Skip to contentSkip to conversation

netfilter fixes for net

PATCH Applied Florian Westphal 2 participants 11 messages Aug 15, 2023 – Aug 16, 2023

Subsystems (5)
lore
Share & tools
lore

Florian Westphal

author maintainer
#0 lore raw

cover netfilter fixes for net

Hello,

These are netfilter fixes for the *net* tree.

First patch resolves a false-positive lockdep splat:
rcu_dereference is used outside of rcu read lock. Let lockdep validate that the transaction mutex is locked.

Second patch fixes a kdoc warning added in previous PR.

Third patch fixes a memory leak:
The catchall element isn't disabled correctly, this allows userspace to deactivate the element again. This results in refcount underflow which in turn prevents memory release. This was always broken since the feature was added in 5.13.

Patch 4 fixes an incorrect change in the previous pull request:
Adding a duplicate key to a set should work if the duplicate key has expired, restore this behaviour. All from myself.

Patch #5 resolves an old historic artifact in sctp conntrack:
a 300ms timeout for shutdown_ack. Increase this to 3s. From Xin Long.

Patch #6 fixes a sysctl data race in ipvs, two threads can clobber the sysctl value, from Sishuai Gong. This is a day-0 bug that predates git history.

Patches 7, 8 and 9, from Pablo Neira Ayuso, are also followups for the previous GC rework in nf_tables: The netlink notifier and the netns exit path must both increment the gc worker seqcount, else worker may encounter stale (free'd) pointers.

The following changes since commit e4dd0d3a2f64b8bd8029ec70f52bdbebd0644408:

net: fix the RTO timer retransmitting skb every 1ms if linear option is enabled (2023-08-15 20:24:04 +0100)

are available in the Git repository at:

https://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf.git tags/nf-23-08-16

for you to fetch changes up to 23185c6aed1ffb8fc44087880ba2767aba493779:

netfilter: nft_dynset: disallow object maps (2023-08-16 00:05:15 +0200)

----------------------------------------------------------------
nf pull request 2023-08-16

----------------------------------------------------------------
Florian Westphal (4):
netfilter: nf_tables: fix false-positive lockdep splat
netfilter: nf_tables: fix kdoc warnings after gc rework
netfilter: nf_tables: deactivate catchall elements in next generation
netfilter: nf_tables: don't fail inserts if duplicate has expired

Pablo Neira Ayuso (3):
netfilter: nf_tables: fix GC transaction races with netns and netlink event exit path
netfilter: nf_tables: GC transaction race with netns dismantle
netfilter: nft_dynset: disallow object maps

Sishuai Gong (1):
ipvs: fix racy memcpy in proc_do_sync_threshold

Xin Long (1):
netfilter: set default timeout to 3 secs for sctp shutdown send and recv state

 Documentation/networking/nf_conntrack-sysctl.rst |  4 +--
 include/net/netfilter/nf_tables.h                |  1 +
 net/netfilter/ipvs/ip_vs_ctl.c                   |  4 +++
 net/netfilter/nf_conntrack_proto_sctp.c          |  6 ++--
 net/netfilter/nf_tables_api.c                    | 44 +++++++++++++++++++++---
 net/netfilter/nft_dynset.c                       |  3 ++
 net/netfilter/nft_set_pipapo.c                   | 38 +++++++++-----------
 7 files changed, 69 insertions(+), 31 deletions(-)
author maintainer
#1 lore raw

PATCH 1/9 netfilter: nf_tables: fix false-positive lockdep splat

->abort invocation may cause splat on debug kernels:

WARNING: suspicious RCU usage
net/netfilter/nft_set_pipapo.c:1697 suspicious rcu_dereference_check() usage! [..]
rcu_scheduler_active = 2, debug_locks = 1
1 lock held by nft/133554: [..] (nft_net->commit_mutex){+.+.}-{3:3}, at: nf_tables_valid_genid [..]

 lockdep_rcu_suspicious+0x1ad/0x260
 nft_pipapo_abort+0x145/0x180
 __nf_tables_abort+0x5359/0x63d0
 nf_tables_abort+0x24/0x40
 nfnetlink_rcv+0x1a0a/0x22c0
 netlink_unicast+0x73c/0x900
 netlink_sendmsg+0x7f0/0xc20
 ____sys_sendmsg+0x48d/0x760

Transaction mutex is held, so parallel updates are not possible. Switch to _protected and check mutex is held for lockdep enabled builds.

