA_LEN: mapping data length (NLA_U32) * @NFTA_SET_POLICY: selection policy (NLA_U32) * @NFTA_SET_DESC: set description (NLA_NESTED) + * @NFTA_SET_ID: uniquely identifies a set in a transaction (NLA_U32) */ enum nft_set_attributes { NFTA_SET_UNSPEC, @@ -258,6 +259,7 @@ enum nft_set_attributes { NFTA_SET_DATA_LEN, NFTA_SET_POLICY, NFTA_SET_DESC, + NFTA_SET_ID, __NFTA_SET_MAX }; #define NFTA_SET_MAX (__NFTA_SET_MAX - 1) @@ -293,12 +295,14 @@ enum nft_set_elem_attributes { * @NFTA_SET_ELEM_LIST_TABLE: table of the set to be changed (NLA_STRING) * @NFTA_SET_ELEM_LIST_SET: name of the set to be changed (NLA_STRING) * @NFTA_SET_ELEM_LIST_ELEMENTS: list of set elements (NLA_NESTED: nft_set_elem_attributes) + * @NFTA_SET_ELEM_LIST_SET_ID: uniquely identifies a set in a transaction (NLA_U32) */ enum nft_set_elem_list_attributes { NFTA_SET_ELEM_LIST_UNSPEC, NFTA_SET_ELEM_LIST_TABLE, NFTA_SET_ELEM_LIST_SET, NFTA_SET_ELEM_LIST_ELEMENTS, + NFTA_SET_ELEM_LIST_SET_ID, __NFTA_SET_ELEM_LIST_MAX }; #define NFTA_SET_ELEM_LIST_MAX (__NFTA_SET_ELEM_LIST_MAX - 1) @@ -484,12 +488,14 @@ enum nft_cmp_attributes { * @NFTA_LOOKUP_SET: name of the set where to look for (NLA_STRING) * @NFTA_LOOKUP_SREG: source register of the data to look for (NLA_U32: nft_registers) * @NFTA_LOOKUP_DREG: destination register (NLA_U32: nft_registers) + * @NFTA_LOOKUP_SET_ID: uniquely identifies a set in a transaction (NLA_U32) */ enum nft_lookup_attributes { NFTA_LOOKUP_UNSPEC, NFTA_LOOKUP_SET, NFTA_LOOKUP_SREG, NFTA_LOOKUP_DREG, + NFTA_LOOKUP_SET_ID, __NFTA_LOOKUP_MAX }; #define NFTA_LOOKUP_MAX (__NFTA_LOOKUP_MAX - 1) diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 34aa3d507542..02d3cc65690c 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -1934,6 +1934,7 @@ static const struct nla_policy nft_set_policy[NFTA_SET_MAX + 1] = { [NFTA_SET_DATA_LEN] = { .type = NLA_U32 }, [NFTA_SET_POLICY] = { .type = NLA_U32 }, [NFTA_SET_DESC] = { .type = NLA_NESTED }, + [NFTA_SET_ID] = { .type = NLA_U32 }, }; static const struct nla_policy nft_set_desc_policy[NFTA_SET_DESC_MAX + 1] = { @@ -1984,6 +1985,20 @@ struct nft_set *nf_tables_set_lookup(const struct nft_table *table, return ERR_PTR(-ENOENT); } +struct nft_set *nf_tables_set_lookup_byid(const struct net *net, + const struct nlattr *nla) +{ + struct nft_trans *trans; + u32 id = ntohl(nla_get_be32(nla)); + + list_for_each_entry(trans, &net->nft.commit_list, list) { + if (trans->msg_type == NFT_MSG_NEWSET && + id == nft_trans_set_id(trans)) + return nft_trans_set(trans); + } + return ERR_PTR(-ENOENT); +} + static int nf_tables_set_alloc_name(struct nft_ctx *ctx, struct nft_set *set, const char *name) { @@ -2259,6 +2274,8 @@ static int nf_tables_dump_sets(struct sk_buff *skb, struct netlink_callback *cb) return ret; } +#define NFT_SET_INACTIVE (1 << 15) /* Internal set flag */ + static int nf_tables_getset(struct sock *nlsk, struct sk_buff *skb, const struct nlmsghdr *nlh, const struct nlattr * const nla[]) @@ -2288,6 +2305,8 @@ static int nf_tables_getset(struct sock *nlsk, struct sk_buff *skb, set = nf_tables_set_lookup(ctx.table, nla[NFTA_SET_NAME]); if (IS_ERR(set)) return PTR_ERR(set); + if (set->flags & NFT_SET_INACTIVE) + return -ENOENT; skb2 = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL); if (skb2 == NULL) @@ -2321,6 +2340,26 @@ static int nf_tables_set_desc_parse(const struct nft_ctx *ctx, return 0; } +static int nft_trans_set_add(struct nft_ctx *ctx, int msg_type, + struct nft_set *set) +{ + struct nft_trans *trans; + + trans = nft_trans_alloc(ctx, msg_type, sizeof(struct nft_trans_set)); + if (trans == NULL) + return -ENOMEM; + + if (msg_type == NFT_MSG_NEWSET && ctx->nla[NFTA_SET_ID] != NULL) { + nft_trans_set_id(trans) = + ntohl(nla_get_be32(ctx->nla[NFTA_SET_ID])); + set->flags |= NFT_SET_INACTIVE; + } + nft_trans_set(trans) = set; + list_add_tail(&trans->list, &ctx->net->nft.commit_list); + + return 0; +} + static int nf_tables_newset(struct sock *nlsk, struct sk_buff *skb, const struct nlmsghdr *nlh, const struct nlattr * const nla[]) @@ -2341,7 +2380,8 @@ static int nf_tables_newset(struct sock *nlsk, struct sk_buff *skb, if (nla[NFTA_SET_TABLE] == NULL || nla[NFTA_SET_NAME] == NULL || - nla[NFTA_SET_KEY_LEN] == NULL) + nla[NFTA_SET_KEY_LEN] == NULL || + nla[NFTA_SET_ID] == NULL) return -EINVAL; memset(&desc, 0, sizeof(desc)); @@ -2458,8 +2498,11 @@ static int nf_tables_newset(struct sock *nlsk, struct sk_buff *skb, if (err < 0) goto err2; + err = nft_trans_set_add(&ctx, NFT_MSG_NEWSET, set); + if (err < 0) + goto err2; + list_add_tail(&set->list, &table->sets); - nf_tables_set_notify(&ctx, set, NFT_MSG_NEWSET); return 0; err2: @@ -2469,14 +2512,18 @@ err1: return err; } +static void nft_set_destroy(struct nft_set *set) +{ + set->ops->destroy(set); + module_put(set->ops->owner); + kfree(set); +} + static void nf_tables_set_destroy(const struct nft_ctx *ctx, struct nft_set *set) { list_del(&set->list); nf_tables_set_notify(ctx, set, NFT_MSG_DELSET); - - set->ops->destroy(set); - module_put(set->ops->owner); - kfree(set); + nft_set_destroy(set); } static int nf_tables_delset(struct sock *nlsk, struct sk_buff *skb, @@ -2500,10 +2547,16 @@ static int nf_tables_delset(struct sock *nlsk, struct sk_buff *skb, set = nf_tables_set_lookup(ctx.table, nla[NFTA_SET_NAME]); if (IS_ERR(set)) return PTR_ERR(set); + if (set->flags & NFT_SET_INACTIVE) + return -ENOENT; if (!list_empty(&set->bindings)) return -EBUSY; - nf_tables_set_destroy(&ctx, set); + err = nft_trans_set_add(&ctx, NFT_MSG_DELSET, set); + if (err < 0) + return err; + + list_del(&set->list); return 0; } @@ -2563,7 +2616,8 @@ void nf_tables_unbind_set(const struct nft_ctx *ctx, struct nft_set *set, { list_del(&binding->list); - if (list_empty(&set->bindings) && set->flags & NFT_SET_ANONYMOUS) + if (list_empty(&set->bindings) && set->flags & NFT_SET_ANONYMOUS && + !(set->flags & NFT_SET_INACTIVE)) nf_tables_set_destroy(ctx, set); } @@ -2581,6 +2635,7 @@ static const struct nla_policy nft_set_elem_list_policy[NFTA_SET_ELEM_LIST_MAX + [NFTA_SET_ELEM_LIST_TABLE] = { .type = NLA_STRING }, [NFTA_SET_ELEM_LIST_SET] = { .type = NLA_STRING }, [NFTA_SET_ELEM_LIST_ELEMENTS] = { .type = NLA_NESTED }, + [NFTA_SET_ELEM_LIST_SET_ID] = { .type = NLA_U32 }, }; static int nft_ctx_init_from_elemattr(struct nft_ctx *ctx, @@ -2680,6 +2735,8 @@ static int nf_tables_dump_set(struct sk_buff *skb, struct netlink_callback *cb) set = nf_tables_set_lookup(ctx.table, nla[NFTA_SET_ELEM_LIST_SET]); if (IS_ERR(set)) return PTR_ERR(set); + if (set->flags & NFT_SET_INACTIVE) + return -ENOENT; event = NFT_MSG_NEWSETELEM; event |= NFNL_SUBSYS_NFTABLES << 8; @@ -2743,6 +2800,8 @@ static int nf_tables_getsetelem(struct sock *nlsk, struct sk_buff *skb, set = nf_tables_set_lookup(ctx.table, nla[NFTA_SET_ELEM_LIST_SET]); if (IS_ERR(set)) return PTR_ERR(set); + if (set->flags & NFT_SET_INACTIVE) + return -ENOENT; if (nlh->nlmsg_flags & NLM_F_DUMP) { struct netlink_dump_control c = { @@ -2928,6 +2987,7 @@ static int nf_tables_newsetelem(struct sock *nlsk, struct sk_buff *skb, const struct nlmsghdr *nlh, const struct nlattr * const nla[]) { + struct net *net = sock_net(skb->sk); const struct nlattr *attr; struct nft_set *set; struct nft_ctx ctx; @@ -2938,8 +2998,15 @@ static int nf_tables_newsetelem(struct sock *nlsk, struct sk_buff *skb, return err; set = nf_tables_set_lookup(ctx.table, nla[NFTA_SET_ELEM_LIST_SET]); - if (IS_ERR(set)) - return PTR_ERR(set); + if (IS_ERR(set)) { + if (nla[NFTA_SET_ELEM_LIST_SET_ID]) { + set = nf_tables_set_lookup_byid(net, + nla[NFTA_SET_ELEM_LIST_SET_ID]); + } + if (IS_ERR(set)) + return PTR_ERR(set); + } + if (!list_empty(&set->bindings) && set->flags & NFT_SET_CONSTANT) return -EBUSY; @@ -3069,7 +3136,7 @@ static const struct nfnl_callback nf_tables_cb[NFT_MSG_MAX] = { .policy = nft_rule_policy, }, [NFT_MSG_NEWSET] = { - .call = nf_tables_newset, + .call_batch = nf_tables_newset, .attr_count = NFTA_SET_MAX, .policy = nft_set_policy, }, @@ -3079,12 +3146,12 @@ static const struct nfnl_callback nf_tables_cb[NFT_MSG_MAX] = { .policy = nft_set_policy, }, [NFT_MSG_DELSET] = { - .call = nf_tables_delset, + .call_batch = nf_tables_delset, .attr_count = NFTA_SET_MAX, .policy = nft_set_policy, }, [NFT_MSG_NEWSETELEM] = { - .call = nf_tables_newsetelem, + .call_batch = nf_tables_newsetelem, .attr_count = NFTA_SET_ELEM_LIST_MAX, .policy = nft_set_elem_list_policy, }, @@ -3094,7 +3161,7 @@ static