Merge tag 'nf-next-23-10-10' of https://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf-next

Florian Westphal says:

====================
netfilter updates for next

First 5 patches, from Phil Sutter, clean up nftables dumpers to
use the context buffer in the netlink_callback structure rather
than a kmalloc'd buffer.

Patch 6, from myself, zaps dead code and replaces the helper function
with a small inlined helper.

Patch 7, also from myself, removes another pr_debug and replaces it
with the existing nf_log-based debug helpers.

Last patch, from George Guo, gets nft_table comments back in
sync with the structure members.

* tag 'nf-next-23-10-10' of https://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf-next:
  netfilter: cleanup struct nft_table
  netfilter: conntrack: prefer tcp_error_log to pr_debug
  netfilter: conntrack: simplify nf_conntrack_alter_reply
  netfilter: nf_tables: Don't allocate nft_rule_dump_ctx
  netfilter: nf_tables: Carry s_idx in nft_rule_dump_ctx
  netfilter: nf_tables: Carry reset flag in nft_rule_dump_ctx
  netfilter: nf_tables: Drop pointless memset when dumping rules
  netfilter: nf_tables: Always allocate nft_rule_dump_ctx
====================

Link: https://lore.kernel.org/r/20231010145343.12551-1-fw@strlen.deSigned-off-by: Jakub Kicinski <kuba@kernel.org>

include/net/netfilter/nf_conntrack.h

@@ -160,10 +160,6 @@ static inline struct net *nf_ct_net(const struct nf_conn *ct)
 	return read_pnet(&ct->ct_net);
 }
 
-/* Alter reply tuple (maybe alter helper). */
-void nf_conntrack_alter_reply(struct nf_conn *ct,
-			      const struct nf_conntrack_tuple *newreply);
-
 /* Is this tuple taken? (ignoring any belonging to the given
    conntrack). */
 int nf_conntrack_tuple_taken(const struct nf_conntrack_tuple *tuple,
@@ -284,6 +280,16 @@ static inline bool nf_is_loopback_packet(const struct sk_buff *skb)
 	return skb->dev && skb->skb_iif && skb->dev->flags & IFF_LOOPBACK;
 }
 
+static inline void nf_conntrack_alter_reply(struct nf_conn *ct,
+					    const struct nf_conntrack_tuple *newreply)
+{
+	/* Must be unconfirmed, so not in hash table yet */
+	if (WARN_ON(nf_ct_is_confirmed(ct)))
+		return;
+
+	ct->tuplehash[IP_CT_DIR_REPLY].tuple = *newreply;
+}
+
 #define nfct_time_stamp ((u32)(jiffies))
 
 /* jiffies until ct expires, 0 if already expired */

include/net/netfilter/nf_tables.h

@@ -1198,10 +1198,13 @@ static inline void nft_use_inc_restore(u32 *use)
  *	@hgenerator: handle generator state
  *	@handle: table handle
  *	@use: number of chain references to this table
+ *	@family:address family
  *	@flags: table flag (see enum nft_table_flags)
  *	@genmask: generation mask
- *	@afinfo: address family info
+ *	@nlpid: netlink port ID
  *	@name: name of the table
+ *	@udlen: length of the user data
+ *	@udata: user data
  *	@validate_state: internal, set when transaction adds jumps
  */
 struct nft_table {

net/netfilter/nf_conntrack_core.c

@@ -2042,24 +2042,6 @@ nf_conntrack_in(struct sk_buff *skb, const struct nf_hook_state *state)
 }
 EXPORT_SYMBOL_GPL(nf_conntrack_in);
 
-/* Alter reply tuple (maybe alter helper).  This is for NAT, and is
-   implicitly racy: see __nf_conntrack_confirm */
-void nf_conntrack_alter_reply(struct nf_conn *ct,
-			      const struct nf_conntrack_tuple *newreply)
-{
-	struct nf_conn_help *help = nfct_help(ct);
-
-	/* Should be unconfirmed, so not in hash table yet */
-	WARN_ON(nf_ct_is_confirmed(ct));
-
-	nf_ct_dump_tuple(newreply);
-
-	ct->tuplehash[IP_CT_DIR_REPLY].tuple = *newreply;
-	if (ct->master || (help && !hlist_empty(&help->expectations)))
-		return;
-}
-EXPORT_SYMBOL_GPL(nf_conntrack_alter_reply);
-
 /* Refresh conntrack for this many jiffies and do accounting if do_acct is 1 */
 void __nf_ct_refresh_acct(struct nf_conn *ct,
 			  enum ip_conntrack_info ctinfo,

net/netfilter/nf_conntrack_helper.c

@@ -194,12 +194,7 @@ int __nf_ct_try_assign_helper(struct nf_conn *ct, struct nf_conn *tmpl,
 	struct nf_conntrack_helper *helper = NULL;
 	struct nf_conn_help *help;
 
