From 52d1aa8b8249ff477aaa38b6f74a8ced780d079c Mon Sep 17 00:00:00 2001 From: Daniel Xu Date: Wed, 9 Nov 2022 12:39:07 -0700 Subject: netfilter: conntrack: Fix data-races around ct mark nf_conn:mark can be read from and written to in parallel. Use READ_ONCE()/WRITE_ONCE() for reads and writes to prevent unwanted compiler optimizations. Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2") Signed-off-by: Daniel Xu Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_conntrack_core.c | 2 +- net/netfilter/nf_conntrack_netlink.c | 24 ++++++++++++++---------- net/netfilter/nf_conntrack_standalone.c | 2 +- net/netfilter/nft_ct.c | 6 +++--- net/netfilter/xt_connmark.c | 18 ++++++++++-------- 5 files changed, 29 insertions(+), 23 deletions(-) (limited to 'net/netfilter') diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c index f97bda06d2a9..2692139ce417 100644 --- a/net/netfilter/nf_conntrack_core.c +++ b/net/netfilter/nf_conntrack_core.c @@ -1781,7 +1781,7 @@ init_conntrack(struct net *net, struct nf_conn *tmpl, } #ifdef CONFIG_NF_CONNTRACK_MARK - ct->mark = exp->master->mark; + ct->mark = READ_ONCE(exp->master->mark); #endif #ifdef CONFIG_NF_CONNTRACK_SECMARK ct->secmark = exp->master->secmark; diff --git a/net/netfilter/nf_conntrack_netlink.c b/net/netfilter/nf_conntrack_netlink.c index 7562b215b932..d71150a40fb0 100644 --- a/net/netfilter/nf_conntrack_netlink.c +++ b/net/netfilter/nf_conntrack_netlink.c @@ -328,9 +328,9 @@ nla_put_failure: } #ifdef CONFIG_NF_CONNTRACK_MARK -static int ctnetlink_dump_mark(struct sk_buff *skb, const struct nf_conn *ct) +static int ctnetlink_dump_mark(struct sk_buff *skb, u32 mark) { - if (nla_put_be32(skb, CTA_MARK, htonl(ct->mark))) + if (nla_put_be32(skb, CTA_MARK, htonl(mark))) goto nla_put_failure; return 0; @@ -543,7 +543,7 @@ static int ctnetlink_dump_extinfo(struct sk_buff *skb, static int ctnetlink_dump_info(struct sk_buff *skb, struct nf_conn *ct) { if (ctnetlink_dump_status(skb, ct) < 0 || - ctnetlink_dump_mark(skb, ct) < 0 || + ctnetlink_dump_mark(skb, READ_ONCE(ct->mark)) < 0 || ctnetlink_dump_secctx(skb, ct) < 0 || ctnetlink_dump_id(skb, ct) < 0 || ctnetlink_dump_use(skb, ct) < 0 || @@ -722,6 +722,7 @@ ctnetlink_conntrack_event(unsigned int events, const struct nf_ct_event *item) struct sk_buff *skb; unsigned int type; unsigned int flags = 0, group; + u32 mark; int err; if (events & (1 << IPCT_DESTROY)) { @@ -826,8 +827,9 @@ ctnetlink_conntrack_event(unsigned int events, const struct nf_ct_event *item) } #ifdef CONFIG_NF_CONNTRACK_MARK - if ((events & (1 << IPCT_MARK) || ct->mark) - && ctnetlink_dump_mark(skb, ct) < 0) + mark = READ_ONCE(ct->mark); + if ((events & (1 << IPCT_MARK) || mark) && + ctnetlink_dump_mark(skb, mark) < 0) goto nla_put_failure; #endif nlmsg_end(skb, nlh); @@ -1154,7 +1156,7 @@ static int ctnetlink_filter_match(struct nf_conn *ct, void *data) } #ifdef CONFIG_NF_CONNTRACK_MARK - if ((ct->mark & filter->mark.mask) != filter->mark.val) + if ((READ_ONCE(ct->mark) & filter->mark.mask) != filter->mark.val) goto ignore_entry; #endif status = (u32)READ_ONCE(ct->status); @@ -2002,9 +2004,9 @@ static void ctnetlink_change_mark(struct nf_conn *ct, mask = ~ntohl(nla_get_be32(cda[CTA_MARK_MASK])); mark = ntohl(nla_get_be32(cda[CTA_MARK])); - newmark = (ct->mark & mask) ^ mark; - if (newmark != ct->mark) - ct->mark = newmark; + newmark = (READ_ONCE(ct->mark) & mask) ^ mark; + if (newmark != READ_ONCE(ct->mark)) + WRITE_ONCE(ct->mark, newmark); } #endif @@ -2669,6 +2671,7 @@ static int __ctnetlink_glue_build(struct sk_buff *skb, struct nf_conn *ct) { const struct nf_conntrack_zone *zone; struct nlattr *nest_parms; + u32 mark; zone = nf_ct_zone(ct); @@ -2730,7 +2733,8 @@ static int __ctnetlink_glue_build(struct sk_buff *skb, struct nf_conn *ct) goto nla_put_failure; #ifdef CONFIG_NF_CONNTRACK_MARK - if (ct->mark && ctnetlink_dump_mark(skb, ct) < 0) + mark = READ_ONCE(ct->mark); + if (mark && ctnetlink_dump_mark(skb, mark) < 0) goto nla_put_failure; #endif if (ctnetlink_dump_labels(skb, ct) < 0) diff --git a/net/netfilter/nf_conntrack_standalone.c b/net/netfilter/nf_conntrack_standalone.c index 4ffe84c5a82c..bca839ab1ae8 100644 --- a/net/netfilter/nf_conntrack_standalone.c +++ b/net/netfilter/nf_conntrack_standalone.c @@ -366,7 +366,7 @@ static int ct_seq_show(struct seq_file *s, void *v) goto release; #if defined(CONFIG_NF_CONNTRACK_MARK) - seq_printf(s, "mark=%u ", ct->mark); + seq_printf(s, "mark=%u ", READ_ONCE(ct->mark)); #endif ct_show_secctx(s, ct); diff --git a/net/netfilter/nft_ct.c b/net/netfilter/nft_ct.c index a3f01f209a53..641dc21f92b4 100644 --- a/net/netfilter/nft_ct.c +++ b/net/netfilter/nft_ct.c @@ -98,7 +98,7 @@ static void nft_ct_get_eval(const struct nft_expr *expr, return; #ifdef CONFIG_NF_CONNTRACK_MARK case NFT_CT_MARK: - *dest = ct->mark; + *dest = READ_ONCE(ct->mark); return; #endif #ifdef CONFIG_NF_CONNTRACK_SECMARK @@ -297,8 +297,8 @@ static void nft_ct_set_eval(const struct nft_expr *expr, switch (priv->key) { #ifdef CONFIG_NF_CONNTRACK_MARK case NFT_CT_MARK: - if (ct->mark != value) { - ct->mark = value; + if (READ_ONCE(ct->mark) != value) { + WRITE_ONCE(ct->mark, value); nf_conntrack_event_cache(IPCT_MARK, ct); } break; diff --git a/net/netfilter/xt_connmark.c b/net/netfilter/xt_connmark.c index e5ebc0810675..ad3c033db64e 100644 --- a/net/netfilter/xt_connmark.c +++ b/net/netfilter/xt_connmark.c @@ -30,6 +30,7 @@ connmark_tg_shift(struct sk_buff *skb, const struct xt_connmark_tginfo2 *info) u_int32_t new_targetmark; struct nf_conn *ct; u_int32_t newmark; + u_int32_t oldmark; ct = nf_ct_get(skb, &ctinfo); if (ct == NULL) @@ -37,14 +38,15 @@ connmark_tg_shift(struct sk_buff *skb, const struct xt_connmark_tginfo2 *info) switch (info->mode) { case XT_CONNMARK_SET: - newmark = (ct->mark & ~info->ctmask) ^ info->ctmark; + oldmark = READ_ONCE(ct->mark); + newmark = (oldmark & ~info->ctmask) ^ info->ctmark; if (info->shift_dir == D_SHIFT_RIGHT) newmark >>= info->shift_bits; else newmark <<= info->shift_bits; - if (ct->mark != newmark) { - ct->mark = newmark; + if (READ_ONCE(ct->mark) != newmark) { + WRITE_ONCE(ct->mark, newmark); nf_conntrack_event_cache(IPCT_MARK, ct); } break; @@ -55,15 +57,15 @@ connmark_tg_shift(struct sk_buff *skb, const struct xt_connmark_tginfo2 *info) else new_targetmark <<= info->shift_bits; - newmark = (ct->mark & ~info->ctmask) ^ + newmark = (READ_ONCE(ct->mark) & ~info->ctmask) ^ new_targetmark; - if (ct->mark != newmark) { - ct->mark = newmark; + if (READ_ONCE(ct->mark) != newmark) { + WRITE_ONCE(ct->mark, newmark); nf_conntrack_event_cache(IPCT_MARK, ct); } break; case XT_CONNMARK_RESTORE: - new_targetmark = (ct->mark & info->ctmask); + new_targetmark = (READ_ONCE(ct->mark) & info->ctmask); if (info->shift_dir == D_SHIFT_RIGHT) new_targetmark >>= info->shift_bits; else @@ -126,7 +128,7 @@ connmark_mt(const struct sk_buff *skb, struct xt_action_param *par) if (ct == NULL) return false; - return ((ct->mark & info->mask) == info->mark) ^ info->invert; + return ((READ_ONCE(ct->mark) & info->mask) == info->mark) ^ info->invert; } static int connmark_mt_check(const struct xt_mtchk_param *par) -- cgit v1.2.3 From 33c7aba0b4ffd6d7cdab862a034eb582a5120a38 Mon Sep 17 00:00:00 2001 From: Pablo Neira Ayuso Date: Mon, 14 Nov 2022 11:31:54 +0100 Subject: netfilter: nf_tables: do not set up extensions for end interval Elements with an end interval flag set on do not store extensions. The global set definition is currently setting on the timeout and stateful expression for end interval elements. This leads to skipping end interval elements from the set->ops->walk() path as the expired check bogusly reports true. Moreover, do not set up stateful expressions for elements with end interval flag set on since this is never used. Fixes: 65038428b2c6 ("netfilter: nf_tables: allow to specify stateful expression in set definition") Fixes: 8d8540c4f5e0 ("netfilter: nft_set_rbtree: add timeout support") Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_tables_api.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'net/netfilter') diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c index e7152d599d73..7a09421f19e1 100644 --- a/net/netfilter/nf_tables_api.c +++ b/net/netfilter/nf_tables_api.c @@ -5958,7 +5958,8 @@ static int nft_add_set_elem(struct nft_ctx *ctx, struct nft_set *set, &timeout); if (err) return err; - } else if (set->flags & NFT_SET_TIMEOUT) { + } else if (set->flags & NFT_SET_TIMEOUT && + !(flags & NFT_SET_ELEM_INTERVAL_END)) { timeout = set->timeout; } @@ -6024,7 +6025,8 @@ static int nft_add_set_elem(struct nft_ctx *ctx, struct nft_set *set, err = -EOPNOTSUPP; goto err_set_elem_expr; } - } else if (set->num_exprs > 0) { + } else if (set->num_exprs > 0 && + !(flags & NFT_SET_ELEM_INTERVAL_END)) { err = nft_set_elem_expr_clone(ctx, set, expr_array); if (err < 0) goto err_set_elem_expr_clone; -- cgit v1.2.3 From c7aa1a76d4a0a3c401025b60c401412bbb60f8c6 Mon Sep 17 00:00:00 2001 From: Vishwanath Pai Date: Wed, 28 Sep 2022 14:26:50 -0400 Subject: netfilter: ipset: regression in ip_set_hash_ip.c This patch introduced a regression: commit 48596a8ddc46 ("netfilter: ipset: Fix adding an IPv4 range containing more than 2^31 addresses") The variable e.ip is passed to adtfn() function which finally adds the ip address to the set. The patch above refactored the for loop and moved e.ip = htonl(ip) to the end of the for loop. What this means is that if the value of "ip" changes between the first assignement of e.ip and the forloop, then e.ip is pointing to a different ip address than "ip". Test case: $ ipset create jdtest_tmp hash:ip family inet hashsize 2048 maxelem 100000 $ ipset add jdtest_tmp 10.0.1.1/31 ipset v6.21.1: Element cannot be added to the set: it's already added The value of ip gets updated inside the "else if (tb[IPSET_ATTR_CIDR])" block but e.ip is still pointing to the old value. Fixes: 48596a8ddc46 ("netfilter: ipset: Fix adding an IPv4 range containing more than 2^31 addresses") Reviewed-by: Joshua Hunt Signed-off-by: Vishwanath Pai Signed-off-by: Pablo Neira Ayuso --- net/netfilter/ipset/ip_set_hash_ip.