const struct nfnl_callback nf_tables_cb[NFT_MSG_MAX] = { .policy = nft_set_elem_list_policy, }, [NFT_MSG_DELSETELEM] = { - .call = nf_tables_delsetelem, + .call_batch = nf_tables_delsetelem, .attr_count = NFTA_SET_ELEM_LIST_MAX, .policy = nft_set_elem_list_policy, }, @@ -3136,6 +3203,16 @@ static int nf_tables_commit(struct sk_buff *skb) nft_trans_rule(trans), NFT_MSG_DELRULE, 0, trans->ctx.afi->family); break; + case NFT_MSG_NEWSET: + nft_trans_set(trans)->flags &= ~NFT_SET_INACTIVE; + nf_tables_set_notify(&trans->ctx, nft_trans_set(trans), + NFT_MSG_NEWSET); + nft_trans_destroy(trans); + break; + case NFT_MSG_DELSET: + nf_tables_set_notify(&trans->ctx, nft_trans_set(trans), + NFT_MSG_DELSET); + break; } } @@ -3148,9 +3225,12 @@ static int nf_tables_commit(struct sk_buff *skb) case NFT_MSG_DELRULE: nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); - nft_trans_destroy(trans); + break; + case NFT_MSG_DELSET: + nft_set_destroy(nft_trans_set(trans)); break; } + nft_trans_destroy(trans); } return 0; @@ -3170,6 +3250,14 @@ static int nf_tables_abort(struct sk_buff *skb) nft_rule_clear(trans->ctx.net, nft_trans_rule(trans)); nft_trans_destroy(trans); break; + case NFT_MSG_NEWSET: + list_del(&nft_trans_set(trans)->list); + break; + case NFT_MSG_DELSET: + list_add_tail(&nft_trans_set(trans)->list, + &trans->ctx.table->sets); + nft_trans_destroy(trans); + break; } } @@ -3181,9 +3269,12 @@ static int nf_tables_abort(struct sk_buff *skb) case NFT_MSG_NEWRULE: nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); - nft_trans_destroy(trans); + break; + case NFT_MSG_NEWSET: + nft_set_destroy(nft_trans_set(trans)); break; } + nft_trans_destroy(trans); } return 0; diff --git a/net/netfilter/nft_lookup.c b/net/netfilter/nft_lookup.c index 7fd2bea8aa23..6404a726d17b 100644 --- a/net/netfilter/nft_lookup.c +++ b/net/netfilter/nft_lookup.c @@ -56,8 +56,14 @@ static int nft_lookup_init(const struct nft_ctx *ctx, return -EINVAL; set = nf_tables_set_lookup(ctx->table, tb[NFTA_LOOKUP_SET]); - if (IS_ERR(set)) - return PTR_ERR(set); + if (IS_ERR(set)) { + if (tb[NFTA_LOOKUP_SET_ID]) { + set = nf_tables_set_lookup_byid(ctx->net, + tb[NFTA_LOOKUP_SET_ID]); + } + if (IS_ERR(set)) + return PTR_ERR(set); + } priv->sreg = ntohl(nla_get_be32(tb[NFTA_LOOKUP_SREG])); err = nft_validate_input_register(priv->sreg); From ff3cd7b3c9225017b546785add039e1b1a22dff7 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Wed, 9 Apr 2014 11:53:06 +0200 Subject: [PATCH 17/25] netfilter: nf_tables: refactor chain statistic routines Add new routines to encapsulate chain statistics allocation and replacement. Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_tables_api.c | 45 +++++++++++++++++++---------------- 1 file changed, 24 insertions(+), 21 deletions(-) diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 02d3cc65690c..92cec68c03ec 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -805,8 +805,7 @@ static const struct nla_policy nft_counter_policy[NFTA_COUNTER_MAX + 1] = { [NFTA_COUNTER_BYTES] = { .type = NLA_U64 }, }; -static int -nf_tables_counters(struct nft_base_chain *chain, const struct nlattr *attr) +static struct nft_stats __percpu *nft_stats_alloc(const struct nlattr *attr) { struct nlattr *tb[NFTA_COUNTER_MAX+1]; struct nft_stats __percpu *newstats; @@ -815,14 +814,14 @@ nf_tables_counters(struct nft_base_chain *chain, const struct nlattr *attr) err = nla_parse_nested(tb, NFTA_COUNTER_MAX, attr, nft_counter_policy); if (err < 0) - return err; + return ERR_PTR(err); if (!tb[NFTA_COUNTER_BYTES] || !tb[NFTA_COUNTER_PACKETS]) - return -EINVAL; + return ERR_PTR(-EINVAL); newstats = alloc_percpu(struct nft_stats); if (newstats == NULL) - return -ENOMEM; + return ERR_PTR(-ENOMEM); /* Restore old counters on this cpu, no problem. Per-cpu statistics * are not exposed to userspace. @@ -831,6 +830,12 @@ nf_tables_counters(struct nft_base_chain *chain, const struct nlattr *attr) stats->bytes = be64_to_cpu(nla_get_be64(tb[NFTA_COUNTER_BYTES])); stats->pkts = be64_to_cpu(nla_get_be64(tb[NFTA_COUNTER_PACKETS])); + return newstats; +} + +static void nft_chain_stats_replace(struct nft_base_chain *chain, + struct nft_stats __percpu *newstats) +{ if (chain->stats) { struct nft_stats __percpu *oldstats = nft_dereference(chain->stats); @@ -840,8 +845,6 @@ nf_tables_counters(struct nft_base_chain *chain, const struct nlattr *attr) free_percpu(oldstats); } else rcu_assign_pointer(chain->stats, newstats); - - return 0; } static int nf_tables_newchain(struct sock *nlsk, struct sk_buff *skb, @@ -860,6 +863,7 @@ static int nf_tables_newchain(struct sock *nlsk, struct sk_buff *skb, u8 policy = NF_ACCEPT; u64 handle = 0; unsigned int i; + struct nft_stats __percpu *stats; int err; bool create; @@ -920,10 +924,11 @@ static int nf_tables_newchain(struct sock *nlsk, struct sk_buff *skb, if (!(chain->flags & NFT_BASE_CHAIN)) return -EOPNOTSUPP; - err = nf_tables_counters(nft_base_chain(chain), - nla[NFTA_CHAIN_COUNTERS]); - if (err < 0) - return err; + stats = nft_stats_alloc(nla[NFTA_CHAIN_COUNTERS]); + if (IS_ERR(stats)) + return PTR_ERR(stats); + + nft_chain_stats_replace(nft_base_chain(chain), stats); } if (nla[NFTA_CHAIN_POLICY]) @@ -977,23 +982,21 @@ static int nf_tables_newchain(struct sock *nlsk, struct sk_buff *skb, return -ENOMEM; if (nla[NFTA_CHAIN_COUNTERS]) { - err = nf_tables_counters(basechain, - nla[NFTA_CHAIN_COUNTERS]); - if (err < 0) { + stats = nft_stats_alloc(nla[NFTA_CHAIN_COUNTERS]); + if (IS_ERR(stats)) { module_put(type->owner); kfree(basechain); - return err; + return PTR_ERR(stats); } + basechain->stats = stats; } else { - struct nft_stats __percpu *newstats; - - newstats = alloc_percpu(struct nft_stats); - if (newstats == NULL) { + stats = alloc_percpu(struct nft_stats); + if (IS_ERR(stats)) { module_put(type->owner); kfree(basechain); - return -ENOMEM; + return PTR_ERR(stats); } - rcu_assign_pointer(basechain->stats, newstats); + rcu_assign_pointer(basechain->stats, stats); } basechain->type = type; From 91c7b38dc9f0de4f7f444b796d14476bc12df7bc Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Wed, 9 Apr 2014 11:58:08 +0200 Subject: [PATCH 18/25] netfilter: nf_tables: use new transaction infrastructure to handle chain This patch speeds up rule-set updates and it also introduces a way to revert chain updates if the batch is aborted. The idea is to store the changes in the transaction to apply that in the commit step. Signed-off-by: Pablo Neira Ayuso --- include/net/netfilter/nf_tables.h | 17 +++ net/netfilter/nf_tables_api.c | 203 +++++++++++++++++++++++------- 2 files changed, 175 insertions(+), 45 deletions(-) diff --git a/include/net/netfilter/nf_tables.h b/include/net/netfilter/nf_tables.h index 0f472d668cbe..7b2361c559b5 100644 --- a/include/net/netfilter/nf_tables.h +++ b/include/net/netfilter/nf_tables.h @@ -420,6 +420,22 @@ struct nft_trans_set { #define nft_trans_set_id(trans) \ (((struct nft_trans_set *)trans->data)->set_id) +struct nft_trans_chain { + bool update; + char name[NFT_CHAIN_MAXNAMELEN]; + struct nft_stats __percpu *stats; + u8 policy; +}; + +#define nft_trans_chain_update(trans) \ + (((struct nft_trans_chain *)trans->data)->update) +#define nft_trans_chain_name(trans) \ + (((struct nft_trans_chain *)trans->data)->name) +#define nft_trans_chain_stats(trans) \ + (((struct nft_trans_chain *)trans->data)->stats) +#define nft_trans_chain_policy(trans) \ + (((struct nft_trans_chain *)trans->data)->policy) + static inline struct nft_expr *nft_expr_first(const struct nft_rule *rule) { return (struct nft_expr *)&rule->data[0]; @@ -452,6 +468,7 @@ static inline void *nft_userdata(const struct nft_rule *rule) enum nft_chain_flags { NFT_BASE_CHAIN = 0x1, + NFT_CHAIN_INACTIVE = 0x2, }; /** diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 92cec68c03ec..7a1149fe2100 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -782,6 +782,8 @@ static int nf_tables_getchain(struct sock *nlsk, struct sk_buff *skb, chain = nf_tables_chain_lookup(table, nla[NFTA_CHAIN_NAME]); if (IS_ERR(chain)) return PTR_ERR(chain); + if (chain->flags & NFT_CHAIN_INACTIVE) + return -ENOENT; skb2 = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL); if (!skb2) @@ -847,6 +849,34 @@ static void nft_chain_stats_replace(struct nft_base_chain *chain, rcu_assign_pointer(chain->stats, newstats); } +static int nft_trans_chain_add(struct nft_ctx *ctx, int msg_type) +{ + struct nft_trans *trans; + + trans = nft_trans_alloc(ctx, msg_type, sizeof(struct nft_trans_chain)); + if (trans == NULL) + return -ENOMEM; + + if (msg_type == NFT_MSG_NEWCHAIN) + ctx->chain->flags |= NFT_CHAIN_INACTIVE; + + list_add_tail(&trans->list, &ctx->net->nft.commit_list); + return 0; +} + +static void nf_tables_chain_destroy(struct nft_chain *chain) +{ + BUG_ON(chain->use > 0); + + if (chain->flags & NFT_BASE_CHAIN) { + module_put(nft_base_chain(chain)->type->owner); + free_percpu(nft_base_chain(chain)->stats); + kfree(nft_base_chain(chain)); + } else { + kfree(chain); + } +} + static int nf_tables_newchain(struct sock *nlsk, struct sk_buff *skb, const struct nlmsghdr *nlh, const struct nlattr * const nla[]) @@ -866,6 +896,7 @@ static int nf_tables_newchain(struct sock *nlsk, struct sk_buff *skb, struct nft_stats __percpu *stats; int err; bool create; + struct nft_ctx ctx; create = nlh->nlmsg_flags & NLM_F_CREATE ? true : false; @@ -911,6 +942,11 @@ static int nf_tables_newchain(struct sock *nlsk, struct sk_buff *skb, } if (chain != NULL) { + struct nft_stats *stats = NULL; + struct nft_trans *trans; + + if (chain->flags & NFT_CHAIN_INACTIVE) + return -ENOENT; if (nlh->nlmsg_flags & NLM_F_EXCL) return -EEXIST; if (nlh->nlmsg_flags & NLM_F_REPLACE) @@ -927,17 +963,28 @@ static int nf_tables_newchain(struct sock *nlsk, struct sk_buff *skb, stats = nft_stats_alloc(nla[NFTA_CHAIN_COUNTERS]); if (IS_ERR(stats)) return PTR_ERR(stats); - - nft_chain_stats_replace(nft_base_chain(chain), stats); } + nft_ctx_init(&ctx, skb, nlh, afi, table, chain, nla); + trans = nft_trans_alloc(&ctx, NFT_MSG_NEWCHAIN, + sizeof(struct nft_trans_chain)); + if (trans == NULL) + return -ENOMEM; + + nft_trans_chain_stats(trans) = stats; + nft_trans_chain_update(trans) = true; + if (nla[NFTA_CHAIN_POLICY]) - nft_base_chain(chain)->policy = policy; + nft_trans_chain_policy(trans) = policy; + else + nft_trans_chain_policy(trans) = -1; - if (nla[NFTA_CHAIN_HANDLE] && name) - nla_strlcpy(chain->name, name, NFT_CHAIN_MAXNAMELEN); - - goto notify; + if (nla[NFTA_CHAIN_HANDLE] && name) { + nla_strlcpy(nft_trans_chain_name(trans), name, + NFT_CHAIN_MAXNAMELEN); + } + list_add_tail(&trans->list, &net->nft.commit_list); + return 0; } if (table->use == UINT_MAX) @@ -1033,31 +1080,26 @@ static int nf_tables_newchain(struct sock *nlsk, struct sk_buff *skb, if (!(table->flags & NFT_TABLE_F_DORMANT) && chain->flags & NFT_BASE_CHAIN) { err = nf_register_hooks(nft_base_chain(chain)->ops, afi->nops); - if (err < 0) { - module_put(basechain->type->owner); - free_percpu(basechain->stats); - kfree(basechain); - return err; - } + if (err < 0) + goto err1; } + + nft_ctx_init(&ctx, skb, nlh, afi, table, chain, nla); + err = nft_trans_chain_add(&ctx, NFT_MSG_NEWCHAIN); + if (err < 0) + goto err2; + list_add_tail(&chain->list, &table->chains); - table->use++; -notify: - nf_tables_chain_notify(skb, nlh, table, chain, NFT_MSG_NEWCHAIN, - family); return 0; -} - -static void nf_tables_chain_destroy(struct nft_chain *chain) -{ - BUG_ON(chain->use > 0); - - if (chain->flags & NFT_BASE_CHAIN) { - module_put(nft_base_chain(chain)->type->owner); - free_percpu(nft_base_chain(chain)->stats); - kfree(nft_base_chain(chain)); - } else - kfree(chain); +err2: + if (!(table->flags & NFT_TABLE_F_DORMANT) && + chain->flags & NFT_BASE_CHAIN) { + nf_unregister_hooks(nft_base_chain(chain)->ops, + afi->nops); + } +err1: + nf_tables_chain_destroy(chain); + return err; } static int nf_tables_delchain(struct sock *nlsk, struct sk_buff *skb, @@ -1070,6 +1112,8 @@ static int nf_tables_delchain(struct sock *nlsk, struct sk_buff *skb, struct nft_chain *chain; struct net *net = sock_net(skb->sk); int family = nfmsg->nfgen_family; + struct nft_ctx ctx; + int err; afi = nf_tables_afinfo_lookup(net, family, false); if (IS_ERR(afi)) @@ -1082,24 +1126,17 @@ static int nf_tables_delchain(struct sock *nlsk, struct sk_buff *skb, chain = nf_tables_chain_lookup(table, nla[NFTA_CHAIN_NAME]); if (IS_ERR(chain)) return PTR_ERR(chain); - + if (chain->flags & NFT_CHAIN_INACTIVE) + return -ENOENT; if (!list_empty(&chain->rules) || chain->use > 0) return -EBUSY; + nft_ctx_init(&ctx, skb, nlh, afi, table, chain, nla); + err = nft_trans_chain_add(&ctx, NFT_MSG_DELCHAIN); + if (err < 0) + return err; + list_del(&chain->list); - table->use--; - - if (!(table->flags & NFT_TABLE_F_DORMANT) && - chain->flags & NFT_BASE_CHAIN) - nf_unregister_hooks(nft_base_chain(chain)->ops, afi->nops); - - nf_tables_chain_notify(skb, nlh, table, chain, NFT_MSG_DELCHAIN, - family); - - /* Make sure all rule references are gone before this is released */ - synchronize_rcu(); - - nf_tables_chain_destroy(chain); return 0; } @@ -1542,6 +1579,8 @@ static int nf_tables_getrule(struct sock *nlsk, struct sk_buff *skb, chain = nf_tables_chain_lookup(table, nla[NFTA_RULE_CHAIN]); if (IS_ERR(chain)) return PTR_ERR(chain); + if (chain->flags & NFT_CHAIN_INACTIVE) + return -ENOENT; rule = nf_tables_rule_lookup(chain, nla[NFTA_RULE_HANDLE]); if (IS_ERR(rule)) @@ -3109,7 +3148,7 @@ static const struct nfnl_callback nf_tables_cb[NFT_MSG_MAX] = { .policy = nft_table_policy, }, [NFT_MSG_NEWCHAIN] = { - .call = nf_tables_newchain, + .call_batch = nf_tables_newchain, .attr_count = NFTA_CHAIN_MAX, .policy = nft_chain_policy, }, @@ -3119,7 +3158,7 @@ static const struct nfnl_callback nf_tables_cb[NFT_MSG_MAX] = { .policy = nft_chain_policy, }, [NFT_MSG_DELCHAIN] = { - .call = nf_tables_delchain, + .call_batch = nf_tables_delchain, .attr_count = NFTA_CHAIN_MAX, .policy = nft_chain_policy, }, @@ -3170,6 +3209,27 @@ static const struct nfnl_callback nf_tables_cb[NFT_MSG_MAX] = { }, }; +static void nft_chain_commit_update(struct nft_trans *trans) +{ + struct nft_base_chain *basechain; + + if (nft_trans_chain_name(trans)[0]) + strcpy(trans->ctx.chain->name, nft_trans_chain_name(trans)); + + if (!(trans->ctx.chain->flags & NFT_BASE_CHAIN)) + return; + + basechain = nft_base_chain(trans->ctx.chain); + nft_chain_stats_replace(basechain, nft_trans_chain_stats(trans)); + + switch (nft_trans_chain_policy(trans)) { + case NF_DROP: + case NF_ACCEPT: + basechain->policy = nft_trans_chain_policy(trans); + break; + } +} + static int nf_tables_commit(struct sk_buff *skb) { struct net *net = sock_net(skb->sk); @@ -3188,6 +3248,33 @@ static int nf_tables_commit(struct sk_buff *skb) list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { switch (trans->msg_type) { + case NFT_MSG_NEWCHAIN: + if (nft_trans_chain_update(trans)) + nft_chain_commit_update(trans); + else { + trans->ctx.chain->flags &= ~NFT_CHAIN_INACTIVE; + trans->ctx.table->use++; + } + nf_tables_chain_notify(trans->ctx.skb, trans->ctx.nlh, + trans->ctx.table, + trans->ctx.chain, + NFT_MSG_NEWCHAIN, + trans->ctx.afi->family); + nft_trans_destroy(trans); + break; + case NFT_MSG_DELCHAIN: + trans->ctx.table->use--; + nf_tables_chain_notify(trans->ctx.skb, trans->ctx.nlh, + trans->ctx.table, + trans->ctx.chain, + NFT_MSG_DELCHAIN, + trans->ctx.afi->family); + if (!(trans->ctx.table->flags & NFT_TABLE_F_DORMANT) && + trans->ctx.chain->flags & NFT_BASE_CHAIN) { + nf_unregister_hooks(nft_base_chain(trans->ctx.chain)->ops, + trans->ctx.afi->nops); + } + break; case NFT_MSG_NEWRULE: nft_rule_clear(trans->ctx.net, nft_trans_rule(trans)); nf_tables_rule_notify(trans->ctx.skb, trans->ctx.nlh, @@ -3225,6 +3312,9 @@ static int nf_tables_commit(struct sk_buff *skb) /* Now we can safely release unused old rules */ list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { switch (trans->msg_type) { + case NFT_MSG_DELCHAIN: + nf_tables_chain_destroy(trans->ctx.chain); + break; case NFT_MSG_DELRULE: nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); @@ -3246,6 +3336,26 @@ static int nf_tables_abort(struct sk_buff *skb) list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { switch (trans->msg_type) { + case NFT_MSG_NEWCHAIN: + if (nft_trans_chain_update(trans)) { + if (nft_trans_chain_stats(trans)) + free_percpu(nft_trans_chain_stats(trans)); + + nft_trans_destroy(trans); + } else { + list_del(&trans->ctx.chain->list); + if (!