Fixes 212ed75dc5fb ("netfilter: nf_tables: integrate pipapo into commit protocol")
---
 net/netfilter/nft_set_pipapo.c | 13 ++++++++++++-
 1 file changed, 12 insertions(+), 1 deletion(-)
net/netfilter/nft_set_pipapo.c +12 -1
index a5b8301afe4a..5fa12cfc7b84 100644
@@ -1697,6 +1697,17 @@ static void nft_pipapo_commit(const struct nft_set *set)
 	priv->clone = new_clone;
 }
 
+static bool nft_pipapo_transaction_mutex_held(const struct nft_set *set)
+{
+#ifdef CONFIG_PROVE_LOCKING
+	const struct net *net = read_pnet(&set->net);
+
+	return lockdep_is_held(&nft_pernet(net)->commit_mutex);
+#else
+	return true;
+#endif
+}
+
 static void nft_pipapo_abort(const struct nft_set *set)
 {
 	struct nft_pipapo *priv = nft_set_priv(set);
@@ -1705,7 +1716,7 @@ static void nft_pipapo_abort(const struct nft_set *set)
 	if (!priv->dirty)
 		return;
 
-	m = rcu_dereference(priv->match);
+	m = rcu_dereference_protected(priv->match, nft_pipapo_transaction_mutex_held(set));
 
 	new_clone = pipapo_clone(m);
 	if (IS_ERR(new_clone))
Signature · 1 line
2.41.0
#2 lore raw

Re: [PATCH net 1/9] netfilter: nf_tables: fix false-positive lockdep splat

This series was applied to netdev/net.git (main) by Florian Westphal <[email protected]>:

Hello:

This series was applied to netdev/net.git (main)
by Florian Westphal <[email protected]>:

Quoted 17 lines · On Wed, 16 Aug 2023 00:29:51 +0200 you wrote:

->abort invocation may cause splat on debug kernels:

WARNING: suspicious RCU usage
net/netfilter/nft_set_pipapo.c:1697 suspicious rcu_dereference_check() usage! [..]
rcu_scheduler_active = 2, debug_locks = 1
1 lock held by nft/133554: [..] (nft_net->commit_mutex){+.+.}-{3:3}, at: nf_tables_valid_genid [..]

 lockdep_rcu_suspicious+0x1ad/0x260
 nft_pipapo_abort+0x145/0x180
 __nf_tables_abort+0x5359/0x63d0
 nf_tables_abort+0x24/0x40
 nfnetlink_rcv+0x1a0a/0x22c0
 netlink_unicast+0x73c/0x900
 netlink_sendmsg+0x7f0/0xc20
 ____sys_sendmsg+0x48d/0x760
Quoted 1 line

[...]

Here is the summary with links:
- [net,1/9] netfilter: nf_tables: fix false-positive lockdep splat

    https://git.kernel.org/netdev/net/c/b9f052dc68f6
  - [net,2/9] netfilter: nf_tables: fix kdoc warnings after gc rework
    https://git.kernel.org/netdev/net/c/08713cb006b6
  - [net,3/9] netfilter: nf_tables: deactivate catchall elements in next generation
    https://git.kernel.org/netdev/net/c/90e5b3462efa
  - [net,4/9] netfilter: nf_tables: don't fail inserts if duplicate has expired
    https://git.kernel.org/netdev/net/c/7845914f45f0
  - [net,5/9] netfilter: set default timeout to 3 secs for sctp shutdown send and recv state
    https://git.kernel.org/netdev/net/c/9bfab6d23a28
  - [net,6/9] ipvs: fix racy memcpy in proc_do_sync_threshold
    https://git.kernel.org/netdev/net/c/5310760af1d4
  - [net,7/9] netfilter: nf_tables: fix GC transaction races with netns and netlink event exit path
    https://git.kernel.org/netdev/net/c/6a33d8b73dfa
  - [net,8/9] netfilter: nf_tables: GC transaction race with netns dismantle
    https://git.kernel.org/netdev/net/c/02c6c24402bf
  - [net,9/9] netfilter: nft_dynset: disallow object maps
    https://git.kernel.org/netdev/net/c/23185c6aed1f

