From c4a3922d2d20c710f827d3a115ee338e8d0467df Mon Sep 17 00:00:00 2001 From: Cong Wang Date: Sun, 2 Feb 2020 20:30:52 -0800 Subject: netfilter: xt_hashlimit: reduce hashlimit_mutex scope for htable_put() It is unnecessary to hold hashlimit_mutex for htable_destroy() as it is already removed from the global hashtable and its refcount is already zero. Also, switch hinfo->use to refcount_t so that we don't have to hold the mutex until it reaches zero in htable_put(). Reported-and-tested-by: syzbot+adf6c6c2be1c3a718121@syzkaller.appspotmail.com Acked-by: Florian Westphal Signed-off-by: Cong Wang Signed-off-by: Pablo Neira Ayuso --- net/netfilter/xt_hashlimit.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) (limited to 'net') diff --git a/net/netfilter/xt_hashlimit.c b/net/netfilter/xt_hashlimit.c index bccd47cd7190..cc475a608f81 100644 --- a/net/netfilter/xt_hashlimit.c +++ b/net/netfilter/xt_hashlimit.c @@ -36,6 +36,7 @@ #include #include #include +#include #include #define XT_HASHLIMIT_ALL (XT_HASHLIMIT_HASH_DIP | XT_HASHLIMIT_HASH_DPT | \ @@ -114,7 +115,7 @@ struct dsthash_ent { struct xt_hashlimit_htable { struct hlist_node node; /* global list of all htables */ - int use; + refcount_t use; u_int8_t family; bool rnd_initialized; @@ -315,7 +316,7 @@ static int htable_create(struct net *net, struct hashlimit_cfg3 *cfg, for (i = 0; i < hinfo->cfg.size; i++) INIT_HLIST_HEAD(&hinfo->hash[i]); - hinfo->use = 1; + refcount_set(&hinfo->use, 1); hinfo->count = 0; hinfo->family = family; hinfo->rnd_initialized = false; @@ -420,7 +421,7 @@ static struct xt_hashlimit_htable *htable_find_get(struct net *net, hlist_for_each_entry(hinfo, &hashlimit_net->htables, node) { if (!strcmp(name, hinfo->name) && hinfo->family == family) { - hinfo->use++; + refcount_inc(&hinfo->use); return hinfo; } } @@ -429,12 +430,11 @@ static struct xt_hashlimit_htable *htable_find_get(struct net *net, static void htable_put(struct xt_hashlimit_htable *hinfo) { - mutex_lock(&hashlimit_mutex); - if (--hinfo->use == 0) { + if (refcount_dec_and_mutex_lock(&hinfo->use, &hashlimit_mutex)) { hlist_del(&hinfo->node); + mutex_unlock(&hashlimit_mutex); htable_destroy(hinfo); } - mutex_unlock(&hashlimit_mutex); } /* The algorithm used is the Simple Token Bucket Filter (TBF) -- cgit v1.2.3-59-g8ed1b From 8d0015a7ab76b8b1e89a3e5f5710a6e5103f2dd5 Mon Sep 17 00:00:00 2001 From: Cong Wang Date: Sun, 2 Feb 2020 20:30:53 -0800 Subject: netfilter: xt_hashlimit: limit the max size of hashtable The user-specified hashtable size is unbound, this could easily lead to an OOM or a hung task as we hold the global mutex while allocating and initializing the new hashtable. Add a max value to cap both cfg->size and cfg->max, as suggested by Florian. Reported-and-tested-by: syzbot+adf6c6c2be1c3a718121@syzkaller.appspotmail.com Signed-off-by: Cong Wang Reviewed-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- net/netfilter/xt_hashlimit.c | 10 ++++++++++ 1 file changed, 10 insertions(+) (limited to 'net') diff --git a/net/netfilter/xt_hashlimit.c b/net/netfilter/xt_hashlimit.c index cc475a608f81..7a2c4b8408c4 100644 --- a/net/netfilter/xt_hashlimit.c +++ b/net/netfilter/xt_hashlimit.c @@ -837,6 +837,8 @@ hashlimit_mt(const struct sk_buff *skb, struct xt_action_param *par) return hashlimit_mt_common(skb, par, hinfo, &info->cfg, 3); } +#define HASHLIMIT_MAX_SIZE 1048576 + static int hashlimit_mt_check_common(const struct xt_mtchk_param *par, struct xt_hashlimit_htable **hinfo, struct hashlimit_cfg3 *cfg, @@ -847,6 +849,14 @@ static int hashlimit_mt_check_common(const struct xt_mtchk_param *par, if (cfg->gc_interval == 0 || cfg->expire == 0) return -EINVAL; + if (cfg->size > HASHLIMIT_MAX_SIZE) { + cfg->size = HASHLIMIT_MAX_SIZE; + pr_info_ratelimited("size too large, truncated to %u\n", cfg->size); + } + if (cfg->max > HASHLIMIT_MAX_SIZE) { + cfg->max = HASHLIMIT_MAX_SIZE; + pr_info_ratelimited("max too large, truncated to %u\n", cfg->max); + } if (par->family == NFPROTO_IPV4) { if (cfg->srcmask > 32 || cfg->dstmask > 32) return -EINVAL; -- cgit v1.2.3-59-g8ed1b From a7da92c2c8a1faf253a3b3e292fda6910deba540 Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Mon, 3 Feb 2020 13:06:18 +0100 Subject: netfilter: flowtable: skip offload setup if disabled nftables test case tests/shell/testcases/flowtable/0001flowtable_0 results in a crash. After the refactor, if we leave early via nf_flowtable_hw_offload(), then "struct flow_block_offload" is left in an uninitialized state, but later users assume its initialised. Fixes: a7965d58ddab02 ("netfilter: flowtable: add nf_flow_table_offload_cmd()") Signed-off-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_flow_table_offload.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'net') diff --git a/net/netfilter/nf_flow_table_offload.c b/net/netfilter/nf_flow_table_offload.c index 83e1db37c3b0..06f00cdc3891 100644 --- a/net/netfilter/nf_flow_table_offload.c +++ b/net/netfilter/nf_flow_table_offload.c @@ -847,9 +847,6 @@ static int nf_flow_table_offload_cmd(struct flow_block_offload *bo, { int err; - if (!nf_flowtable_hw_offload(flowtable)) - return 0; - if (!dev->netdev_ops->ndo_setup_tc) return -EOPNOTSUPP; @@ -876,6 +873,9 @@ int nf_flow_table_offload_setup(struct nf_flowtable *flowtable, struct flow_block_offload bo; int err; + if (!nf_flowtable_hw_offload(flowtable)) + return 0; + err = nf_flow_table_offload_cmd(&bo, flowtable, dev, cmd, &extack); if (err < 0) return err; -- cgit v1.2.3-59-g8ed1b From 3d1e0b406de16508de96f4a07fc3f94cfc678372 Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Mon, 3 Feb 2020 17:37:04 +0100 Subject: netfilter: conntrack: remove two args from resolve_clash ctinfo is whats taken from the skb, i.e. ct = nf_ct_get(skb, &ctinfo). We do not pass 'ct' and instead re-fetch it from the skb. Just do the same for both netns and ctinfo. Also add a comment on what clash resolution is supposed to do. While at it, one indent level can be removed. Signed-off-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_conntrack_core.c | 69 +++++++++++++++++++++++++++++---------- 1 file changed, 51 insertions(+), 18 deletions(-) (limited to 'net') diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c index d1305423640f..5e332b01f3c0 100644 --- a/net/netfilter/nf_conntrack_core.c +++ b/net/netfilter/nf_conntrack_core.c @@ -894,31 +894,64 @@ static void nf_ct_acct_merge(struct nf_conn *ct, enum ip_conntrack_info ctinfo, } } -/* Resolve race on insertion if this protocol allows this. */ +/** + * nf_ct_resolve_clash - attempt to handle clash without packet drop + * + * @skb: skb that causes the clash + * @h: tuplehash of the clashing entry already in table + * + * A conntrack entry can be inserted to the connection tracking table + * if there is no existing entry with an identical tuple. + * + * If there is one, @skb (and the assocated, unconfirmed conntrack) has + * to be dropped. In case @skb is retransmitted, next conntrack lookup + * will find the already-existing entry. + * + * The major problem with such packet drop is the extra delay added by + * the packet loss -- it will take some time for a retransmit to occur + * (or the sender to time out when waiting for a reply). + * + * This function attempts to