From 719774377622bc4025d2a74f551b5dc2158c6c30 Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Fri, 7 Jan 2022 05:03:22 +0100 Subject: netfilter: conntrack: convert to refcount_t api Convert nf_conn reference counting from atomic_t to refcount_t based api. refcount_t api provides more runtime sanity checks and will warn on certain constructs, e.g. refcount_inc() on a zero reference count, which usually indicates use-after-free. For this reason template allocation is changed to init the refcount to 1, the subsequenct add operations are removed. Likewise, init_conntrack() is changed to set the initial refcount to 1 instead refcount_inc(). This is safe because the new entry is not (yet) visible to other cpus. Signed-off-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- include/linux/netfilter/nf_conntrack_common.h | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'include/linux') diff --git a/include/linux/netfilter/nf_conntrack_common.h b/include/linux/netfilter/nf_conntrack_common.h index 700ea077ce2d..a03f7a80b9ab 100644 --- a/include/linux/netfilter/nf_conntrack_common.h +++ b/include/linux/netfilter/nf_conntrack_common.h @@ -2,7 +2,7 @@ #ifndef _NF_CONNTRACK_COMMON_H #define _NF_CONNTRACK_COMMON_H -#include +#include #include struct ip_conntrack_stat { @@ -25,19 +25,19 @@ struct ip_conntrack_stat { #define NFCT_PTRMASK ~(NFCT_INFOMASK) struct nf_conntrack { - atomic_t use; + refcount_t use; }; void nf_conntrack_destroy(struct nf_conntrack *nfct); static inline void nf_conntrack_put(struct nf_conntrack *nfct) { - if (nfct && atomic_dec_and_test(&nfct->use)) + if (nfct && refcount_dec_and_test(&nfct->use)) nf_conntrack_destroy(nfct); } static inline void nf_conntrack_get(struct nf_conntrack *nfct) { if (nfct) - atomic_inc(&nfct->use); + refcount_inc(&nfct->use); } #endif /* _NF_CONNTRACK_COMMON_H */ -- cgit v1.2.3 From 3fce16493dc1aa2c9af3d7e7bd360dfe203a3e6a Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Fri, 7 Jan 2022 05:03:23 +0100 Subject: netfilter: core: move ip_ct_attach indirection to struct nf_ct_hook ip_ct_attach predates struct nf_ct_hook, we can place it there and remove the exported symbol. Signed-off-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- include/linux/netfilter.h | 2 +- net/netfilter/core.c | 19 ++++++++----------- net/netfilter/nf_conntrack_core.c | 4 +--- 3 files changed, 10 insertions(+), 15 deletions(-) (limited to 'include/linux') diff --git a/include/linux/netfilter.h b/include/linux/netfilter.h index 3fda1a508733..e0e3f3355ab1 100644 --- a/include/linux/netfilter.h +++ b/include/linux/netfilter.h @@ -440,7 +440,6 @@ nf_nat_decode_session(struct sk_buff *skb, struct flowi *fl, u_int8_t family) #if IS_ENABLED(CONFIG_NF_CONNTRACK) #include -extern void (*ip_ct_attach)(struct sk_buff *, const struct sk_buff *) __rcu; void nf_ct_attach(struct sk_buff *, const struct sk_buff *); struct nf_conntrack_tuple; bool nf_ct_get_tuple_skb(struct nf_conntrack_tuple *dst_tuple, @@ -463,6 +462,7 @@ struct nf_ct_hook { void (*destroy)(struct nf_conntrack *); bool (*get_tuple_skb)(struct nf_conntrack_tuple *, const struct sk_buff *); + void (*attach)(struct sk_buff *nskb, const struct sk_buff *skb); }; extern struct nf_ct_hook __rcu *nf_ct_hook; diff --git a/net/netfilter/core.c b/net/netfilter/core.c index 6dec9cd395f1..dc806dc9fe28 100644 --- a/net/netfilter/core.c +++ b/net/netfilter/core.c @@ -673,25 +673,22 @@ struct nf_ct_hook __rcu *nf_ct_hook __read_mostly; EXPORT_SYMBOL_GPL(nf_ct_hook); #if IS_ENABLED(CONFIG_NF_CONNTRACK) -/* This