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) (limited to 'net/netfilter') diff --git a/net/netfilter/ipset/ip_set_hash_ip.c b/net/netfilter/ipset/ip_set_hash_ip.c index dd30c03d5a23..75d556d71652 100644 --- a/net/netfilter/ipset/ip_set_hash_ip.c +++ b/net/netfilter/ipset/ip_set_hash_ip.c @@ -151,18 +151,16 @@ hash_ip4_uadt(struct ip_set *set, struct nlattr *tb[], if (((u64)ip_to - ip + 1) >> (32 - h->netmask) > IPSET_MAX_RANGE) return -ERANGE; - if (retried) { + if (retried) ip = ntohl(h->next.ip); - e.ip = htonl(ip); - } for (; ip <= ip_to;) { + e.ip = htonl(ip); ret = adtfn(set, &e, &ext, &ext, flags); if (ret && !ip_set_eexist(ret, flags)) return ret; ip += hosts; - e.ip = htonl(ip); - if (e.ip == 0) + if (ip == 0) return 0; ret = 0; -- cgit v1.2.3 From 6a66ce44a51bdfc47721f0c591137df2d4b21247 Mon Sep 17 00:00:00 2001 From: Jozsef Kadlecsik Date: Tue, 22 Nov 2022 20:18:58 +0100 Subject: netfilter: ipset: restore allowing 64 clashing elements in hash:net,iface The commit 510841da1fcc ("netfilter: ipset: enforce documented limit to prevent allocating huge memory") was too strict and prevented to add up to 64 clashing elements to a hash:net,iface type of set. This patch fixes the issue and now the type behaves as documented. Fixes: 510841da1fcc ("netfilter: ipset: enforce documented limit to prevent allocating huge memory") Signed-off-by: Jozsef Kadlecsik Signed-off-by: Pablo Neira Ayuso --- net/netfilter/ipset/ip_set_hash_gen.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net/netfilter') diff --git a/net/netfilter/ipset/ip_set_hash_gen.h b/net/netfilter/ipset/ip_set_hash_gen.h index 3adc291d9ce1..7499192af586 100644 --- a/net/netfilter/ipset/ip_set_hash_gen.h +++ b/net/netfilter/ipset/ip_set_hash_gen.h @@ -916,7 +916,7 @@ mtype_add(struct ip_set *set, void *value, const struct ip_set_ext *ext, #ifdef IP_SET_HASH_WITH_MULTI if (h->bucketsize >= AHASH_MAX_TUNED) goto set_full; - else if (h->bucketsize < multi) + else if (h->bucketsize <= multi) h->bucketsize += AHASH_INIT_SIZE; #endif if (n->size >= AHASH_MAX(h)) { -- cgit v1.2.3 From bcd9e3c1656d0f7dd9743598c65c3ae24efb38d0 Mon Sep 17 00:00:00 2001 From: Felix Fietkau Date: Mon, 21 Nov 2022 19:26:15 +0100 Subject: netfilter: flowtable_offload: add missing locking nf_flow_table_block_setup and the driver TC_SETUP_FT call can modify the flow block cb list while they are being traversed elsewhere, causing a crash. Add a write lock around the calls to protect readers Fixes: c29f74e0df7a ("netfilter: nf_flow_table: hardware offload support") Reported-by: Chad Monroe Signed-off-by: Felix Fietkau Signed-off-by: Pablo Neira Ayuso --- net/netfilter/nf_flow_table_offload.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'net/netfilter') diff --git a/net/netfilter/nf_flow_table_offload.c b/net/netfilter/nf_flow_table_offload.c index b04645ced89b..00b522890d77 100644 --- a/net/netfilter/nf_flow_table_offload.c +++ b/net/netfilter/nf_flow_table_offload.c @@ -1098,6 +1098,7 @@ static int nf_flow_table_block_setup(struct nf_flowtable *flowtable, struct flow_block_cb *block_cb, *next; int err = 0; + down_write(&flowtable->flow_block_lock); switch (cmd) { case FLOW_BLOCK_BIND: list_splice(&bo->cb_list, &flowtable->flow_block.cb_list); @@ -1112,6 +1113,7 @@ static int nf_flow_table_block_setup(struct nf_flowtable *flowtable, WARN_ON_ONCE(1); err = -EOPNOTSUPP; } + up_write(&flowtable->flow_block_lock); return err; } @@ -1168,7 +1170,9 @@ static int nf_flow_table_offload_cmd(struct flow_block_offload *bo, nf_flow_table_block_offload_init(bo, dev_net(dev), cmd, flowtable, extack); + down_write(&flowtable->flow_block_lock); err = dev->netdev_ops->ndo_setup_tc(dev, TC_SETUP_FT, bo); + up_write(&flowtable->flow_block_lock); if (err < 0) return err; -- cgit v1.2.3