handle the situation without packet drop. + * + * If @skb has no NAT transformation or if the colliding entries are + * exactly the same, only the to-be-confirmed conntrack entry is discarded + * and @skb is associated with the conntrack entry already in the table. + * + * Returns NF_DROP if the clash could not be resolved. + */ static __cold noinline int -nf_ct_resolve_clash(struct net *net, struct sk_buff *skb, - enum ip_conntrack_info ctinfo, - struct nf_conntrack_tuple_hash *h) +nf_ct_resolve_clash(struct sk_buff *skb, struct nf_conntrack_tuple_hash *h) { /* This is the conntrack entry already in hashes that won race. */ struct nf_conn *ct = nf_ct_tuplehash_to_ctrack(h); const struct nf_conntrack_l4proto *l4proto; - enum ip_conntrack_info oldinfo; - struct nf_conn *loser_ct = nf_ct_get(skb, &oldinfo); + enum ip_conntrack_info ctinfo; + struct nf_conn *loser_ct; + struct net *net; + + loser_ct = nf_ct_get(skb, &ctinfo); l4proto = nf_ct_l4proto_find(nf_ct_protonum(ct)); - if (l4proto->allow_clash && - !nf_ct_is_dying(ct) && - atomic_inc_not_zero(&ct->ct_general.use)) { - if (((ct->status & IPS_NAT_DONE_MASK) == 0) || - nf_ct_match(ct, loser_ct)) { - nf_ct_acct_merge(ct, ctinfo, loser_ct); - nf_conntrack_put(&loser_ct->ct_general); - nf_ct_set(skb, ct, oldinfo); - return NF_ACCEPT; - } - nf_ct_put(ct); + if (!l4proto->allow_clash) + goto drop; + + if (nf_ct_is_dying(ct)) + goto drop; + + if (!atomic_inc_not_zero(&ct->ct_general.use)) + goto drop; + + if (((ct->status & IPS_NAT_DONE_MASK) == 0) || + nf_ct_match(ct, loser_ct)) { + nf_ct_acct_merge(ct, ctinfo, loser_ct); + nf_conntrack_put(&loser_ct->ct_general); + nf_ct_set(skb, ct, ctinfo); + return NF_ACCEPT; } + + nf_ct_put(ct); +drop: + net = nf_ct_net(loser_ct); NF_CT_STAT_INC(net, drop); return NF_DROP; } @@ -1036,7 +1069,7 @@ __nf_conntrack_confirm(struct sk_buff *skb) out: nf_ct_add_to_dying_list(ct); - ret = nf_ct_resolve_clash(net, skb, ctinfo, h); + ret = nf_ct_resolve_clash(skb, h); dying: nf_conntrack_double_unlock(hash, reply_hash); NF_CT_STAT_INC(net, insert_failed); -- cgit v1.2.3-59-g8ed1b From b1b32552c1d81f0cf6a8e79043a2a47e769ff071 Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Mon, 3 Feb 2020 17:37:05 +0100 Subject: netfilter: conntrack: place confirm-bit setting in a helper ... so it can be re-used from clash resolution in followup patch. Signed-off-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_conntrack_core.c | 21 ++++++++++++++------- 1 file changed, 14 insertions(+), 7 deletions(-) (limited to 'net') diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c index 5e332b01f3c0..5fda5bd10160 100644 --- a/net/netfilter/nf_conntrack_core.c +++ b/net/netfilter/nf_conntrack_core.c @@ -894,6 +894,19 @@ static void nf_ct_acct_merge(struct nf_conn *ct, enum ip_conntrack_info ctinfo, } } +static void __nf_conntrack_insert_prepare(struct nf_conn *ct) +{ + struct nf_conn_tstamp *tstamp; + + atomic_inc(&ct->ct_general.use); + ct->status |= IPS_CONFIRMED; + + /* set conntrack timestamp, if enabled. */ + tstamp = nf_conn_tstamp_find(ct); + if (tstamp) + tstamp->start = ktime_get_real_ns(); +} + /** * nf_ct_resolve_clash - attempt to handle clash without packet drop * @@ -965,7 +978,6 @@ __nf_conntrack_confirm(struct sk_buff *skb) struct nf_conntrack_tuple_hash *h; struct nf_conn *ct; struct nf_conn_help *help; - struct nf_conn_tstamp *tstamp; struct hlist_nulls_node *n; enum ip_conntrack_info ctinfo; struct net *net; @@ -1042,13 +1054,8 @@ __nf_conntrack_confirm(struct sk_buff *skb) setting time, otherwise we'd get timer wrap in weird delay cases. */ ct->timeout += nfct_time_stamp; - atomic_inc(&ct->ct_general.use); - ct->status |= IPS_CONFIRMED; - /* set conntrack timestamp, if enabled. */ - tstamp = nf_conn_tstamp_find(ct); - if (tstamp) - tstamp->start = ktime_get_real_ns(); + __nf_conntrack_insert_prepare(ct); /* Since the lookup is lockless, hash insertion must be done after * starting the timer and setting the CONFIRMED bit. The RCU barriers -- cgit v1.2.3-59-g8ed1b From bb89abe52bf426f1f40850c441efc77426cc31e1 Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Mon, 3 Feb 2020 17:37:06 +0100 Subject: netfilter: conntrack: split resolve_clash function Followup patch will need a helper function with the 'clashing entries refer to the identical tuple in both directions' resolution logic. This patch will add another resolve_clash helper where loser_ct must not be added to the dying list because it will be inserted into the table. Therefore this also moves the stat counters and dying-list insertion of the losing ct. Signed-off-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_conntrack_core.c | 58 +++++++++++++++++++++++++++------------ 1 file changed, 41 insertions(+), 17 deletions(-) (limited to 'net') diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c index 5fda5bd10160..3f069eb0f0fc 100644 --- a/net/netfilter/nf_conntrack_core.c +++ b/net/netfilter/nf_conntrack_core.c @@ -907,6 +907,39 @@ static void __nf_conntrack_insert_prepare(struct nf_conn *ct) tstamp->start = ktime_get_real_ns(); } +static int __nf_ct_resolve_clash(struct sk_buff *skb, + struct nf_conntrack_tuple_hash *h) +{ + /* This is the conntrack entry already in hashes that won race. */ + struct nf_conn *ct = nf_ct_tuplehash_to_ctrack(h); + enum ip_conntrack_info ctinfo; + struct nf_conn *loser_ct; + + loser_ct = nf_ct_get(skb, &ctinfo); + + if (nf_ct_is_dying(ct)) + return NF_DROP; + + if (!atomic_inc_not_zero(&ct->ct_general.use)) + return NF_DROP; + + if (((ct->status & IPS_NAT_DONE_MASK) == 0) || + nf_ct_match(ct, loser_ct)) { + struct net *net = nf_ct_net(ct); + + nf_ct_acct_merge(ct, ctinfo, loser_ct); + nf_ct_add_to_dying_list(loser_ct); + nf_conntrack_put(&loser_ct->ct_general); + nf_ct_set(skb, ct, ctinfo); + + NF_CT_STAT_INC(net, insert_failed); + return NF_ACCEPT; + } + + nf_ct_put(ct); + return NF_DROP; +} + /** * nf_ct_resolve_clash - attempt to handle clash without packet drop * @@ -941,31 +974,23 @@ nf_ct_resolve_clash(struct sk_buff *skb, struct nf_conntrack_tuple_hash *h) enum ip_conntrack_info ctinfo; struct nf_conn *loser_ct; struct net *net; + int ret; loser_ct = nf_ct_get(skb, &ctinfo); + net = nf_ct_net(loser_ct); l4proto = nf_ct_l4proto_find(nf_ct_protonum(ct)); if (!l4proto->allow_clash) goto drop; - if (nf_ct_is_dying(ct)) - goto drop; - - if (!atomic_inc_not_zero(&ct->ct_general.use)) - goto drop; - - if (((ct->status & IPS_NAT_DONE_MASK) == 0) || - nf_ct_match(ct, loser_ct)) { - nf_ct_acct_merge(ct, ctinfo, loser_ct); - nf_conntrack_put(&loser_ct->ct_general); - nf_ct_set(skb, ct, ctinfo); - return NF_ACCEPT; - } + ret = __nf_ct_resolve_clash(skb, h); + if (ret == NF_ACCEPT) + return ret; - nf_ct_put(ct); drop: - net = nf_ct_net(loser_ct); + nf_ct_add_to_dying_list(loser_ct); NF_CT_STAT_INC(net, drop); + NF_CT_STAT_INC(net, insert_failed); return NF_DROP; } @@ -1034,6 +1059,7 @@ __nf_conntrack_confirm(struct sk_buff *skb) if (unlikely(nf_ct_is_dying(ct))) { nf_ct_add_to_dying_list(ct); + NF_CT_STAT_INC(net, insert_failed); goto dying; } @@ -1075,11 +1101,9 @@ __nf_conntrack_confirm(struct sk_buff *skb) return NF_ACCEPT; out: - nf_ct_add_to_dying_list(ct); ret = nf_ct_resolve_clash(skb, h); dying: nf_conntrack_double_unlock(hash, reply_hash); - NF_CT_STAT_INC(net, insert_failed); local_bh_enable(); return ret; } -- cgit v1.2.3-59-g8ed1b From 6a757c07e51f80ac34325fcd558490d2d1439e1b Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Mon, 3 Feb 2020 17:37:07 +0100 Subject: netfilter: conntrack: allow insertion of clashing entries This patch further relaxes the need to drop an skb due to a clash with an existing conntrack entry. Current clash resolution handles the case where the clash occurs between two identical entries (distinct nf_conn objects with same tuples), i.e.: Original Reply existing: 10.2.3.4:42 -> 10.8.8.8:53 10.2.3.4:42 <- 10.0.0.6:5353 clashing: 10.2.3.4:42 -> 10.8.8.8:53 10.2.3.4:42 <- 10.0.0.6:5353 ... existing handling will discard the unconfirmed clashing entry and makes skb->_nfct point to the existing one. The skb can then be processed normally just as if the clash would not have existed in the first place. For other clashes, the skb needs to be dropped. This frequently happens with DNS resolvers that send A and AAAA queries back-to-back when NAT rules are present that cause packets to get different DNAT transformations applied, for example: -m statistics --mode random ... -j DNAT --dnat-to 10.0.0.6:5353 -m statistics --mode random ... -j DNAT --dnat-to 10.0.0.7:5353 In this case the A or AAAA query is dropped which incurs a costly delay during name resolution. This patch also allows this collision type: Original Reply existing: 10.2.3.4:42 -> 10.8.8.8:53 10.2.3.4:42 <- 10.0.0.6:5353 clashing: 10.2.3.4:42 -> 10.8.8.8:53 10.2.3.4:42 <- 10.0.0.7:5353 In this case, clash is in original direction -- the reply direction is still unique. The change makes it so that when the 2nd colliding packet is received, the clashing conntrack is tagged with new IPS_NAT_CLASH_BIT, gets a fixed 1 second timeout and is inserted in the reply direction only. The entry is hidden from 'conntrack -L', it will time out quickly and it can be early dropped because it will never progress to the ASSURED state. To avoid special-casing the delete code path to special case the ORIGINAL hlist_nulls node, a new helper, "hlist_nulls_add_fake", is added so hlist_nulls_del() will work. Example: CPU A: CPU B: 1. 10.2.3.4:42 -> 10.8.8.8:53 (A) 2. 10.2.3.4:42 -> 10.8.8.8:53 (AAAA) 3. Apply DNAT, reply changed to 10.0.0.6 4. 10.2.3.4:42 -> 10.8.8.8:53 (AAAA) 5. Apply DNAT, reply changed to 10.0.0.7 6. confirm/commit to conntrack table, no collisions 7. commit clashing entry Reply comes in: 10.2.3.4:42 <- 10.0.0.6:5353 (A) -> Finds a conntrack, DNAT is reversed & packet forwarded to 10.2.3.4:42 10.2.3.4:42 <- 10.0.0.7:5353 (AAAA) -> Finds a conntrack, DNAT is reversed & packet forwarded to 10.2.3.4:42 The conntrack entry is deleted from table, as it has the NAT_CLASH bit set. In case of a retransmit from ORIGINAL dir, all further packets will get the DNAT transformation to 10.0.0.6. I tried to come up with other solutions but they all have worse problems. Alternatives considered were: 1. Confirm ct entries at allocation time, not in postrouting. a. will cause uneccesarry work when the skb that creates the conntrack is dropped by ruleset. b. in case nat is applied, ct entry would need to be moved in the table, which requires another spinlock pair to be taken. c. breaks the 'unconfirmed entry is private to cpu' assumption: we would need to guard all nfct->ext allocation requests with ct->lock spinlock. 2. Make the unconfirmed list a hash table instead of a pcpu list. Shares drawback c) of the first alternative. 