does not belong here, but locally generated errors need it if connection - tracking in use: without this, connection may not be in hash table, and hence - manufactured ICMP or RST packets will not be associated with it. */ -void (*ip_ct_attach)(struct sk_buff *, const struct sk_buff *) - __rcu __read_mostly; -EXPORT_SYMBOL(ip_ct_attach); - struct nf_nat_hook __rcu *nf_nat_hook __read_mostly; EXPORT_SYMBOL_GPL(nf_nat_hook); +/* This does not belong here, but locally generated errors need it if connection + * tracking in use: without this, connection may not be in hash table, and hence + * manufactured ICMP or RST packets will not be associated with it. + */ void nf_ct_attach(struct sk_buff *new, const struct sk_buff *skb) { - void (*attach)(struct sk_buff *, const struct sk_buff *); + const struct nf_ct_hook *ct_hook; if (skb->_nfct) { rcu_read_lock(); - attach = rcu_dereference(ip_ct_attach); - if (attach) - attach(new, skb); + ct_hook = rcu_dereference(nf_ct_hook); + if (ct_hook) + ct_hook->attach(new, skb); rcu_read_unlock(); } } diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c index 328d359fcabe..89f1e5f0090b 100644 --- a/net/netfilter/nf_conntrack_core.c +++ b/net/netfilter/nf_conntrack_core.c @@ -2455,7 +2455,6 @@ static int kill_all(struct nf_conn *i, void *data) void nf_conntrack_cleanup_start(void) { conntrack_gc_work.exiting = true; - RCU_INIT_POINTER(ip_ct_attach, NULL); } void nf_conntrack_cleanup_end(void) @@ -2774,12 +2773,11 @@ static struct nf_ct_hook nf_conntrack_hook = { .update = nf_conntrack_update, .destroy = destroy_conntrack, .get_tuple_skb = nf_conntrack_get_tuple_skb, + .attach = nf_conntrack_attach, }; void nf_conntrack_init_end(void) { - /* For use by REJECT target */ - RCU_INIT_POINTER(ip_ct_attach, nf_conntrack_attach); RCU_INIT_POINTER(nf_ct_hook, &nf_conntrack_hook); } -- cgit v1.2.3 From 285c8a7a58158cb1805c97ff03875df2ba2ea1fe Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Fri, 7 Jan 2022 05:03:24 +0100 Subject: netfilter: make function op structures const No functional changes, these structures should be const. Signed-off-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- include/linux/netfilter.h | 8 ++++---- net/netfilter/core.c | 10 +++++----- net/netfilter/nf_conntrack_core.c | 4 ++-- net/netfilter/nf_conntrack_netlink.c | 4 ++-- net/netfilter/nf_nat_core.c | 2 +- net/netfilter/nfnetlink_queue.c | 8 ++++---- 6 files changed, 18 insertions(+), 18 deletions(-) (limited to 'include/linux') diff --git a/include/linux/netfilter.h b/include/linux/netfilter.h index e0e3f3355ab1..15e71bfff726 100644 --- a/include/linux/netfilter.h +++ b/include/linux/netfilter.h @@ -381,13 +381,13 @@ struct nf_nat_hook { enum ip_conntrack_dir dir); }; -extern struct nf_nat_hook __rcu *nf_nat_hook; +extern const struct nf_nat_hook __rcu *nf_nat_hook; static inline void nf_nat_decode_session(struct sk_buff *skb, struct flowi *fl, u_int8_t family) { #if IS_ENABLED(CONFIG_NF_NAT) - struct nf_nat_hook *nat_hook; + const struct nf_nat_hook *nat_hook; rcu_read_lock(); nat_hook = rcu_dereference(nf_nat_hook); @@ -464,7 +464,7 @@ struct nf_ct_hook { const struct sk_buff *); void (*attach)(struct sk_buff *nskb, const struct sk_buff *skb); }; -extern struct nf_ct_hook __rcu *nf_ct_hook; +extern const struct nf_ct_hook __rcu *nf_ct_hook; struct nlattr; @@ -479,7 +479,7 @@ struct nfnl_ct_hook { void (*seq_adjust)(struct sk_buff *skb, struct nf_conn *ct, enum ip_conntrack_info ctinfo, s32 off); }; -extern struct nfnl_ct_hook __rcu *nfnl_ct_hook; +extern