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
author maintainer
#3 lore raw

PATCH 2/9 netfilter: nf_tables: fix kdoc warnings after gc rework

In reply to Florian Westphal

Jakub Kicinski says:
We've got some new kdoc warnings here:
net/netfilter/nft_set_pipapo.c:1557: warning: Function parameter or member '_set' not described in 'pipapo_gc' net/netfilter/nft_set_pipapo.c:1557: warning: Excess function parameter 'set' description in 'pipapo_gc' include/net/netfilter/nf_tables.h:577: warning: Function parameter or member 'dead' not described in 'nft_set'

Fixes 5f68718b34a5 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
Fixes f6c383b8c31a ("netfilter: nf_tables: adapt set backend to use GC transaction API")
---
 include/net/netfilter/nf_tables.h | 1 +
 net/netfilter/nft_set_pipapo.c    | 2 +-
 2 files changed, 2 insertions(+), 1 deletion(-)
include/net/netfilter/nf_tables.h +1 -0
index 35870858ddf2..e9ae567c037d 100644
@@ -534,6 +534,7 @@ struct nft_set_elem_expr {
  *	@expr: stateful expression
  * 	@ops: set ops
  * 	@flags: set flags
+ *	@dead: set will be freed, never cleared
  *	@genmask: generation mask
  * 	@klen: key length
  * 	@dlen: data length
net/netfilter/nft_set_pipapo.c +1 -1
index 5fa12cfc7b84..f95b3844162e 100644
@@ -1549,7 +1549,7 @@ static void nft_pipapo_gc_deactivate(struct net *net, struct nft_set *set,
 
 /**
  * pipapo_gc() - Drop expired entries from set, destroy start and end elements
- * @set:	nftables API set representation
+ * @_set:	nftables API set representation
  * @m:		Matching data
  */
 static void pipapo_gc(const struct nft_set *_set, struct nft_pipapo_match *m)
Signature · 1 line
2.41.0
author maintainer
#4 lore raw

PATCH 3/9 netfilter: nf_tables: deactivate catchall elements in next generation

In reply to Florian Westphal

When flushing, individual set elements are disabled in the next generation via the ->flush callback.

Catchall elements are not disabled. This is incorrect and may lead to double-deactivations of catchall elements which then results in memory leaks:

WARNING: CPU: 1 PID: 3300 at include/net/netfilter/nf_tables.h:1172 nft_map_deactivate+0x549/0x730
CPU: 1 PID: 3300 Comm: nft Not tainted 6.5.0-rc5+ #60
RIP: 0010:nft_map_deactivate+0x549/0x730
 [..]
 ? nft_map_deactivate+0x549/0x730
 nf_tables_delset+0xb66/0xeb0

(the warn is due to nft_use_dec() detecting underflow).

Fixes aaa31047a6d2 ("netfilter: nftables: add catch-all set element support")
---
 net/netfilter/nf_tables_api.c | 1 +
 1 file changed, 1 insertion(+)
net/netfilter/nf_tables_api.c +1 -0
index c62227ae7746..6f31022cacc6 100644
@@ -7091,6 +7091,7 @@ static int nft_set_catchall_flush(const struct nft_ctx *ctx,
 		ret = __nft_set_catchall_flush(ctx, set, &elem);
 		if (ret < 0)
 			break;
+		nft_set_elem_change_active(ctx->net, set, ext);
 	}
 
 	return ret;
Signature · 1 line
2.41.0
author maintainer
#5 lore raw

PATCH 4/9 netfilter: nf_tables: don't fail inserts if duplicate has expired

In reply to Florian Westphal

nftables selftests fail:
run-tests.sh testcases/sets/0044interval_overlap_0
Expected: 0-2 . 0-3, got:
W: [FAILED] ./testcases/sets/0044interval_overlap_0: got 1

Insertion must ignore duplicate but expired entries.

Moreover, there is a strange asymmetry in nft_pipapo_activate:

It refetches the current element, whereas the other ->activate callbacks (bitmap, hash, rhash, rbtree) use elem->priv.
Same for .remove: other set implementations take elem->priv, nft_pipapo_remove fetches elem->priv, then does a relookup, remove this.

I suspect this was the reason for the change that prompted the removal of the expired check in pipapo_get() in the first place, but skipping exired elements there makes no sense to me, this helper is used for normal get requests, insertions (duplicate check) and deactivate callback.