3. Document this is expected and force users to rearrange their ruleset (e.g. by using "-m cluster" instead of "-m statistics"). nft has the 'jhash' expression which can be used instead of 'numgen'. Major drawback: doesn't fix what I consider a bug, not very realistic and I believe its reasonable to have the existing rulesets to 'just work'. 4. Document this is expected and force users to steer problematic packets to the same CPU -- this would serialize the "allocate new conntrack entry/nat table evaluation/perform nat/confirm entry", so no race can occur. Similar drawback to 3. Another advantage of this patch compared to 1) and 2) is that there are no changes to the hot path; things are handled in the udp tracker and the clash resolution path. Cc: rcu@vger.kernel.org Cc: "Paul E. McKenney" Cc: Josh Triplett Cc: Jozsef Kadlecsik Signed-off-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- include/linux/rculist_nulls.h | 7 ++ include/uapi/linux/netfilter/nf_conntrack_common.h | 12 +++- net/netfilter/nf_conntrack_core.c | 76 +++++++++++++++++++++- net/netfilter/nf_conntrack_proto_udp.c | 20 ++++-- 4 files changed, 108 insertions(+), 7 deletions(-) (limited to 'net') diff --git a/include/linux/rculist_nulls.h b/include/linux/rculist_nulls.h index e5b752027a03..9670b54b484a 100644 --- a/include/linux/rculist_nulls.h +++ b/include/linux/rculist_nulls.h @@ -145,6 +145,13 @@ static inline void hlist_nulls_add_tail_rcu(struct hlist_nulls_node *n, } } +/* after that hlist_nulls_del will work */ +static inline void hlist_nulls_add_fake(struct hlist_nulls_node *n) +{ + n->pprev = &n->next; + n->next = (struct hlist_nulls_node *)NULLS_MARKER(NULL); +} + /** * hlist_nulls_for_each_entry_rcu - iterate over rcu list of given type * @tpos: the type * to use as a loop cursor. diff --git a/include/uapi/linux/netfilter/nf_conntrack_common.h b/include/uapi/linux/netfilter/nf_conntrack_common.h index 336014bf8868..b6f0bb1dc799 100644 --- a/include/uapi/linux/netfilter/nf_conntrack_common.h +++ b/include/uapi/linux/netfilter/nf_conntrack_common.h @@ -97,6 +97,15 @@ enum ip_conntrack_status { IPS_UNTRACKED_BIT = 12, IPS_UNTRACKED = (1 << IPS_UNTRACKED_BIT), +#ifdef __KERNEL__ + /* Re-purposed for in-kernel use: + * Tags a conntrack entry that clashed with an existing entry + * on insert. + */ + IPS_NAT_CLASH_BIT = IPS_UNTRACKED_BIT, + IPS_NAT_CLASH = IPS_UNTRACKED, +#endif + /* Conntrack got a helper explicitly attached via CT target. */ IPS_HELPER_BIT = 13, IPS_HELPER = (1 << IPS_HELPER_BIT), @@ -110,7 +119,8 @@ enum ip_conntrack_status { */ IPS_UNCHANGEABLE_MASK = (IPS_NAT_DONE_MASK | IPS_NAT_MASK | IPS_EXPECTED | IPS_CONFIRMED | IPS_DYING | - IPS_SEQ_ADJUST | IPS_TEMPLATE | IPS_OFFLOAD), + IPS_SEQ_ADJUST | IPS_TEMPLATE | IPS_UNTRACKED | + IPS_OFFLOAD), __IPS_MAX_BIT = 15, }; diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c index 3f069eb0f0fc..1927fc296f95 100644 --- a/net/netfilter/nf_conntrack_core.c +++ b/net/netfilter/nf_conntrack_core.c @@ -940,11 +940,71 @@ static int __nf_ct_resolve_clash(struct sk_buff *skb, return NF_DROP; } +/** + * nf_ct_resolve_clash_harder - attempt to insert clashing conntrack entry + * + * @skb: skb that causes the collision + * @repl_idx: hash slot for reply direction + * + * Called when origin or reply direction had a clash. + * The skb can be handled without packet drop provided the reply direction + * is unique or there the existing entry has the identical tuple in both + * directions. + * + * Caller must hold conntrack table locks to prevent concurrent updates. + * + * Returns NF_DROP if the clash could not be handled. + */ +static int nf_ct_resolve_clash_harder(struct sk_buff *skb, u32 repl_idx) +{ + struct nf_conn *loser_ct = (struct