const struct nfnl_ct_hook __rcu *nfnl_ct_hook; /** * nf_skb_duplicated - TEE target has sent a packet diff --git a/net/netfilter/core.c b/net/netfilter/core.c index dc806dc9fe28..354cb472f386 100644 --- a/net/netfilter/core.c +++ b/net/netfilter/core.c @@ -666,14 +666,14 @@ EXPORT_SYMBOL(nf_hook_slow_list); /* This needs to be compiled in any case to avoid dependencies between the * nfnetlink_queue code and nf_conntrack. */ -struct nfnl_ct_hook __rcu *nfnl_ct_hook __read_mostly; +const struct nfnl_ct_hook __rcu *nfnl_ct_hook __read_mostly; EXPORT_SYMBOL_GPL(nfnl_ct_hook); -struct nf_ct_hook __rcu *nf_ct_hook __read_mostly; +const struct nf_ct_hook __rcu *nf_ct_hook __read_mostly; EXPORT_SYMBOL_GPL(nf_ct_hook); #if IS_ENABLED(CONFIG_NF_CONNTRACK) -struct nf_nat_hook __rcu *nf_nat_hook __read_mostly; +const struct nf_nat_hook __rcu *nf_nat_hook __read_mostly; EXPORT_SYMBOL_GPL(nf_nat_hook); /* This does not belong here, but locally generated errors need it if connection @@ -696,7 +696,7 @@ EXPORT_SYMBOL(nf_ct_attach); void nf_conntrack_destroy(struct nf_conntrack *nfct) { - struct nf_ct_hook *ct_hook; + const struct nf_ct_hook *ct_hook; rcu_read_lock(); ct_hook = rcu_dereference(nf_ct_hook); @@ -709,7 +709,7 @@ EXPORT_SYMBOL(nf_conntrack_destroy); bool nf_ct_get_tuple_skb(struct nf_conntrack_tuple *dst_tuple, const struct sk_buff *skb) { - struct nf_ct_hook *ct_hook; + const struct nf_ct_hook *ct_hook; bool ret = false; rcu_read_lock(); diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c index 89f1e5f0090b..cd3d07e418b5 100644 --- a/net/netfilter/nf_conntrack_core.c +++ b/net/netfilter/nf_conntrack_core.c @@ -2085,9 +2085,9 @@ static int __nf_conntrack_update(struct net *net, struct sk_buff *skb, struct nf_conn *ct, enum ip_conntrack_info ctinfo) { + const struct nf_nat_hook *nat_hook; struct nf_conntrack_tuple_hash *h; struct nf_conntrack_tuple tuple; - struct nf_nat_hook *nat_hook; unsigned int status; int dataoff; u16 l3num; @@ -2769,7 +2769,7 @@ err_cachep: return ret; } -static struct nf_ct_hook nf_conntrack_hook = { +static const struct nf_ct_hook nf_conntrack_hook = { .update = nf_conntrack_update, .destroy = destroy_conntrack, .get_tuple_skb = nf_conntrack_get_tuple_skb, diff --git a/net/netfilter/nf_conntrack_netlink.c b/net/netfilter/nf_conntrack_netlink.c index 13e2e0390c77..01cc69ab0b4e 100644 --- a/net/netfilter/nf_conntrack_netlink.c +++ b/net/netfilter/nf_conntrack_netlink.c @@ -1819,7 +1819,7 @@ ctnetlink_parse_nat_setup(struct nf_conn *ct, const struct nlattr *attr) __must_hold(RCU) { - struct nf_nat_hook *nat_hook; + const struct nf_nat_hook *nat_hook; int err; nat_hook = rcu_dereference(nf_nat_hook); @@ -2921,7 +2921,7 @@ static void ctnetlink_glue_seqadj(struct sk_buff *skb, struct nf_conn *ct, nf_ct_tcp_seqadj_set(skb, ct, ctinfo, diff); } -static struct nfnl_ct_hook ctnetlink_glue_hook = { +static const struct nfnl_ct_hook ctnetlink_glue_hook = { .build_size = ctnetlink_glue_build_size, .build = ctnetlink_glue_build, .parse = ctnetlink_glue_parse, diff --git a/net/netfilter/nf_nat_core.c b/net/netfilter/nf_nat_core.c index 3dd130487b5b..2d06a66899b2 100644 --- a/net/netfilter/nf_nat_core.c +++ b/net/netfilter/nf_nat_core.c @@ -1167,7 +1167,7 @@ static struct pernet_operations nat_net_ops = { .size = sizeof(struct nat_net), }; -static struct nf_nat_hook nat_hook = { +static const struct nf_nat_hook nat_hook = { .parse_nat_setup = nfnetlink_parse_nat_setup, #ifdef CONFIG_XFRM .decode_session = __nf_nat_decode_session, diff --git a/net/netfilter/nfnetlink_queue.c