(trans->ctx.table->flags & NFT_TABLE_F_DORMANT) && + trans->ctx.chain->flags & NFT_BASE_CHAIN) { + nf_unregister_hooks(nft_base_chain(trans->ctx.chain)->ops, + trans->ctx.afi->nops); + } + } + break; + case NFT_MSG_DELCHAIN: + list_add_tail(&trans->ctx.chain->list, + &trans->ctx.table->chains); + nft_trans_destroy(trans); + break; case NFT_MSG_NEWRULE: list_del_rcu(&nft_trans_rule(trans)->list); break; @@ -3269,6 +3379,9 @@ static int nf_tables_abort(struct sk_buff *skb) list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { switch (trans->msg_type) { + case NFT_MSG_NEWCHAIN: + nf_tables_chain_destroy(trans->ctx.chain); + break; case NFT_MSG_NEWRULE: nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); From f75edf5e9c9773cf6d0bd9b8d50ead41b4984569 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Sun, 30 Mar 2014 14:04:52 +0200 Subject: [PATCH 19/25] netfilter: nf_tables: disabling table hooks always succeeds nf_tables_table_disable() always succeeds, make this function void. Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_tables_api.c | 9 +++------ 1 file changed, 3 insertions(+), 6 deletions(-) diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 7a1149fe2100..5f717032b2a5 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -381,7 +381,7 @@ err: return err; } -static int nf_tables_table_disable(const struct nft_af_info *afi, +static void nf_tables_table_disable(const struct nft_af_info *afi, struct nft_table *table) { struct nft_chain *chain; @@ -391,8 +391,6 @@ static int nf_tables_table_disable(const struct nft_af_info *afi, nf_unregister_hooks(nft_base_chain(chain)->ops, afi->nops); } - - return 0; } static int nf_tables_updtable(struct sock *nlsk, struct sk_buff *skb, @@ -412,9 +410,8 @@ static int nf_tables_updtable(struct sock *nlsk, struct sk_buff *skb, if ((flags & NFT_TABLE_F_DORMANT) && !(table->flags & NFT_TABLE_F_DORMANT)) { - ret = nf_tables_table_disable(afi, table); - if (ret >= 0) - table->flags |= NFT_TABLE_F_DORMANT; + nf_tables_table_disable(afi, table); + table->flags |= NFT_TABLE_F_DORMANT; } else if (!(flags & NFT_TABLE_F_DORMANT) && table->flags & NFT_TABLE_F_DORMANT) { ret = nf_tables_table_enable(afi, table); From e1aaca93ee66de5b4c92c7c7f3c9722d90852729 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Sun, 30 Mar 2014 14:04:53 +0200 Subject: [PATCH 20/25] netfilter: nf_tables: pass context to nf_tables_updtable() So nf_tables_uptable() only takes one single parameter. Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_tables_api.c | 49 ++++++++++++++++++----------------- 1 file changed, 25 insertions(+), 24 deletions(-) diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 5f717032b2a5..d71125a0a341 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -393,36 +393,34 @@ static void nf_tables_table_disable(const struct nft_af_info *afi, } } -static int nf_tables_updtable(struct sock *nlsk, struct sk_buff *skb, - const struct nlmsghdr *nlh, - const struct nlattr * const nla[], - struct nft_af_info *afi, struct nft_table *table) +static int nf_tables_updtable(struct nft_ctx *ctx) { - const struct nfgenmsg *nfmsg = nlmsg_data(nlh); + const struct nfgenmsg *nfmsg = nlmsg_data(ctx->nlh); int family = nfmsg->nfgen_family, ret = 0; + u32 flags; - if (nla[NFTA_TABLE_FLAGS]) { - u32 flags; + if (!ctx->nla[NFTA_TABLE_FLAGS]) + return 0; - flags = ntohl(nla_get_be32(nla[NFTA_TABLE_FLAGS])); - if (flags & ~NFT_TABLE_F_DORMANT) - return -EINVAL; + flags = ntohl(nla_get_be32(ctx->nla[NFTA_TABLE_FLAGS])); + if (flags & ~NFT_TABLE_F_DORMANT) + return -EINVAL; - if ((flags & NFT_TABLE_F_DORMANT) && - !(table->flags & NFT_TABLE_F_DORMANT)) { - nf_tables_table_disable(afi, table); - table->flags |= NFT_TABLE_F_DORMANT; - } else if (!(flags & NFT_TABLE_F_DORMANT) && - table->flags & NFT_TABLE_F_DORMANT) { - ret = nf_tables_table_enable(afi, table); - if (ret >= 0) - table->flags &= ~NFT_TABLE_F_DORMANT; - } - if (ret < 0) - goto err; + if ((flags & NFT_TABLE_F_DORMANT) && + !(ctx->table->flags & NFT_TABLE_F_DORMANT)) { + nf_tables_table_disable(ctx->afi, ctx->table); + ctx->table->flags |= NFT_TABLE_F_DORMANT; + } else if (!(flags & NFT_TABLE_F_DORMANT) && + ctx->table->flags & NFT_TABLE_F_DORMANT) { + ret = nf_tables_table_enable(ctx->afi, ctx->table); + if (ret >= 0) + ctx->table->flags &= ~NFT_TABLE_F_DORMANT; } + if (ret < 0) + goto err; - nf_tables_table_notify(skb, nlh, table, NFT_MSG_NEWTABLE, family); + nf_tables_table_notify(ctx->skb, ctx->nlh, ctx->table, + NFT_MSG_NEWTABLE, family); err: return ret; } @@ -438,6 +436,7 @@ static int nf_tables_newtable(struct sock *nlsk, struct sk_buff *skb, struct net *net = sock_net(skb->sk); int family = nfmsg->nfgen_family; u32 flags = 0; + struct nft_ctx ctx; afi = nf_tables_afinfo_lookup(net, family, true); if (IS_ERR(afi)) @@ -456,7 +455,9 @@ static int nf_tables_newtable(struct sock *nlsk, struct sk_buff *skb, return -EEXIST; if (nlh->nlmsg_flags & NLM_F_REPLACE) return -EOPNOTSUPP; - return nf_tables_updtable(nlsk, skb, nlh, nla, afi, table); + + nft_ctx_init(&ctx, skb, nlh, afi, table, NULL, nla); + return nf_tables_updtable(&ctx); } if (nla[NFTA_TABLE_FLAGS]) { From 55dd6f93076bb82aa8911191125418dcfcbf2c9b Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Thu, 3 Apr 2014 11:53:37 +0200 Subject: [PATCH 21/25] netfilter: nf_tables: use new transaction infrastructure to handle table This patch speeds up rule-set updates and it also provides a way to revert updates and leave things in consistent state in case that the batch needs to be aborted. Signed-off-by: Pablo Neira Ayuso --- include/net/netfilter/nf_tables.h | 10 +++ net/netfilter/nf_tables_api.c | 145 ++++++++++++++++++++++++++---- 2 files changed, 136 insertions(+), 19 deletions(-) diff --git a/include/net/netfilter/nf_tables.h b/include/net/netfilter/nf_tables.h index 7b2361c559b5..15bf745f198d 100644 --- a/include/net/netfilter/nf_tables.h +++ b/include/net/netfilter/nf_tables.h @@ -436,6 +436,16 @@ struct nft_trans_chain { #define nft_trans_chain_policy(trans) \ (((struct nft_trans_chain *)trans->data)->policy) +struct nft_trans_table { + bool update; + bool enable; +}; + +#define nft_trans_table_update(trans) \ + (((struct nft_trans_table *)trans->data)->update) +#define nft_trans_table_enable(trans) \ + (((struct nft_trans_table *)trans->data)->enable) + static inline struct nft_expr *nft_expr_first(const struct nft_rule *rule) { return (struct nft_expr *)&rule->data[0]; diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index d71125a0a341..3e685dbb2921 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -307,6 +307,9 @@ done: return skb->len; } +/* Internal table flags */ +#define NFT_TABLE_INACTIVE (1 << 15) + static int nf_tables_gettable(struct sock *nlsk, struct sk_buff *skb, const struct nlmsghdr *nlh, const struct nlattr * const nla[]) @@ -333,6 +336,8 @@ static int nf_tables_gettable(struct sock *nlsk, struct sk_buff *skb, table = nf_tables_table_lookup(afi, nla[NFTA_TABLE_NAME]); if (IS_ERR(table)) return PTR_ERR(table); + if (table->flags & NFT_TABLE_INACTIVE) + return -ENOENT; skb2 = alloc_skb(NLMSG_GOODSIZE, GFP_KERNEL); if (!skb2) @@ -395,9 +400,9 @@ static void nf_tables_table_disable(const struct nft_af_info *afi, static int nf_tables_updtable(struct nft_ctx *ctx) { - const struct nfgenmsg *nfmsg = nlmsg_data(ctx->nlh); - int family = nfmsg->nfgen_family, ret = 0; + struct nft_trans *trans; u32 flags; + int ret = 0; if (!ctx->nla[NFTA_TABLE_FLAGS]) return 0; @@ -406,25 +411,48 @@ static int nf_tables_updtable(struct nft_ctx *ctx) if (flags & ~NFT_TABLE_F_DORMANT) return -EINVAL; + trans = nft_trans_alloc(ctx, NFT_MSG_NEWTABLE, + sizeof(struct nft_trans_table)); + if (trans == NULL) + return -ENOMEM; + if ((flags & NFT_TABLE_F_DORMANT) && !(ctx->table->flags & NFT_TABLE_F_DORMANT)) { - nf_tables_table_disable(ctx->afi, ctx->table); - ctx->table->flags |= NFT_TABLE_F_DORMANT; + nft_trans_table_enable(trans) = false; } else if (!