In first two cases expired elements must be skipped.

For ->deactivate(), this gets called for DELSETELEM, so it seems to me that expired elements should be skipped as well, i.e. delete request should fail with -ENOENT error.

Fixes 24138933b97b ("netfilter: nf_tables: don't skip expired elements during walk")
---
 net/netfilter/nft_set_pipapo.c | 23 ++++-------------------
 1 file changed, 4 insertions(+), 19 deletions(-)
net/netfilter/nft_set_pipapo.c +4 -19
index f95b3844162e..3757fcc55723 100644
@@ -566,6 +566,8 @@ static struct nft_pipapo_elem *pipapo_get(const struct net *net,
 			goto out;
 
 		if (last) {
+			if (nft_set_elem_expired(&f->mt[b].e->ext))
+				goto next_match;
 			if ((genmask &&
 			     !nft_set_elem_active(&f->mt[b].e->ext, genmask)))
 				goto next_match;
@@ -600,17 +602,8 @@ static struct nft_pipapo_elem *pipapo_get(const struct net *net,
 static void *nft_pipapo_get(const struct net *net, const struct nft_set *set,
 			    const struct nft_set_elem *elem, unsigned int flags)
 {
-	struct nft_pipapo_elem *ret;
-
-	ret = pipapo_get(net, set, (const u8 *)elem->key.val.data,
+	return pipapo_get(net, set, (const u8 *)elem->key.val.data,
 			 nft_genmask_cur(net));
-	if (IS_ERR(ret))
-		return ret;
-
-	if (nft_set_elem_expired(&ret->ext))
-		return ERR_PTR(-ENOENT);
-
-	return ret;
 }
 
 /**
@@ -1743,11 +1736,7 @@ static void nft_pipapo_activate(const struct net *net,
 				const struct nft_set *set,
 				const struct nft_set_elem *elem)
 {
-	struct nft_pipapo_elem *e;
-
-	e = pipapo_get(net, set, (const u8 *)elem->key.val.data, 0);
-	if (IS_ERR(e))
-		return;
+	struct nft_pipapo_elem *e = elem->priv;
 
 	nft_set_elem_change_active(net, set, &e->ext);
 }
@@ -1961,10 +1950,6 @@ static void nft_pipapo_remove(const struct net *net, const struct nft_set *set,
 
 	data = (const u8 *)nft_set_ext_key(&e->ext);
 
-	e = pipapo_get(net, set, data, 0);
-	if (IS_ERR(e))
-		return;
-
 	while ((rules_f0 = pipapo_rules_same_key(m->f, first_rule))) {
 		union nft_pipapo_map_bucket rulemap[NFT_PIPAPO_MAX_FIELDS];
 		const u8 *match_start, *match_end;
Signature · 1 line
2.41.0
author maintainer
#6 lore raw

PATCH 5/9 netfilter: set default timeout to 3 secs for sctp shutdown send and recv state

In reply to Florian Westphal

From: Xin Long <[email protected]>

In SCTP protocol, it is using the same timer (T2 timer) for SHUTDOWN and SHUTDOWN_ACK retransmission. However in sctp conntrack the default timeout value for SCTP_CONNTRACK_SHUTDOWN_ACK_SENT state is 3 secs while it's 300 msecs for SCTP_CONNTRACK_SHUTDOWN_SEND/RECV state.

As Paolo Valerio noticed, this might cause unwanted expiration of the ct entry. In my test, with 1s tc netem delay set on the NAT path, after the SHUTDOWN is sent, the sctp ct entry enters SCTP_CONNTRACK_SHUTDOWN_SEND state. However, due to 300ms (too short) delay, when the SHUTDOWN_ACK is sent back from the peer, the sctp ct entry has expired and been deleted, and then the SHUTDOWN_ACK has to be dropped.

Also, it is confusing these two sysctl options always show 0 due to all timeout values using sec as unit:

net.netfilter.nf_conntrack_sctp_timeout_shutdown_recd = 0 net.netfilter.nf_conntrack_sctp_timeout_shutdown_sent = 0

This patch fixes it by also using 3 secs for sctp shutdown send and recv state in sctp conntrack, which is also RTO.initial value in SCTP protocol.