b/net/netfilter/nfnetlink_queue.c index 08771f47d469..862d8a3a0911 100644 --- a/net/netfilter/nfnetlink_queue.c +++ b/net/netfilter/nfnetlink_queue.c @@ -225,7 +225,7 @@ find_dequeue_entry(struct nfqnl_instance *queue, unsigned int id) static void nfqnl_reinject(struct nf_queue_entry *entry, unsigned int verdict) { - struct nf_ct_hook *ct_hook; + const struct nf_ct_hook *ct_hook; int err; if (verdict == NF_ACCEPT || @@ -388,7 +388,7 @@ nfqnl_build_packet_message(struct net *net, struct nfqnl_instance *queue, struct net_device *outdev; struct nf_conn *ct = NULL; enum ip_conntrack_info ctinfo = 0; - struct nfnl_ct_hook *nfnl_ct; + const struct nfnl_ct_hook *nfnl_ct; bool csum_verify; char *secdata = NULL; u32 seclen = 0; @@ -1103,7 +1103,7 @@ static int nfqnl_recv_verdict_batch(struct sk_buff *skb, return 0; } -static struct nf_conn *nfqnl_ct_parse(struct nfnl_ct_hook *nfnl_ct, +static struct nf_conn *nfqnl_ct_parse(const struct nfnl_ct_hook *nfnl_ct, const struct nlmsghdr *nlh, const struct nlattr * const nfqa[], struct nf_queue_entry *entry, @@ -1170,11 +1170,11 @@ static int nfqnl_recv_verdict(struct sk_buff *skb, const struct nfnl_info *info, { struct nfnl_queue_net *q = nfnl_queue_pernet(info->net); u_int16_t queue_num = ntohs(info->nfmsg->res_id); + const struct nfnl_ct_hook *nfnl_ct; struct nfqnl_msg_verdict_hdr *vhdr; enum ip_conntrack_info ctinfo; struct nfqnl_instance *queue; struct nf_queue_entry *entry; - struct nfnl_ct_hook *nfnl_ct; struct nf_conn *ct = NULL; unsigned int verdict; int err; -- cgit v1.2.3 From 6ae7989c9af0d98ab64196f4f4c6f6499454bd23 Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Fri, 7 Jan 2022 05:03:25 +0100 Subject: netfilter: conntrack: avoid useless indirection during conntrack destruction nf_ct_put() results in a usesless indirection: nf_ct_put -> nf_conntrack_put -> nf_conntrack_destroy -> rcu readlock + indirect call of ct_hooks->destroy(). There are two _put helpers: nf_ct_put and nf_conntrack_put. The latter is what should be used in code that MUST NOT cause a linker dependency on the conntrack module (e.g. calls from core network stack). Everyone else should call nf_ct_put() instead. A followup patch will convert a few nf_conntrack_put() calls to nf_ct_put(), in particular from modules that already have a conntrack dependency such as act_ct or even nf_conntrack itself. Signed-off-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- include/linux/netfilter/nf_conntrack_common.h | 2 ++ include/net/netfilter/nf_conntrack.h | 8 ++++++-- net/netfilter/nf_conntrack_core.c | 12 ++++++------ 3 files changed, 14 insertions(+), 8 deletions(-) (limited to 'include/linux') diff --git a/include/linux/netfilter/nf_conntrack_common.h b/include/linux/netfilter/nf_conntrack_common.h index a03f7a80b9ab..2770db2fa080 100644 --- a/include/linux/netfilter/nf_conntrack_common.h +++ b/include/linux/netfilter/nf_conntrack_common.h @@ -29,6 +29,8 @@ struct nf_conntrack { }; void nf_conntrack_destroy(struct nf_conntrack *nfct); + +/* like nf_ct_put, but without module dependency on nf_conntrack */ static inline void nf_conntrack_put(struct nf_conntrack *nfct) { if (nfct && refcount_dec_and_test(&nfct->use)) diff --git a/include/net/netfilter/nf_conntrack.h b/include/net/netfilter/nf_conntrack.h index a4a14f3a5e38..8731d5bcb47d 100644 --- a/include/net/netfilter/nf_conntrack.h +++ b/include/net/netfilter/nf_conntrack.h @@ -76,6 +76,8 @@ struct nf_conn { * Hint, SKB address this struct and refcnt via skb->_nfct and * helpers nf_conntrack_get() and