(flags & NFT_TABLE_F_DORMANT) && ctx->table->flags & NFT_TABLE_F_DORMANT) { ret = nf_tables_table_enable(ctx->afi, ctx->table); - if (ret >= 0) + if (ret >= 0) { ctx->table->flags &= ~NFT_TABLE_F_DORMANT; + nft_trans_table_enable(trans) = true; + } } if (ret < 0) goto err; - nf_tables_table_notify(ctx->skb, ctx->nlh, ctx->table, - NFT_MSG_NEWTABLE, family); + nft_trans_table_update(trans) = true; + list_add_tail(&trans->list, &ctx->net->nft.commit_list); + return 0; err: + nft_trans_destroy(trans); return ret; } +static int nft_trans_table_add(struct nft_ctx *ctx, int msg_type) +{ + struct nft_trans *trans; + + trans = nft_trans_alloc(ctx, msg_type, sizeof(struct nft_trans_table)); + if (trans == NULL) + return -ENOMEM; + + if (msg_type == NFT_MSG_NEWTABLE) + ctx->table->flags |= NFT_TABLE_INACTIVE; + + list_add_tail(&trans->list, &ctx->net->nft.commit_list); + return 0; +} + static int nf_tables_newtable(struct sock *nlsk, struct sk_buff *skb, const struct nlmsghdr *nlh, const struct nlattr * const nla[]) @@ -437,6 +465,7 @@ static int nf_tables_newtable(struct sock *nlsk, struct sk_buff *skb, int family = nfmsg->nfgen_family; u32 flags = 0; struct nft_ctx ctx; + int err; afi = nf_tables_afinfo_lookup(net, family, true); if (IS_ERR(afi)) @@ -451,6 +480,8 @@ static int nf_tables_newtable(struct sock *nlsk, struct sk_buff *skb, } if (table != NULL) { + if (table->flags & NFT_TABLE_INACTIVE) + return -ENOENT; if (nlh->nlmsg_flags & NLM_F_EXCL) return -EEXIST; if (nlh->nlmsg_flags & NLM_F_REPLACE) @@ -480,8 +511,14 @@ static int nf_tables_newtable(struct sock *nlsk, struct sk_buff *skb, INIT_LIST_HEAD(&table->sets); table->flags = flags; + nft_ctx_init(&ctx, skb, nlh, afi, table, NULL, nla); + err = nft_trans_table_add(&ctx, NFT_MSG_NEWTABLE); + if (err < 0) { + kfree(table); + module_put(afi->owner); + return err; + } list_add_tail(&table->list, &afi->tables); - nf_tables_table_notify(skb, nlh, table, NFT_MSG_NEWTABLE, family); return 0; } @@ -493,7 +530,8 @@ static int nf_tables_deltable(struct sock *nlsk, struct sk_buff *skb, struct nft_af_info *afi; struct nft_table *table; struct net *net = sock_net(skb->sk); - int family = nfmsg->nfgen_family; + int family = nfmsg->nfgen_family, err; + struct nft_ctx ctx; afi = nf_tables_afinfo_lookup(net, family, false); if (IS_ERR(afi)) @@ -502,17 +540,27 @@ static int nf_tables_deltable(struct sock *nlsk, struct sk_buff *skb, table = nf_tables_table_lookup(afi, nla[NFTA_TABLE_NAME]); if (IS_ERR(table)) return PTR_ERR(table); + if (table->flags & NFT_TABLE_INACTIVE) + return -ENOENT; if (!list_empty(&table->chains) || !list_empty(&table->sets)) return -EBUSY; + nft_ctx_init(&ctx, skb, nlh, afi, table, NULL, nla); + err = nft_trans_table_add(&ctx, NFT_MSG_DELTABLE); + if (err < 0) + return err; + list_del(&table->list); - nf_tables_table_notify(skb, nlh, table, NFT_MSG_DELTABLE, family); - kfree(table); - module_put(afi->owner); return 0; } +static void nf_tables_table_destroy(struct nft_ctx *ctx) +{ + kfree(ctx->table); + module_put(ctx->afi->owner); +} + int nft_register_chain_type(const struct nf_chain_type *ctype) { int err = 0; @@ -776,6 +824,8 @@ static int nf_tables_getchain(struct sock *nlsk, struct sk_buff *skb, table = nf_tables_table_lookup(afi, nla[NFTA_CHAIN_TABLE]); if (IS_ERR(table)) return PTR_ERR(table); + if (table->flags & NFT_TABLE_INACTIVE) + return -ENOENT; chain = nf_tables_chain_lookup(table, nla[NFTA_CHAIN_NAME]); if (IS_ERR(chain)) @@ -1120,6 +1170,8 @@ static int nf_tables_delchain(struct sock *nlsk, struct sk_buff *skb, table = nf_tables_table_lookup(afi, nla[NFTA_CHAIN_TABLE]); if (IS_ERR(table)) return PTR_ERR(table); + if (table->flags & NFT_TABLE_INACTIVE) + return -ENOENT; chain = nf_tables_chain_lookup(table, nla[NFTA_CHAIN_NAME]); if (IS_ERR(chain)) @@ -1573,6 +1625,8 @@ static int nf_tables_getrule(struct sock *nlsk, struct sk_buff *skb, table = nf_tables_table_lookup(afi, nla[NFTA_RULE_TABLE]); if (IS_ERR(table)) return PTR_ERR(table); + if (table->flags & NFT_TABLE_INACTIVE) + return -ENOENT; chain = nf_tables_chain_lookup(table, nla[NFTA_RULE_CHAIN]); if (IS_ERR(chain)) @@ -1838,6 +1892,8 @@ static int nf_tables_delrule(struct sock *nlsk, struct sk_buff *skb, table = nf_tables_table_lookup(afi, nla[NFTA_RULE_TABLE]); if (IS_ERR(table)) return PTR_ERR(table); + if (table->flags & NFT_TABLE_INACTIVE) + return -ENOENT; if (nla[NFTA_RULE_CHAIN]) { chain = nf_tables_chain_lookup(table, nla[NFTA_RULE_CHAIN]); @@ -2004,6 +2060,8 @@ static int nft_ctx_init_from_setattr(struct nft_ctx *ctx, table = nf_tables_table_lookup(afi, nla[NFTA_SET_TABLE]); if (IS_ERR(table)) return PTR_ERR(table); + if (table->flags & NFT_TABLE_INACTIVE) + return -ENOENT; } nft_ctx_init(ctx, skb, nlh, afi, table, NULL, nla); @@ -2681,7 +2739,8 @@ static const struct nla_policy nft_set_elem_list_policy[NFTA_SET_ELEM_LIST_MAX + static int nft_ctx_init_from_elemattr(struct nft_ctx *ctx, const struct sk_buff *skb, const struct nlmsghdr *nlh, - const struct nlattr * const nla[]) + const struct nlattr * const nla[], + bool trans) { const struct nfgenmsg *nfmsg = nlmsg_data(nlh); struct nft_af_info *afi; @@ -2695,6 +2754,8 @@ static int nft_ctx_init_from_elemattr(struct nft_ctx *ctx, table = nf_tables_table_lookup(afi, nla[NFTA_SET_ELEM_LIST_TABLE]); if (IS_ERR(table)) return PTR_ERR(table); + if (!trans && (table->flags & NFT_TABLE_INACTIVE)) + return -ENOENT; nft_ctx_init(ctx, skb, nlh, afi, table, NULL, nla); return 0; @@ -2768,7 +2829,8 @@ static int nf_tables_dump_set(struct sk_buff *skb, struct netlink_callback *cb) if (err < 0) return err; - err = nft_ctx_init_from_elemattr(&ctx, cb->skb, cb->nlh, (void *)nla); + err = nft_ctx_init_from_elemattr(&ctx, cb->skb, cb->nlh, (void *)nla, + false); if (err < 0) return err; @@ -2833,7 +2895,7 @@ static int nf_tables_getsetelem(struct sock *nlsk, struct sk_buff *skb, struct nft_ctx ctx; int err; - err = nft_ctx_init_from_elemattr(&ctx, skb, nlh, nla); + err = nft_ctx_init_from_elemattr(&ctx, skb, nlh, nla, false); if (err < 0) return err; @@ -3033,7 +3095,7 @@ static int nf_tables_newsetelem(struct sock *nlsk, struct sk_buff *skb, struct nft_ctx ctx; int rem, err; - err = nft_ctx_init_from_elemattr(&ctx, skb, nlh, nla); + err = nft_ctx_init_from_elemattr(&ctx, skb, nlh, nla, true); if (err < 0) return err; @@ -3111,7 +3173,7 @@ static int nf_tables_delsetelem(struct sock *nlsk, struct sk_buff *skb, struct nft_ctx ctx; int rem, err; - err = nft_ctx_init_from_elemattr(&ctx, skb, nlh, nla); + err = nft_ctx_init_from_elemattr(&ctx, skb, nlh, nla, false); if (err < 0) return err; @@ -3131,7 +3193,7 @@ static int nf_tables_delsetelem(struct sock *nlsk, struct sk_buff *skb, static const struct nfnl_callback nf_tables_cb[NFT_MSG_MAX] = { [NFT_MSG_NEWTABLE] = { - .call = nf_tables_newtable, + .call_batch = nf_tables_newtable, .attr_count = NFTA_TABLE_MAX, .policy = nft_table_policy, }, @@ -3141,7 +3203,7 @@ static const struct nfnl_callback nf_tables_cb[NFT_MSG_MAX] = { .policy = nft_table_policy, }, [NFT_MSG_DELTABLE] = { - .call = nf_tables_deltable, + .call_batch = nf_tables_deltable, .attr_count = NFTA_TABLE_MAX, .policy = nft_table_policy, }, @@ -3246,6 +3308,28 @@ static int nf_tables_commit(struct sk_buff *skb) list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { switch (trans->msg_type) { + case NFT_MSG_NEWTABLE: + if (nft_trans_table_update(trans)) { + if (!nft_trans_table_enable(trans)) { + nf_tables_table_disable(trans->ctx.afi, + trans->ctx.table); + trans->ctx.table->flags |= NFT_TABLE_F_DORMANT; + } + } else { + trans->ctx.table->flags &= ~NFT_TABLE_INACTIVE; + } + nf_tables_table_notify(trans->ctx.skb, trans->ctx.nlh, + trans->ctx.table, + NFT_MSG_NEWTABLE, + trans->ctx.afi->family); + nft_trans_destroy(trans); + break; + case NFT_MSG_DELTABLE: + nf_tables_table_notify(trans->ctx.skb, trans->ctx.nlh, + trans->ctx.table, + NFT_MSG_DELTABLE, + trans->ctx.afi->family); + break; case NFT_MSG_NEWCHAIN: if (nft_trans_chain_update(trans)) nft_chain_commit_update(trans); @@ -3310,6 +3394,9 @@ static int nf_tables_commit(struct sk_buff *skb) /* Now we can safely release unused old rules */ list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { switch (trans->msg_type) { + case NFT_MSG_DELTABLE: + nf_tables_table_destroy(&trans->ctx); + break; case NFT_MSG_DELCHAIN: nf_tables_chain_destroy(trans->ctx.chain); break; @@ -3334,6 +3421,23 @@ static int nf_tables_abort(struct sk_buff *skb) list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { switch (trans->msg_type) { + case NFT_MSG_NEWTABLE: + if (nft_trans_table_update(trans)) { + if (nft_trans_table_enable(trans)) { + nf_tables_table_disable(trans->ctx.afi, + trans->ctx.table); + trans->ctx.table->flags |= NFT_TABLE_F_DORMANT; + } + nft_trans_destroy(trans); + } else { + list_del(&trans->ctx.table->list); + } + break; + case NFT_MSG_DELTABLE: + list_add_tail(&trans->ctx.table->list, + &trans->ctx.afi->tables); + nft_trans_destroy(trans); + break; case NFT_MSG_NEWCHAIN: if (nft_trans_chain_update(trans)) { if (nft_trans_chain_stats(trans)) @@ -3377,6 +3481,9 @@ static int nf_tables_abort(struct sk_buff *skb) list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { switch (trans->msg_type) { + case NFT_MSG_NEWTABLE: + nf_tables_table_destroy(&trans->ctx); + break; case NFT_MSG_NEWCHAIN: nf_tables_chain_destroy(trans->ctx.chain); break; From 60319eb1ca351aa36e29d58d2e60ba9a9836265a Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Fri, 4 Apr 2014 03:36:42 +0200 Subject: [PATCH 22/25] netfilter: nf_tables: use new transaction infrastructure to handle elements Leave the set content in consistent state if we fail to load the batch. Use the new generic transaction infrastructure to achieve this. Signed-off-by: Pablo Neira Ayuso --- include/net/netfilter/nf_tables.h | 10 ++++ net/netfilter/nf_tables_api.c | 82 +++++++++++++++++++++++++------ 2 files changed, 78 insertions(+), 14 deletions(-) diff --git a/include/net/netfilter/nf_tables.h b/include/net/netfilter/nf_tables.h index 15bf745f198d..b08f2a941007 100644 --- a/include/net/netfilter/nf_tables.h +++ b/include/net/netfilter/nf_tables.h @@ -446,6 +446,16 @@ struct nft_trans_table { #define nft_trans_table_enable(trans) \ (((struct nft_trans_table *)trans->data)->enable) +struct nft_trans_elem { + struct nft_set *set; + struct nft_set_elem elem; +}; + +#define nft_trans_elem_set(trans) \ + (((struct nft_trans_elem *)trans->data)->set) +#define nft_trans_elem(trans) \ + (((struct nft_trans_elem *)trans->data)->elem) + static inline struct nft_expr *nft_expr_first(const struct nft_rule *rule) { return (struct nft_expr *)&rule->data[0]; diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 3e685dbb2921..cd002935e6b2 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -2993,7 +2993,21 @@ err: return err; } -static int nft_add_set_elem(const struct nft_ctx *ctx, struct nft_set *set, +static struct nft_trans *nft_trans_elem_alloc(struct nft_ctx *ctx, + int msg_type, + struct nft_set *set) +{ + struct nft_trans *trans; + + trans = nft_trans_alloc(ctx, msg_type, sizeof(struct nft_trans_elem)); + if (trans == NULL) + return NULL; + + nft_trans_elem_set(trans) = set; + return trans; +} + +static int nft_add_set_elem(struct nft_ctx *ctx, struct nft_set *set, const struct nlattr *attr) { struct nlattr *nla[NFTA_SET_ELEM_MAX + 1]; @@ -3001,6 +3015,7 @@ static int nft_add_set_elem(const struct nft_ctx *ctx, struct nft_set *set, struct nft_set_elem elem; struct nft_set_binding *binding; enum nft_registers dreg; + struct nft_trans *trans; int err; if (set->size && set->nelems == set->size) @@ -3068,14 +3083,20 @@ static int nft_add_set_elem(const struct nft_ctx *ctx, struct nft_set *set, } } + trans = nft_trans_elem_alloc(ctx, NFT_MSG_NEWSETELEM, set); + if (trans == NULL) + goto err3; + err = set->ops->insert(set, &elem); if (err < 0) - goto err3; - set->nelems++; + goto err4; - nf_tables_setelem_notify(ctx, set, &elem, NFT_MSG_NEWSETELEM, 0); + nft_trans_elem(trans) = elem; + list_add(&trans->list, &ctx->net->nft.commit_list); return 0; +err4: + kfree(trans); err3: if (nla[NFTA_SET_ELEM_DATA] != NULL) nft_data_uninit(&elem.data, d2.type); @@ -3093,7 +3114,7 @@ static int nf_tables_newsetelem(struct sock *nlsk, struct sk_buff *skb, const struct nlattr *attr; struct nft_set *set; struct nft_ctx ctx; - int rem, err; + int rem, err = 0; err = nft_ctx_init_from_elemattr(&ctx, skb, nlh, nla, true); if (err < 0) @@ -3115,17 +3136,18 @@ static int nf_tables_newsetelem(struct sock *nlsk, struct sk_buff *skb, nla_for_each_nested(attr, nla[NFTA_SET_ELEM_LIST_ELEMENTS], rem) { err = nft_add_set_elem(&ctx, set, attr); if (err < 0) - return err; + break; } - return 0; + return err; } -static int nft_del_setelem(const struct nft_ctx *ctx, struct nft_set *set, +static int nft_del_setelem(struct nft_ctx *ctx, struct nft_set *set, const struct nlattr *attr) { struct nlattr *nla[NFTA_SET_ELEM_MAX + 1]; struct nft_data_desc desc; struct nft_set_elem elem; + struct nft_trans *trans; int err; err = nla_parse_nested(nla, NFTA_SET_ELEM_MAX, attr, @@ -3149,10 +3171,12 @@ static int nft_del_setelem(const struct nft_ctx *ctx, struct nft_set *set, if (err < 0) goto err2; - set->ops->remove(set, &elem); - set->nelems--; + trans = nft_trans_elem_alloc(ctx, NFT_MSG_DELSETELEM, set); + if (trans == NULL) + goto err2; - nf_tables_setelem_notify(ctx, set, &elem, NFT_MSG_DELSETELEM, 0); + nft_trans_elem(trans) = elem; + list_add(&trans->list, &ctx->net->nft.commit_list); nft_data_uninit(&elem.key, NFT_DATA_VALUE); if (set->flags & NFT_SET_MAP) @@ -3171,7 +3195,7 @@ static int nf_tables_delsetelem(struct sock *nlsk, struct sk_buff *skb, const struct nlattr *attr; struct nft_set *set; struct nft_ctx ctx; - int rem, err; + int rem, err = 0; err = nft_ctx_init_from_elemattr(&ctx, skb, nlh, nla, false); if (err < 0) @@ -3186,9 +3210,9 @@ static int nf_tables_delsetelem(struct sock *nlsk, struct sk_buff *skb, nla_for_each_nested(attr, nla[NFTA_SET_ELEM_LIST_ELEMENTS], rem) { err = nft_del_setelem(&ctx, set, attr); if (err < 0) - return err; + break; } - return 0; + return err; } static const struct nfnl_callback nf_tables_cb[NFT_MSG_MAX] = { @@ -3294,6 +3318,7 @@ static int nf_tables_commit(struct sk_buff *skb) { struct net *net = sock_net(skb->sk); struct nft_trans *trans, *next; + struct nft_set *set; /* Bump generation counter, invalidate any dump in progress */ net->nft.genctr++; @@ -3385,6 +3410,25 @@ static int nf_tables_commit(struct sk_buff *skb) nf_tables_set_notify(&trans->ctx, nft_trans_set(trans), NFT_MSG_DELSET); break; + case NFT_MSG_NEWSETELEM: + nft_trans_elem_set(trans)->nelems++; + nf_tables_setelem_notify(&trans->ctx, + nft_trans_elem_set(trans), + &nft_trans_elem(trans), + NFT_MSG_NEWSETELEM, 0); + nft_trans_destroy(trans); + break; + case NFT_MSG_DELSETELEM: + nft_trans_elem_set(trans)->nelems--; + nf_tables_setelem_notify(&trans->ctx, + nft_trans_elem_set(trans), + &nft_trans_elem(trans), + NFT_MSG_DELSETELEM, 0); + set = nft_trans_elem_set(trans); + set->ops->get(set, &nft_trans_elem(trans)); + set->ops->remove(set, &nft_trans_elem(trans)); + nft_trans_destroy(trans); + break; } } @@ -3418,6 +3462,7 @@ static int nf_tables_abort(struct sk_buff *skb) { struct net *net = sock_net(skb->sk); struct nft_trans *trans, *next; + struct nft_set *set; list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { switch (trans->msg_type) { @@ -3473,6 +3518,15 @@ static int nf_tables_abort(struct sk_buff *skb) &trans->ctx.table->sets); nft_trans_destroy(trans); break; + case NFT_MSG_NEWSETELEM: + set = nft_trans_elem_set(trans); + set->ops->get(set, &nft_trans_elem(trans)); + set->ops->remove(set, &nft_trans_elem(trans)); + nft_trans_destroy(trans); + break; + case NFT_MSG_DELSETELEM: + nft_trans_destroy(trans); + break; } } From 35151d840c60ce82692656c73bf6e5b922247ad2 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Mon, 5 May 2014 17:12:40 +0200 Subject: [PATCH 23/25] netfilter: nf_tables: simplify nf_tables_*_notify Now that all these function are called from the commit path, we can pass the context structure to reduce the amount of parameters in all of the nf_tables_*_notify functions. This patch also removes unneeded branches to check for skb, nlh and net that should be always set in the context structure. Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_tables_api.c | 90 +++++++++++++---------------------- 1 file changed, 32 insertions(+), 58 deletions(-) diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index cd002935e6b2..86d055a51f0b 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -235,19 +235,16 @@ nla_put_failure: return -1; } -static int nf_tables_table_notify(const struct sk_buff *oskb, - const struct nlmsghdr *nlh, - const struct nft_table *table, - int event, int family) +static int nf_tables_table_notify(const struct nft_ctx *ctx, int event) { struct sk_buff *skb; - u32 portid = oskb ? NETLINK_CB(oskb).portid : 0; - u32 seq = nlh ? nlh->nlmsg_seq : 0; - struct net *net = oskb ? sock_net(oskb->sk) : &init_net; + u32 portid = NETLINK_CB(ctx->skb).portid; + u32 seq = ctx->nlh->nlmsg_seq; + struct net *net = sock_net(ctx->skb->sk); bool report; int err; - report = nlh ? nlmsg_report(nlh) : false; + report = nlmsg_report(ctx->nlh); if (!report && !nfnetlink_has_listeners(net, NFNLGRP_NFTABLES)) return 0; @@ -257,7 +254,7 @@ static int nf_tables_table_notify(const struct sk_buff *oskb, goto err; err = nf_tables_fill_table_info(skb, portid, seq, event, 0, - family, table); + ctx->afi->family, ctx->table); if (err < 0) { kfree_skb(skb); goto err; @@ -721,20 +718,16 @@ nla_put_failure: return -1; } -static int nf_tables_chain_notify(const