Note that the very short time value for SCTP_CONNTRACK_SHUTDOWN_SEND/RECV was probably used for a rare scenario where SHUTDOWN is sent on 1st path but SHUTDOWN_ACK is replied on 2nd path, then a new connection started immediately on 1st path. So this patch also moves from SHUTDOWN_SEND/RECV to CLOSE when receiving INIT in the ORIGINAL direction.

Fixes 9fb9cbb1082d ("[NETFILTER]: Add nf_conntrack subsystem.")
---
 Documentation/networking/nf_conntrack-sysctl.rst | 4 ++--
 net/netfilter/nf_conntrack_proto_sctp.c          | 6 +++---
 2 files changed, 5 insertions(+), 5 deletions(-)
Documentation/networking/nf_conntrack-sysctl.rst +2 -2
index 8b1045c3b59e..c383a394c665 100644
@@ -178,10 +178,10 @@ nf_conntrack_sctp_timeout_established - INTEGER (seconds)
 	Default is set to (hb_interval * path_max_retrans + rto_max)
 
 nf_conntrack_sctp_timeout_shutdown_sent - INTEGER (seconds)
-	default 0.3
+	default 3
 
 nf_conntrack_sctp_timeout_shutdown_recd - INTEGER (seconds)
-	default 0.3
+	default 3
 
 nf_conntrack_sctp_timeout_shutdown_ack_sent - INTEGER (seconds)
 	default 3
net/netfilter/nf_conntrack_proto_sctp.c +3 -3
index 91eacc9b0b98..b6bcc8f2f46b 100644
@@ -49,8 +49,8 @@ static const unsigned int sctp_timeouts[SCTP_CONNTRACK_MAX] = {
 	[SCTP_CONNTRACK_COOKIE_WAIT]		= 3 SECS,
 	[SCTP_CONNTRACK_COOKIE_ECHOED]		= 3 SECS,
 	[SCTP_CONNTRACK_ESTABLISHED]		= 210 SECS,
-	[SCTP_CONNTRACK_SHUTDOWN_SENT]		= 300 SECS / 1000,
-	[SCTP_CONNTRACK_SHUTDOWN_RECD]		= 300 SECS / 1000,
+	[SCTP_CONNTRACK_SHUTDOWN_SENT]		= 3 SECS,
+	[SCTP_CONNTRACK_SHUTDOWN_RECD]		= 3 SECS,
 	[SCTP_CONNTRACK_SHUTDOWN_ACK_SENT]	= 3 SECS,
 	[SCTP_CONNTRACK_HEARTBEAT_SENT]		= 30 SECS,
 };
@@ -105,7 +105,7 @@ static const u8 sctp_conntracks[2][11][SCTP_CONNTRACK_MAX] = {
 	{
 /*	ORIGINAL	*/
 /*                  sNO, sCL, sCW, sCE, sES, sSS, sSR, sSA, sHS */
-/* init         */ {sCL, sCL, sCW, sCE, sES, sSS, sSR, sSA, sCW},
+/* init         */ {sCL, sCL, sCW, sCE, sES, sCL, sCL, sSA, sCW},
 /* init_ack     */ {sCL, sCL, sCW, sCE, sES, sSS, sSR, sSA, sCL},
 /* abort        */ {sCL, sCL, sCL, sCL, sCL, sCL, sCL, sCL, sCL},
 /* shutdown     */ {sCL, sCL, sCW, sCE, sSS, sSS, sSR, sSA, sCL},
Signature · 1 line
2.41.0
author maintainer
#7 lore raw

PATCH 6/9 ipvs: fix racy memcpy in proc_do_sync_threshold

In reply to Florian Westphal

From: Sishuai Gong <[email protected]>

When two threads run proc_do_sync_threshold() in parallel, data races could happen between the two memcpy():

Thread-1 Thread-2

memcpy(val, valp, sizeof(val));
				memcpy(valp, val, sizeof(val));

This race might mess up the (struct ctl_table *) table->data, so we add a mutex lock to serialize them.