-	/* We already got a helper explicitly attached. The function
-	 * nf_conntrack_alter_reply - in case NAT is in use - asks for looking
-	 * the helper up again. Since now the user is in full control of
-	 * making consistent helper configurations, skip this automatic
-	 * re-lookup, otherwise we'll lose the helper.
-	 */
+	/* We already got a helper explicitly attached (e.g. nft_ct) */
 	if (test_bit(IPS_HELPER_BIT, &ct->status))
 		return 0;
 

net/netfilter/nf_conntrack_proto_tcp.c

@@ -835,7 +835,8 @@ static bool tcp_error(const struct tcphdr *th,
 
 static noinline bool tcp_new(struct nf_conn *ct, const struct sk_buff *skb,
 			     unsigned int dataoff,
-			     const struct tcphdr *th)
+			     const struct tcphdr *th,
+			     const struct nf_hook_state *state)
 {
 	enum tcp_conntrack new_state;
 	struct net *net = nf_ct_net(ct);
@@ -846,7 +847,7 @@ static noinline bool tcp_new(struct nf_conn *ct, const struct sk_buff *skb,
 
 	/* Invalid: delete conntrack */
 	if (new_state >= TCP_CONNTRACK_MAX) {
-		pr_debug("nf_ct_tcp: invalid new deleting.\n");
+		tcp_error_log(skb, state, "invalid new");
 		return false;
 	}
 
@@ -980,7 +981,7 @@ int nf_conntrack_tcp_packet(struct nf_conn *ct,
 	if (tcp_error(th, skb, dataoff, state))
 		return -NF_ACCEPT;
 
-	if (!nf_ct_is_confirmed(ct) && !tcp_new(ct, skb, dataoff, th))
+	if (!nf_ct_is_confirmed(ct) && !tcp_new(ct, skb, dataoff, th, state))
 		return -NF_ACCEPT;
 
 	spin_lock_bh(&ct->lock);

net/netfilter/nf_tables_api.c

@@ -3441,20 +3441,21 @@ static void audit_log_rule_reset(const struct nft_table *table,
 }
 
 struct nft_rule_dump_ctx {
+	unsigned int s_idx;
 	char *table;
 	char *chain;
+	bool reset;
 };
 
 static int __nf_tables_dump_rules(struct sk_buff *skb,
 				  unsigned int *idx,
 				  struct netlink_callback *cb,
 				  const struct nft_table *table,
-				  const struct nft_chain *chain,
-				  bool reset)
+				  const struct nft_chain *chain)
 {
+	struct nft_rule_dump_ctx *ctx = (void *)cb->ctx;
 	struct net *net = sock_net(skb->sk);
 	const struct nft_rule *rule, *prule;
-	unsigned int s_idx = cb->args[0];
 	unsigned int entries = 0;
 	int ret = 0;
 	u64 handle;
@@ -3463,12 +3464,8 @@ static int __nf_tables_dump_rules(struct sk_buff *skb,
 	list_for_each_entry_rcu(rule, &chain->rules, list) {
 		if (!nft_is_active(net, rule))
 			goto cont_skip;
-		if (*idx < s_idx)
+		if (*idx < ctx->s_idx)
 			goto cont;
-		if (*idx > s_idx) {
-			memset(&cb->args[1], 0,
-					sizeof(cb->args) - sizeof(cb->args[0]));
-		}
 		if (prule)
 			handle = prule->handle;
 		else
@@ -3479,7 +3476,7 @@ static int __nf_tables_dump_rules(struct sk_buff *skb,
 					NFT_MSG_NEWRULE,
 					NLM_F_MULTI | NLM_F_APPEND,
 					table->family,
-					table, chain, rule, handle, reset) < 0) {
+					table, chain, rule, handle, ctx->reset) < 0) {
 			ret = 1;
 			break;
 		}
@@ -3491,7 +3488,7 @@ static int __nf_tables_dump_rules(struct sk_buff *skb,
 		(*idx)++;
 	}
 