nf_conn *)skb_nfct(skb); + const struct nf_conntrack_zone *zone; + struct nf_conntrack_tuple_hash *h; + struct hlist_nulls_node *n; + struct net *net; + + zone = nf_ct_zone(loser_ct); + net = nf_ct_net(loser_ct); + + /* Reply direction must never result in a clash, unless both origin + * and reply tuples are identical. + */ + hlist_nulls_for_each_entry(h, n, &nf_conntrack_hash[repl_idx], hnnode) { + if (nf_ct_key_equal(h, + &loser_ct->tuplehash[IP_CT_DIR_REPLY].tuple, + zone, net)) + return __nf_ct_resolve_clash(skb, h); + } + + /* We want the clashing entry to go away real soon: 1 second timeout. */ + loser_ct->timeout = nfct_time_stamp + HZ; + + /* IPS_NAT_CLASH removes the entry automatically on the first + * reply. Also prevents UDP tracker from moving the entry to + * ASSURED state, i.e. the entry can always be evicted under + * pressure. + */ + loser_ct->status |= IPS_FIXED_TIMEOUT | IPS_NAT_CLASH; + + __nf_conntrack_insert_prepare(loser_ct); + + /* fake add for ORIGINAL dir: we want lookups to only find the entry + * already in the table. This also hides the clashing entry from + * ctnetlink iteration, i.e. conntrack -L won't show them. + */ + hlist_nulls_add_fake(&loser_ct->tuplehash[IP_CT_DIR_ORIGINAL].hnnode); + + hlist_nulls_add_head_rcu(&loser_ct->tuplehash[IP_CT_DIR_REPLY].hnnode, + &nf_conntrack_hash[repl_idx]); + return NF_ACCEPT; +} + /** * nf_ct_resolve_clash - attempt to handle clash without packet drop * * @skb: skb that causes the clash * @h: tuplehash of the clashing entry already in table + * @hash_reply: hash slot for reply direction * * A conntrack entry can be inserted to the connection tracking table * if there is no existing entry with an identical tuple. @@ -963,10 +1023,18 @@ static int __nf_ct_resolve_clash(struct sk_buff *skb, * exactly the same, only the to-be-confirmed conntrack entry is discarded * and @skb is associated with the conntrack entry already in the table. * + * Failing that, the new, unconfirmed conntrack is still added to the table + * provided that the collision only occurs in the ORIGINAL direction. + * The new entry will be added after the existing one in the hash list, + * so packets in the ORIGINAL direction will continue to match the existing + * entry. The new entry will also have a fixed timeout so it expires -- + * due to the collision, it will not see bidirectional traffic. + * * Returns NF_DROP if the clash could not be resolved. */ static __cold noinline int -nf_ct_resolve_clash(struct sk_buff *skb, struct nf_conntrack_tuple_hash *h) +nf_ct_resolve_clash(struct sk_buff *skb, struct nf_conntrack_tuple_hash *h, + u32 reply_hash) { /* This is the conntrack entry already in hashes that won race. */ struct nf_conn *ct = nf_ct_tuplehash_to_ctrack(h); @@ -987,6 +1055,10 @@ nf_ct_resolve_clash(struct sk_buff *skb, struct nf_conntrack_tuple_hash *h) if (ret == NF_ACCEPT) return ret; + ret = nf_ct_resolve_clash_harder(skb, reply_hash); + if (ret == NF_ACCEPT) + return ret; + drop: nf_ct_add_to_dying_list(loser_ct); NF_CT_STAT_INC(net, drop); @@ -1101,7 +1173,7 @@ __nf_conntrack_confirm(struct sk_buff *skb) return NF_ACCEPT; out: - ret = nf_ct_resolve_clash(skb, h); + ret = nf_ct_resolve_clash(skb, h, reply_hash); dying: nf_conntrack_double_unlock(hash, reply_hash); local_bh_enable(); diff --git a/net/netfilter/nf_conntrack_proto_udp.c b/net/netfilter/nf_conntrack_proto_udp.c index 7365b43f8f98..760ca2422816 100644 --- a/net/netfilter/nf_conntrack_proto_udp.c +++ b/net/netfilter/nf_conntrack_proto_udp.c @@ -81,6 +81,18 @@ static bool udp_error(struct sk_buff *skb, return false; } +static void