struct sk_buff *oskb, - const struct nlmsghdr *nlh, - const struct nft_table *table, - const struct nft_chain *chain, - int event, int family) +static int nf_tables_chain_notify(const struct nft_ctx *ctx, int event) { struct sk_buff *skb; - u32 portid = oskb ? NETLINK_CB(oskb).portid : 0; - struct net *net = oskb ? sock_net(oskb->sk) : &init_net; - u32 seq = nlh ? nlh->nlmsg_seq : 0; + u32 portid = NETLINK_CB(ctx->skb).portid; + struct net *net = sock_net(ctx->skb->sk); + u32 seq = ctx->nlh->nlmsg_seq; bool report; int err; - report = nlh ? nlmsg_report(nlh) : false; + report = nlmsg_report(ctx->nlh); if (!report && !nfnetlink_has_listeners(net, NFNLGRP_NFTABLES)) return 0; @@ -743,8 +736,9 @@ static int nf_tables_chain_notify(const struct sk_buff *oskb, if (skb == NULL) goto err; - err = nf_tables_fill_chain_info(skb, portid, seq, event, 0, family, - table, chain); + err = nf_tables_fill_chain_info(skb, portid, seq, event, 0, + ctx->afi->family, ctx->table, + ctx->chain); if (err < 0) { kfree_skb(skb); goto err; @@ -1475,21 +1469,19 @@ nla_put_failure: return -1; } -static int nf_tables_rule_notify(const struct sk_buff *oskb, - const struct nlmsghdr *nlh, - const struct nft_table *table, - const struct nft_chain *chain, +static int nf_tables_rule_notify(const struct nft_ctx *ctx, const struct nft_rule *rule, - int event, u32 flags, int family) + int event) { + const struct sk_buff *oskb = ctx->skb; struct sk_buff *skb; u32 portid = NETLINK_CB(oskb).portid; - struct net *net = oskb ? sock_net(oskb->sk) : &init_net; - u32 seq = nlh->nlmsg_seq; + struct net *net = sock_net(oskb->sk); + u32 seq = ctx->nlh->nlmsg_seq; bool report; int err; - report = nlmsg_report(nlh); + report = nlmsg_report(ctx->nlh); if (!report && !nfnetlink_has_listeners(net, NFNLGRP_NFTABLES)) return 0; @@ -1498,8 +1490,9 @@ static int nf_tables_rule_notify(const struct sk_buff *oskb, if (skb == NULL) goto err; - err = nf_tables_fill_rule_info(skb, portid, seq, event, flags, - family, table, chain, rule); + err = nf_tables_fill_rule_info(skb, portid, seq, event, 0, + ctx->afi->family, ctx->table, + ctx->chain, rule); if (err < 0) { kfree_skb(skb); goto err; @@ -3343,17 +3336,11 @@ static int nf_tables_commit(struct sk_buff *skb) } else { trans->ctx.table->flags &= ~NFT_TABLE_INACTIVE; } - nf_tables_table_notify(trans->ctx.skb, trans->ctx.nlh, - trans->ctx.table, - NFT_MSG_NEWTABLE, - trans->ctx.afi->family); + nf_tables_table_notify(&trans->ctx, NFT_MSG_NEWTABLE); nft_trans_destroy(trans); break; case NFT_MSG_DELTABLE: - nf_tables_table_notify(trans->ctx.skb, trans->ctx.nlh, - trans->ctx.table, - NFT_MSG_DELTABLE, - trans->ctx.afi->family); + nf_tables_table_notify(&trans->ctx, NFT_MSG_DELTABLE); break; case NFT_MSG_NEWCHAIN: if (nft_trans_chain_update(trans)) @@ -3362,20 +3349,12 @@ static int nf_tables_commit(struct sk_buff *skb) trans->ctx.chain->flags &= ~NFT_CHAIN_INACTIVE; trans->ctx.table->use++; } - nf_tables_chain_notify(trans->ctx.skb, trans->ctx.nlh, - trans->ctx.table, - trans->ctx.chain, - NFT_MSG_NEWCHAIN, - trans->ctx.afi->family); + nf_tables_chain_notify(&trans->ctx, NFT_MSG_NEWCHAIN); nft_trans_destroy(trans); break; case NFT_MSG_DELCHAIN: trans->ctx.table->use--; - nf_tables_chain_notify(trans->ctx.skb, trans->ctx.nlh, - trans->ctx.table, - trans->ctx.chain, - NFT_MSG_DELCHAIN, - trans->ctx.afi->family); + nf_tables_chain_notify(&trans->ctx, NFT_MSG_DELCHAIN); if (!(trans->ctx.table->flags & NFT_TABLE_F_DORMANT) && trans->ctx.chain->flags & NFT_BASE_CHAIN) { nf_unregister_hooks(nft_base_chain(trans->ctx.chain)->ops, @@ -3384,21 +3363,16 @@ static int nf_tables_commit(struct sk_buff *skb) break; case NFT_MSG_NEWRULE: nft_rule_clear(trans->ctx.net, nft_trans_rule(trans)); - nf_tables_rule_notify(trans->ctx.skb, trans->ctx.nlh, - trans->ctx.table, - trans->ctx.chain, + nf_tables_rule_notify(&trans->ctx, nft_trans_rule(trans), - NFT_MSG_NEWRULE, 0, - trans->ctx.afi->family); + NFT_MSG_NEWRULE); nft_trans_destroy(trans); break; case NFT_MSG_DELRULE: list_del_rcu(&nft_trans_rule(trans)->list); - nf_tables_rule_notify(trans->ctx.skb, trans->ctx.nlh, - trans->ctx.table, - trans->ctx.chain, - nft_trans_rule(trans), NFT_MSG_DELRULE, 0, - trans->ctx.afi->family); + nf_tables_rule_notify(&trans->ctx, + nft_trans_rule(trans), + NFT_MSG_DELRULE); break; case NFT_MSG_NEWSET: nft_trans_set(trans)->flags &= ~NFT_SET_INACTIVE; From 128ad3322ba5de8fa346203c9931d1fdcab8da87 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Fri, 9 May 2014 17:14:24 +0200 Subject: [PATCH 24/25] netfilter: nf_tables: remove skb and nlh from context structure Instead of caching the original skbuff that contains the netlink messages, this stores the netlink message sequence number, the netlink portID and the report flag. This helps to prepare the introduction of the object release via call_rcu. Signed-off-by: Pablo Neira Ayuso --- include/net/netfilter/nf_tables.h | 10 +-- net/netfilter/nf_tables_api.c | 101 ++++++++++++++---------------- 2 files changed, 52 insertions(+), 59 deletions(-) diff --git a/include/net/netfilter/nf_tables.h b/include/net/netfilter/nf_tables.h index b08f2a941007..1ed2797fb964 100644 --- a/include/net/netfilter/nf_tables.h +++ b/include/net/netfilter/nf_tables.h @@ -72,21 +72,23 @@ static inline void nft_data_debug(const struct nft_data *data) * struct nft_ctx - nf_tables rule/set context * * @net: net namespace - * @skb: netlink skb - * @nlh: netlink message header * @afi: address family info * @table: the table the chain is contained in * @chain: the chain the rule is contained in * @nla: netlink attributes + * @portid: netlink portID of the original message + * @seq: netlink sequence number + * @report: notify via unicast netlink message */ struct nft_ctx { struct net *net; - const struct sk_buff *skb; - const struct nlmsghdr *nlh; struct nft_af_info *afi; struct nft_table *table; struct nft_chain *chain; const struct nlattr * const *nla; + u32 portid; + u32 seq; + bool report; }; struct nft_data_desc { diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 86d055a51f0b..4147166ad17c 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -96,13 +96,14 @@ static void nft_ctx_init(struct nft_ctx *ctx, struct nft_chain *chain, const struct nlattr * const *nla) { - ctx->net = sock_net(skb->sk); - ctx->skb = skb; - ctx->nlh = nlh; - ctx->afi = afi; - ctx->table = table; - ctx->chain = chain; - ctx->nla = nla; + ctx->net = sock_net(skb->sk); + ctx->afi = afi; + ctx->table = table; + ctx->chain = chain; + ctx->nla = nla; + ctx->portid = NETLINK_CB(skb).portid; + ctx->report = nlmsg_report(nlh); + ctx->seq = nlh->nlmsg_seq; } static struct nft_trans *nft_trans_alloc(struct nft_ctx *ctx, int msg_type, @@ -238,14 +239,10 @@ nla_put_failure: static int nf_tables_table_notify(const struct nft_ctx *ctx, int event) { struct sk_buff *skb; - u32 portid = NETLINK_CB(ctx->skb).portid; - u32 seq = ctx->nlh->nlmsg_seq; - struct net *net = sock_net(ctx->skb->sk); - bool report; int err; - report = nlmsg_report(ctx->nlh); - if (!report && !nfnetlink_has_listeners(net, NFNLGRP_NFTABLES)) + if (!ctx->report && + !nfnetlink_has_listeners(ctx->net, NFNLGRP_NFTABLES)) return 0; err = -ENOBUFS; @@ -253,18 +250,20 @@ static int nf_tables_table_notify(const struct nft_ctx *ctx, int event) if (skb == NULL) goto err; - err = nf_tables_fill_table_info(skb, portid, seq, event, 0, + err = nf_tables_fill_table_info(skb, ctx->portid, ctx->seq, event, 0, ctx->afi->family, ctx->table); if (err < 0) { kfree_skb(skb); goto err; } - err = nfnetlink_send(skb, net, portid, NFNLGRP_NFTABLES, report, - GFP_KERNEL); + err = nfnetlink_send(skb, ctx->net, ctx->portid, NFNLGRP_NFTABLES, + ctx->report, GFP_KERNEL); err: - if (err < 0) - nfnetlink_set_err(net, portid, NFNLGRP_NFTABLES, err); + if (err < 0) { + nfnetlink_set_err(ctx->net, ctx->portid, NFNLGRP_NFTABLES, + err); + } return err; } @@ -721,14 +720,10 @@ nla_put_failure: static int nf_tables_chain_notify(const struct nft_ctx *ctx, int event) { struct sk_buff *skb; - u32 portid = NETLINK_CB(ctx->skb).portid; - struct net *net = sock_net(ctx->skb->sk); - u32 seq = ctx->nlh->nlmsg_seq; - bool report; int err; - report = nlmsg_report(ctx->nlh); - if (!report && !nfnetlink_has_listeners(net, NFNLGRP_NFTABLES)) + if (!ctx->report && + !nfnetlink_has_listeners(ctx->net, NFNLGRP_NFTABLES)) return 0; err = -ENOBUFS; @@ -736,7 +731,7 @@ static int nf_tables_chain_notify(const