---
 net/netfilter/ipvs/ip_vs_ctl.c | 4 ++++
 1 file changed, 4 insertions(+)
net/netfilter/ipvs/ip_vs_ctl.c +4 -0
index 62606fb44d02..4bb0d90eca1c 100644
@@ -1876,6 +1876,7 @@ static int
 proc_do_sync_threshold(struct ctl_table *table, int write,
 		       void *buffer, size_t *lenp, loff_t *ppos)
 {
+	struct netns_ipvs *ipvs = table->extra2;
 	int *valp = table->data;
 	int val[2];
 	int rc;
@@ -1885,6 +1886,7 @@ proc_do_sync_threshold(struct ctl_table *table, int write,
 		.mode = table->mode,
 	};
 
+	mutex_lock(&ipvs->sync_mutex);
 	memcpy(val, valp, sizeof(val));
 	rc = proc_dointvec(&tmp, write, buffer, lenp, ppos);
 	if (write) {
@@ -1894,6 +1896,7 @@ proc_do_sync_threshold(struct ctl_table *table, int write,
 		else
 			memcpy(valp, val, sizeof(val));
 	}
+	mutex_unlock(&ipvs->sync_mutex);
 	return rc;
 }
 
@@ -4321,6 +4324,7 @@ static int __net_init ip_vs_control_net_init_sysctl(struct netns_ipvs *ipvs)
 	ipvs->sysctl_sync_threshold[0] = DEFAULT_SYNC_THRESHOLD;
 	ipvs->sysctl_sync_threshold[1] = DEFAULT_SYNC_PERIOD;
 	tbl[idx].data = &ipvs->sysctl_sync_threshold;
+	tbl[idx].extra2 = ipvs;
 	tbl[idx++].maxlen = sizeof(ipvs->sysctl_sync_threshold);
 	ipvs->sysctl_sync_refresh_period = DEFAULT_SYNC_REFRESH_PERIOD;
 	tbl[idx++].data = &ipvs->sysctl_sync_refresh_period;
Signature · 1 line
2.41.0
author maintainer
#8 lore raw

PATCH 7/9 netfilter: nf_tables: fix GC transaction races with netns and netlink event exit path

In reply to Florian Westphal

From: Pablo Neira Ayuso <[email protected]>

Netlink event path is missing a synchronization point with GC transactions. Add GC sequence number update to netns release path and netlink event path, any GC transaction losing race will be discarded.

Fixes 5f68718b34a5 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
---
 net/netfilter/nf_tables_api.c | 36 +++++++++++++++++++++++++++++++----
 1 file changed, 32 insertions(+), 4 deletions(-)
net/netfilter/nf_tables_api.c +32 -4
index 6f31022cacc6..8ac4dd8be1a2 100644
@@ -9739,6 +9739,22 @@ static void nft_set_commit_update(struct list_head *set_update_list)
 	}
 }
 
+static unsigned int nft_gc_seq_begin(struct nftables_pernet *nft_net)
+{
+	unsigned int gc_seq;
+
+	/* Bump gc counter, it becomes odd, this is the busy mark. */
+	gc_seq = READ_ONCE(nft_net->gc_seq);
+	WRITE_ONCE(nft_net->gc_seq, ++gc_seq);
+
+	return gc_seq;
+}
+
+static void nft_gc_seq_end(struct nftables_pernet *nft_net, unsigned int gc_seq)
+{
+	WRITE_ONCE(nft_net->gc_seq, ++gc_seq);
+}
+
 static int nf_tables_commit(struct net *net, struct sk_buff *skb)
 {
 	struct nftables_pernet *nft_net = nft_pernet(net);
@@ -9824,9 +9840,7 @@ static int nf_tables_commit(struct net *net, struct sk_buff *skb)
 
 	WRITE_ONCE(nft_net->base_seq, base_seq);
 
-	/* Bump gc counter, it becomes odd, this is the busy mark. */
-	gc_seq = READ_ONCE(nft_net->gc_seq);
-	WRITE_ONCE(nft_net->gc_seq, ++gc_seq);
+	gc_seq = nft_gc_seq_begin(nft_net);
 
 	/* step 3. Start new generation, rules_gen_X now in use. */
 	net->nft.gencursor = nft_gencursor_next(net);
@@ -10039,7 +10053,7 @@ static int nf_tables_commit(struct net *net, struct sk_buff *skb)
 	nf_tables_gen_notify(net, skb, NFT_MSG_NEWGEN);
 	nf_tables_commit_audit_log(&adl, nft_net->base_seq);
 