nf_conntrack_put(). * Helper nf_ct_put() equals nf_conntrack_put() by dec refcnt, + * except that the latter uses internal indirection and does not + * result in a conntrack module dependency. * beware nf_ct_get() is different and don't inc refcnt. */ struct nf_conntrack ct_general; @@ -170,11 +172,13 @@ nf_ct_get(const struct sk_buff *skb, enum ip_conntrack_info *ctinfo) return (struct nf_conn *)(nfct & NFCT_PTRMASK); } +void nf_ct_destroy(struct nf_conntrack *nfct); + /* decrement reference count on a conntrack */ static inline void nf_ct_put(struct nf_conn *ct) { - WARN_ON(!ct); - nf_conntrack_put(&ct->ct_general); + if (ct && refcount_dec_and_test(&ct->ct_general.use)) + nf_ct_destroy(&ct->ct_general); } /* Protocol module loading */ diff --git a/net/netfilter/nf_conntrack_core.c b/net/netfilter/nf_conntrack_core.c index cd3d07e418b5..7a2063abae04 100644 --- a/net/netfilter/nf_conntrack_core.c +++ b/net/netfilter/nf_conntrack_core.c @@ -558,7 +558,7 @@ static void nf_ct_del_from_dying_or_unconfirmed_list(struct nf_conn *ct) #define NFCT_ALIGN(len) (((len) + NFCT_INFOMASK) & ~NFCT_INFOMASK) -/* Released via destroy_conntrack() */ +/* Released via nf_ct_destroy() */ struct nf_conn *nf_ct_tmpl_alloc(struct net *net, const struct nf_conntrack_zone *zone, gfp_t flags) @@ -612,12 +612,11 @@ static void destroy_gre_conntrack(struct nf_conn *ct) #endif } -static void -destroy_conntrack(struct nf_conntrack *nfct) +void nf_ct_destroy(struct nf_conntrack *nfct) { struct nf_conn *ct = (struct nf_conn *)nfct; - pr_debug("destroy_conntrack(%p)\n", ct); + pr_debug("%s(%p)\n", __func__, ct); WARN_ON(refcount_read(&nfct->use) != 0); if (unlikely(nf_ct_is_template(ct))) { @@ -643,9 +642,10 @@ destroy_conntrack(struct nf_conntrack *nfct) if (ct->master) nf_ct_put(ct->master); - pr_debug("destroy_conntrack: returning ct=%p to slab\n", ct); + pr_debug("%s: returning ct=%p to slab\n", __func__, ct); nf_conntrack_free(ct); } +EXPORT_SYMBOL(nf_ct_destroy); static void nf_ct_delete_from_lists(struct nf_conn *ct) { @@ -2771,7 +2771,7 @@ err_cachep: static const struct nf_ct_hook nf_conntrack_hook = { .update = nf_conntrack_update, - .destroy = destroy_conntrack, + .destroy = nf_ct_destroy, .get_tuple_skb = nf_conntrack_get_tuple_skb, .attach = nf_conntrack_attach, }; -- cgit v1.2.3 From 6316136ec6e3dd1c302f7e7289a9ee46ecc610ae Mon Sep 17 00:00:00 2001 From: Florian Westphal Date: Fri, 7 Jan 2022 15:46:16 +0100 Subject: netfilter: egress: avoid a lockdep splat include/linux/netfilter_netdev.h:97 suspicious rcu_dereference_check() usage! 2 locks held by sd-resolve/1100: 0: ..(rcu_read_lock_bh){1:3}, at: ip_finish_output2 1: ..(rcu_read_lock_bh){1:3}, at: __dev_queue_xmit __dev_queue_xmit+0 .. The helper has two callers, one uses rcu_read_lock, the other rcu_read_lock_bh(). Annotate the dereference to reflect this. Fixes: 42df6e1d221dd ("netfilter: Introduce egress hook") Signed-off-by: Florian Westphal Signed-off-by: Pablo Neira Ayuso --- include/linux/netfilter_netdev.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'include/linux') diff --git a/include/linux/netfilter_netdev.h b/include/linux/netfilter_netdev.h index b71b57a83bb4..b4dd96e4dc8d 100644 --- a/include/linux/netfilter_netdev.h +++ b/include/linux/netfilter_netdev.h @@ -94,7 +94,7 @@ static inline struct sk_buff *nf_hook_egress(struct sk_buff *skb, int *rc, return skb; #endif - e = rcu_dereference(dev->nf_hooks_egress); + e = rcu_dereference_check(dev->nf_hooks_egress, rcu_read_lock_bh_held()); if (!e) return skb; -- cgit v1.2.3