struct nft_ctx *ctx, int event) if (skb == NULL) goto err; - err = nf_tables_fill_chain_info(skb, portid, seq, event, 0, + err = nf_tables_fill_chain_info(skb, ctx->portid, ctx->seq, event, 0, ctx->afi->family, ctx->table, ctx->chain); if (err < 0) { @@ -744,11 +739,13 @@ static int nf_tables_chain_notify(const struct nft_ctx *ctx, int event) goto err; } - err = nfnetlink_send(skb, net, portid, NFNLGRP_NFTABLES, report, - GFP_KERNEL); + err = nfnetlink_send(skb, ctx->net, ctx->portid, NFNLGRP_NFTABLES, + ctx->report, GFP_KERNEL); err: - if (err < 0) - nfnetlink_set_err(net, portid, NFNLGRP_NFTABLES, err); + if (err < 0) { + nfnetlink_set_err(ctx->net, ctx->portid, NFNLGRP_NFTABLES, + err); + } return err; } @@ -1473,16 +1470,11 @@ static int nf_tables_rule_notify(const struct nft_ctx *ctx, const struct nft_rule *rule, int event) { - const struct sk_buff *oskb = ctx->skb; struct sk_buff *skb; - u32 portid = NETLINK_CB(oskb).portid; - struct net *net = sock_net(oskb->sk); - u32 seq = ctx->nlh->nlmsg_seq; - bool report; int err; - report = nlmsg_report(ctx->nlh); - if (!report && !nfnetlink_has_listeners(net, NFNLGRP_NFTABLES)) + if (!ctx->report && + !nfnetlink_has_listeners(ctx->net, NFNLGRP_NFTABLES)) return 0; err = -ENOBUFS; @@ -1490,7 +1482,7 @@ static int nf_tables_rule_notify(const struct nft_ctx *ctx, if (skb == NULL) goto err; - err = nf_tables_fill_rule_info(skb, portid, seq, event, 0, + err = nf_tables_fill_rule_info(skb, ctx->portid, ctx->seq, event, 0, ctx->afi->family, ctx->table, ctx->chain, rule); if (err < 0) { @@ -1498,11 +1490,13 @@ static int nf_tables_rule_notify(const struct nft_ctx *ctx, goto err; } - err = nfnetlink_send(skb, net, portid, NFNLGRP_NFTABLES, report, - GFP_KERNEL); + err = nfnetlink_send(skb, ctx->net, ctx->portid, NFNLGRP_NFTABLES, + ctx->report, GFP_KERNEL); err: - if (err < 0) - nfnetlink_set_err(net, portid, NFNLGRP_NFTABLES, err); + if (err < 0) { + nfnetlink_set_err(ctx->net, ctx->portid, NFNLGRP_NFTABLES, + err); + } return err; } @@ -2141,8 +2135,8 @@ static int nf_tables_fill_set(struct sk_buff *skb, const struct nft_ctx *ctx, struct nfgenmsg *nfmsg; struct nlmsghdr *nlh; struct nlattr *desc; - u32 portid = NETLINK_CB(ctx->skb).portid; - u32 seq = ctx->nlh->nlmsg_seq; + u32 portid = ctx->portid; + u32 seq = ctx->seq; event |= NFNL_SUBSYS_NFTABLES << 8; nlh = nlmsg_put(skb, portid, seq, event, sizeof(struct nfgenmsg), @@ -2194,12 +2188,11 @@ static int nf_tables_set_notify(const struct nft_ctx *ctx, int event) { struct sk_buff *skb; - u32 portid = NETLINK_CB(ctx->skb).portid; - bool report; + u32 portid = ctx->portid; int err; - report = nlmsg_report(ctx->nlh); - if (!report && !nfnetlink_has_listeners(ctx->net, NFNLGRP_NFTABLES)) + if (!ctx->report && + !nfnetlink_has_listeners(ctx->net, NFNLGRP_NFTABLES)) return 0; err = -ENOBUFS; @@ -2213,8 +2206,8 @@ static int nf_tables_set_notify(const struct nft_ctx *ctx, goto err; } - err = nfnetlink_send(skb, ctx->net, portid, NFNLGRP_NFTABLES, report, - GFP_KERNEL); + err = nfnetlink_send(skb, ctx->net, portid, NFNLGRP_NFTABLES, + ctx->report, GFP_KERNEL); err: if (err < 0) nfnetlink_set_err(ctx->net, portid, NFNLGRP_NFTABLES, err); @@ -2956,14 +2949,12 @@ static int nf_tables_setelem_notify(const struct nft_ctx *ctx, const struct nft_set_elem *elem, int event, u16 flags) { - const struct sk_buff *oskb = ctx->skb; - struct net *net = sock_net(oskb->sk); - u32 portid = NETLINK_CB(oskb).portid; - bool report = nlmsg_report(ctx->nlh); + struct net *net = ctx->net; + u32 portid = ctx->portid; struct sk_buff *skb; int err; - if (!report && !nfnetlink_has_listeners(net, NFNLGRP_NFTABLES)) + if (!ctx->report && !nfnetlink_has_listeners(net, NFNLGRP_NFTABLES)) return 0; err = -ENOBUFS; @@ -2978,7 +2969,7 @@ static int nf_tables_setelem_notify(const struct nft_ctx *ctx, goto err; } - err = nfnetlink_send(skb, net, portid, NFNLGRP_NFTABLES, report, + err = nfnetlink_send(skb, net, portid, NFNLGRP_NFTABLES, ctx->report, GFP_KERNEL); err: if (err < 0) From c7c32e72cbe23cea97c5d87ffcf6e23cc1ec1a65 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Thu, 10 Apr 2014 00:31:10 +0200 Subject: [PATCH 25/25] netfilter: nf_tables: defer all object release via rcu Now that all objects are released in the reverse order via the transaction infrastructure, we can enqueue the release via call_rcu to save one synchronize_rcu. For small rule-sets loaded via nft -f, it now takes around 50ms less here. Signed-off-by: Pablo Neira Ayuso --- include/net/netfilter/nf_tables.h | 2 + net/netfilter/nf_tables_api.c | 93 ++++++++++++++++++------------- 2 files changed, 56 insertions(+), 39 deletions(-) diff --git a/include/net/netfilter/nf_tables.h b/include/net/netfilter/nf_tables.h index 1ed2797fb964..7ee6ce6564ae 100644 --- a/include/net/netfilter/nf_tables.h +++ b/include/net/netfilter/nf_tables.h @@ -393,12 +393,14 @@ struct nft_rule { /** * struct nft_trans - nf_tables object update in transaction * + * @rcu_head: rcu head to defer release of transaction data * @list: used internally * @msg_type: message type * @ctx: transaction context * @data: internal information related to the transaction */ struct nft_trans { + struct rcu_head rcu_head; struct list_head list; int msg_type; struct nft_ctx ctx; diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index 4147166ad17c..e171c635d08c 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -3298,6 +3298,30 @@ static void nft_chain_commit_update(struct nft_trans *trans) } } +/* Schedule objects for release via rcu to make sure no packets are accesing + * removed rules. + */ +static void nf_tables_commit_release_rcu(struct rcu_head *rt) +{ + struct nft_trans *trans = container_of(rt, struct nft_trans, rcu_head); + + switch (trans->msg_type) { + case NFT_MSG_DELTABLE: + nf_tables_table_destroy(&trans->ctx); + break; + case NFT_MSG_DELCHAIN: + nf_tables_chain_destroy(trans->ctx.chain); + break; + case NFT_MSG_DELRULE: + nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); + break; + case NFT_MSG_DELSET: + nft_set_destroy(nft_trans_set(trans)); + break; + } + kfree(trans); +} + static int nf_tables_commit(struct sk_buff *skb) { struct net *net = sock_net(skb->sk); @@ -3397,32 +3421,39 @@ static int nf_tables_commit(struct sk_buff *skb) } } - /* Make sure we don't see any packet traversing old rules */ - synchronize_rcu(); - - /* Now we can safely release unused old rules */ list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { - switch (trans->msg_type) { - case NFT_MSG_DELTABLE: - nf_tables_table_destroy(&trans->ctx); - break; - case NFT_MSG_DELCHAIN: - nf_tables_chain_destroy(trans->ctx.chain); - break; - case NFT_MSG_DELRULE: - nf_tables_rule_destroy(&trans->ctx, - nft_trans_rule(trans)); - break; - case NFT_MSG_DELSET: - nft_set_destroy(nft_trans_set(trans)); - break; - } - nft_trans_destroy(trans); + list_del(&trans->list); + trans->ctx.nla = NULL; + call_rcu(&trans->rcu_head, nf_tables_commit_release_rcu); } return 0; } +/* Schedule objects for release via rcu to make sure no packets are accesing + * aborted rules. + */ +static void nf_tables_abort_release_rcu(struct rcu_head *rt) +{ + struct nft_trans *trans = container_of(rt, struct nft_trans, rcu_head); + + switch (trans->msg_type) { + case NFT_MSG_NEWTABLE: + nf_tables_table_destroy(&trans->ctx); + break; + case NFT_MSG_NEWCHAIN: + nf_tables_chain_destroy(trans->ctx.chain); + break; + case NFT_MSG_NEWRULE: + nf_tables_rule_destroy(&trans->ctx, nft_trans_rule(trans)); + break; + case NFT_MSG_NEWSET: + nft_set_destroy(nft_trans_set(trans)); + break; + } + kfree(trans); +} + static int nf_tables_abort(struct sk_buff *skb) { struct net *net = sock_net(skb->sk); @@ -3495,26 +3526,10 @@ static int nf_tables_abort(struct sk_buff *skb) } } - /* Make sure we don't see any packet accessing aborted rules */ - synchronize_rcu(); - list_for_each_entry_safe(trans, next, &net->nft.commit_list, list) { - switch (trans->msg_type) { - case NFT_MSG_NEWTABLE: - nf_tables_table_destroy(&trans->ctx); - break; - case NFT_MSG_NEWCHAIN: - nf_tables_chain_destroy(trans->ctx.chain); - break; - case NFT_MSG_NEWRULE: - nf_tables_rule_destroy(&trans->ctx, - nft_trans_rule(trans)); - break; - case NFT_MSG_NEWSET: - nft_set_destroy(nft_trans_set(trans)); - break; - } - nft_trans_destroy(trans); + list_del(&trans->list); + trans->ctx.nla = NULL; + call_rcu(&trans->rcu_head, nf_tables_abort_release_rcu); } return 0;