-	WRITE_ONCE(nft_net->gc_seq, ++gc_seq);
+	nft_gc_seq_end(nft_net, gc_seq);
 	nf_tables_commit_release(net);
 
 	return 0;
@@ -11040,6 +11054,7 @@ static int nft_rcv_nl_event(struct notifier_block *this, unsigned long event,
 	struct net *net = n->net;
 	unsigned int deleted;
 	bool restart = false;
+	unsigned int gc_seq;
 
 	if (event != NETLINK_URELEASE || n->protocol != NETLINK_NETFILTER)
 		return NOTIFY_DONE;
@@ -11047,6 +11062,9 @@ static int nft_rcv_nl_event(struct notifier_block *this, unsigned long event,
 	nft_net = nft_pernet(net);
 	deleted = 0;
 	mutex_lock(&nft_net->commit_mutex);
+
+	gc_seq = nft_gc_seq_begin(nft_net);
+
 	if (!list_empty(&nf_tables_destroy_list))
 		rcu_barrier();
 again:
@@ -11069,6 +11087,8 @@ static int nft_rcv_nl_event(struct notifier_block *this, unsigned long event,
 		if (restart)
 			goto again;
 	}
+	nft_gc_seq_end(nft_net, gc_seq);
+
 	mutex_unlock(&nft_net->commit_mutex);
 
 	return NOTIFY_DONE;
@@ -11106,12 +11126,20 @@ static void __net_exit nf_tables_pre_exit_net(struct net *net)
 static void __net_exit nf_tables_exit_net(struct net *net)
 {
 	struct nftables_pernet *nft_net = nft_pernet(net);
+	unsigned int gc_seq;
 
 	mutex_lock(&nft_net->commit_mutex);
+
+	gc_seq = nft_gc_seq_begin(nft_net);
+
 	if (!list_empty(&nft_net->commit_list) ||
 	    !list_empty(&nft_net->module_list))
 		__nf_tables_abort(net, NFNL_ABORT_NONE);
+
 	__nft_release_tables(net);
+
+	nft_gc_seq_end(nft_net, gc_seq);
+
 	mutex_unlock(&nft_net->commit_mutex);
 	WARN_ON_ONCE(!list_empty(&nft_net->tables));
 	WARN_ON_ONCE(!list_empty(&nft_net->module_list));
Signature · 1 line
2.41.0
author maintainer
#9 lore raw

PATCH 8/9 netfilter: nf_tables: GC transaction race with netns dismantle

In reply to Florian Westphal

From: Pablo Neira Ayuso <[email protected]>

Use maybe_get_net() since GC workqueue might race with netns exit path.

Fixes 5f68718b34a5 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
---
 net/netfilter/nf_tables_api.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
net/netfilter/nf_tables_api.c +6 -1
index 8ac4dd8be1a2..3e841e45f2c0 100644
@@ -9481,9 +9481,14 @@ struct nft_trans_gc *nft_trans_gc_alloc(struct nft_set *set,
 	if (!trans)
 		return NULL;
 
+	trans->net = maybe_get_net(net);
+	if (!trans->net) {
+		kfree(trans);
+		return NULL;
+	}
+
 	refcount_inc(&set->refs);
 	trans->set = set;
-	trans->net = get_net(net);
 	trans->seq = gc_seq;
 
 	return trans;
Signature · 1 line
2.41.0
author maintainer
#10 lore raw

PATCH 9/9 netfilter: nft_dynset: disallow object maps

In reply to Florian Westphal

From: Pablo Neira Ayuso <[email protected]>

Do not allow to insert elements from datapath to objects maps.

Fixes 8aeff920dcc9 ("netfilter: nf_tables: add stateful object reference to set elements")
---
 net/netfilter/nft_dynset.c | 3 +++
 1 file changed, 3 insertions(+)
net/netfilter/nft_dynset.c +3 -0
index 4fb34d76dbea..5c5cc01c73c5 100644
@@ -191,6 +191,9 @@ static int nft_dynset_init(const struct nft_ctx *ctx,
 	if (IS_ERR(set))
 		return PTR_ERR(set);
 
+	if (set->flags & NFT_SET_OBJECT)
+		return -EOPNOTSUPP;
+
 	if (set->ops->update == NULL)
 		return -EOPNOTSUPP;
 
Signature · 1 line
2.41.0