-	if (reset && entries)
+	if (ctx->reset && entries)
 		audit_log_rule_reset(table, cb->seq, entries);
 
 	return ret;
@@ -3501,17 +3498,13 @@ static int nf_tables_dump_rules(struct sk_buff *skb,
 				struct netlink_callback *cb)
 {
 	const struct nfgenmsg *nfmsg = nlmsg_data(cb->nlh);
-	const struct nft_rule_dump_ctx *ctx = cb->data;
+	struct nft_rule_dump_ctx *ctx = (void *)cb->ctx;
 	struct nft_table *table;
 	const struct nft_chain *chain;
 	unsigned int idx = 0;
 	struct net *net = sock_net(skb->sk);
 	int family = nfmsg->nfgen_family;
 	struct nftables_pernet *nft_net;
-	bool reset = false;
-
-	if (NFNL_MSG_TYPE(cb->nlh->nlmsg_type) == NFT_MSG_GETRULE_RESET)
-		reset = true;
 
 	rcu_read_lock();
 	nft_net = nft_pernet(net);
@@ -3521,10 +3514,10 @@ static int nf_tables_dump_rules(struct sk_buff *skb,
 		if (family != NFPROTO_UNSPEC && family != table->family)
 			continue;
 
-		if (ctx && ctx->table && strcmp(ctx->table, table->name) != 0)
+		if (ctx->table && strcmp(ctx->table, table->name) != 0)
 			continue;
 
-		if (ctx && ctx->table && ctx->chain) {
+		if (ctx->table && ctx->chain) {
 			struct rhlist_head *list, *tmp;
 
 			list = rhltable_lookup(&table->chains_ht, ctx->chain,
@@ -3536,7 +3529,7 @@ static int nf_tables_dump_rules(struct sk_buff *skb,
 				if (!nft_is_active(net, chain))
 					continue;
 				__nf_tables_dump_rules(skb, &idx,
-						       cb, table, chain, reset);
+						       cb, table, chain);
 				break;
 			}
 			goto done;
@@ -3544,62 +3537,51 @@ static int nf_tables_dump_rules(struct sk_buff *skb,
 
 		list_for_each_entry_rcu(chain, &table->chains, list) {
 			if (__nf_tables_dump_rules(skb, &idx,
-						   cb, table, chain, reset))
+						   cb, table, chain))
 				goto done;
 		}
 
-		if (ctx && ctx->table)
+		if (ctx->table)
 			break;
 	}
 done:
 	rcu_read_unlock();
 
-	cb->args[0] = idx;
+	ctx->s_idx = idx;
 	return skb->len;
 }
 
 static int nf_tables_dump_rules_start(struct netlink_callback *cb)
 {
+	struct nft_rule_dump_ctx *ctx = (void *)cb->ctx;
 	const struct nlattr * const *nla = cb->data;
-	struct nft_rule_dump_ctx *ctx = NULL;
 
-	if (nla[NFTA_RULE_TABLE] || nla[NFTA_RULE_CHAIN]) {
-		ctx = kzalloc(sizeof(*ctx), GFP_ATOMIC);
-		if (!ctx)
-			return -ENOMEM;
+	BUILD_BUG_ON(sizeof(*ctx) > sizeof(cb->ctx));
 
-		if (nla[NFTA_RULE_TABLE]) {
-			ctx->table = nla_strdup(nla[NFTA_RULE_TABLE],
-							GFP_ATOMIC);
-			if (!ctx->table) {
-				kfree(ctx);
-				return -ENOMEM;
-			}
-		}
-		if (nla[NFTA_RULE_CHAIN]) {
-			ctx->chain = nla_strdup(nla[NFTA_RULE_CHAIN],
-						GFP_ATOMIC);
-			if (!ctx->chain) {
-				kfree(ctx->table);
-				kfree(ctx);
-				return -ENOMEM;
-			}
+	if (nla[NFTA_RULE_TABLE]) {
+		ctx->table = nla_strdup(nla[NFTA_RULE_TABLE], GFP_ATOMIC);
+		if (!ctx->table)
+			return -ENOMEM;
+	}
+	if (nla[NFTA_RULE_CHAIN]) {
+		ctx->chain = nla_strdup(nla[NFTA_RULE_CHAIN], GFP_ATOMIC);
+		if (!ctx->chain) {
+			kfree(ctx->table);
+			return -ENOMEM;
 		}
 	}
+	if (NFNL_MSG_TYPE(cb->nlh->nlmsg_type) == NFT_MSG_GETRULE_RESET)
+		ctx->reset = true;
 
-	cb->data = ctx;
 	return 0;
 }
 
 static int nf_tables_dump_rules_done(struct netlink_callback *cb)
 {
-	struct nft_rule_dump_ctx *ctx = cb->data;
+	struct nft_rule_dump_ctx *ctx = (void *)cb->ctx;
 
-	if (ctx) {
-		kfree(ctx->table);
-		kfree(ctx->chain);
-		kfree(ctx);
-	}
+	kfree(ctx->table);
+	kfree(ctx->chain);
 	return 0;
 }