nf_conntrack_udp_refresh_unreplied(struct nf_conn *ct, + struct sk_buff *skb, + enum ip_conntrack_info ctinfo, + u32 extra_jiffies) +{ + if (unlikely(ctinfo == IP_CT_ESTABLISHED_REPLY && + ct->status & IPS_NAT_CLASH)) + nf_ct_kill(ct); + else + nf_ct_refresh_acct(ct, ctinfo, skb, extra_jiffies); +} + /* Returns verdict for packet, and may modify conntracktype */ int nf_conntrack_udp_packet(struct nf_conn *ct, struct sk_buff *skb, @@ -116,8 +128,8 @@ int nf_conntrack_udp_packet(struct nf_conn *ct, if (!test_and_set_bit(IPS_ASSURED_BIT, &ct->status)) nf_conntrack_event_cache(IPCT_ASSURED, ct); } else { - nf_ct_refresh_acct(ct, ctinfo, skb, - timeouts[UDP_CT_UNREPLIED]); + nf_conntrack_udp_refresh_unreplied(ct, skb, ctinfo, + timeouts[UDP_CT_UNREPLIED]); } return NF_ACCEPT; } @@ -198,8 +210,8 @@ int nf_conntrack_udplite_packet(struct nf_conn *ct, if (!test_and_set_bit(IPS_ASSURED_BIT, &ct->status)) nf_conntrack_event_cache(IPCT_ASSURED, ct); } else { - nf_ct_refresh_acct(ct, ctinfo, skb, - timeouts[UDP_CT_UNREPLIED]); + nf_conntrack_udp_refresh_unreplied(ct, skb, ctinfo, + timeouts[UDP_CT_UNREPLIED]); } return NF_ACCEPT; } -- cgit v1.2.3-59-g8ed1b From bd97ad51a7eb1b02049deca56bc26d96cabbac8a Mon Sep 17 00:00:00 2001 From: Stefano Brivio Date: Fri, 14 Feb 2020 18:14:13 +0100 Subject: netfilter: nft_set_pipapo: Fix mapping table example in comments In both insertion and lookup examples, the two element pointers of rule mapping tables were swapped. Fix that. Reported-by: Pablo Neira Ayuso Fixes: 3c4287f62044 ("nf_tables: Add set type for arbitrary concatenation of ranges") Signed-off-by: Stefano Brivio Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nft_set_pipapo.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'net') diff --git a/net/netfilter/nft_set_pipapo.c b/net/netfilter/nft_set_pipapo.c index f0cb1e13af50..579600b39f39 100644 --- a/net/netfilter/nft_set_pipapo.c +++ b/net/netfilter/nft_set_pipapo.c @@ -203,7 +203,7 @@ * :: * * rule indices in last field: 0 1 - * map to elements: 0x42 0x66 + * map to elements: 0x66 0x42 * * * Matching @@ -298,7 +298,7 @@ * :: * * rule indices in last field: 0 1 - * map to elements: 0x42 0x66 + * map to elements: 0x66 0x42 * * the matching element is at 0x42. * -- cgit v1.2.3-59-g8ed1b From 9a7712048f9d43da5022e75eca3d6b81080e76d3 Mon Sep 17 00:00:00 2001 From: Stefano Brivio Date: Fri, 14 Feb 2020 18:14:14 +0100 Subject: netfilter: nft_set_pipapo: Don't abuse unlikely() in pipapo_refill() I originally used unlikely() in the if (match_only) clause, which we hit on the mapping table for the last field in a set, to ensure we avoid branching to the rest of for loop body, which is executed more frequently. However, Pablo reports, this is confusing as it gives the impression that this is not a common case, and it's actually not the intended usage of unlikely(). I couldn't observe any statistical difference in matching rates on x864_64 and aarch64 without it, so just drop it. Reported-by: Pablo Neira Ayuso Fixes: 3c4287f62044 ("nf_tables: Add set type for arbitrary concatenation of ranges") Signed-off-by: Stefano Brivio Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nft_set_pipapo.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net') diff --git a/net/netfilter/nft_set_pipapo.c b/net/netfilter/nft_set_pipapo.c index 579600b39f39..feac8553f6d9 100644 --- a/net/netfilter/nft_set_pipapo.c +++ b/net/netfilter/nft_set_pipapo.c @@ -503,7 +503,7 @@ static int pipapo_refill(unsigned long *map, int len, int rules, return -1; } - if (unlikely(match_only)) { + if (match_only) { bitmap_clear(map, i, 1); return i; } -- cgit v1.2.3-59-g8ed1b