[PATCH net 0/6] Netfilter fixes for net

[PATCH net 0/6] Netfilter fixes for net

[Pablo Neira Ayuso]

Hi,

The following patchset contains Netfilter fixes for net:

Patch #1 unlike early commit path stage which triggers a call to abort,
         an explicit release of the batch is required on abort, otherwise
         mutex is released and commit_list remains in place.

Patch #2 release mutex after nft_gc_seq_end() in commit path, otherwise
         async GC worker could collect expired objects.

Patch #3 flush pending destroy work in module removal path, otherwise UaF
         is possible.

Patch #4 and #6 restrict the table dormant flag with basechain updates
	 to fix state inconsistency in the hook registration.

Patch #5 adds missing RCU read side lock to flowtable type to avoid races
	 with module removal.

Please, pull these changes from:

  git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf.git nf-24-04-04

Thanks.

----------------------------------------------------------------

The following changes since commit 72076fc9fe60b9143cd971fd8737718719bc512e:

  Revert "tg3: Remove residual error handling in tg3_suspend" (2024-04-04 10:51:01 +0200)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf.git tags/nf-24-04-04

for you to fetch changes up to 1bc83a019bbe268be3526406245ec28c2458a518:

  netfilter: nf_tables: discard table flag update with pending basechain deletion (2024-04-04 11:38:35 +0200)

----------------------------------------------------------------
netfilter pull request 24-04-04

----------------------------------------------------------------
Pablo Neira Ayuso (5):
      netfilter: nf_tables: release batch on table validation from abort path
      netfilter: nf_tables: release mutex after nft_gc_seq_end from abort path
      netfilter: nf_tables: flush pending destroy work before exit_net release
      netfilter: nf_tables: reject new basechain after table flag update
      netfilter: nf_tables: discard table flag update with pending basechain deletion

Ziyang Xuan (1):
      netfilter: nf_tables: Fix potential data-race in __nft_flowtable_type_get()

 net/netfilter/nf_tables_api.c | 50 +++++++++++++++++++++++++++++--------------
 1 file changed, 34 insertions(+), 16 deletions(-)

[PATCH net 1/6] netfilter: nf_tables: release batch on table validation from abort path

[Pablo Neira Ayuso]

Unlike early commit path stage which triggers a call to abort, an
explicit release of the batch is required on abort, otherwise mutex is
released and commit_list remains in place.

Add WARN_ON_ONCE to ensure commit_list is empty from the abort path
before releasing the mutex.

After this patch, commit_list is always assumed to be empty before
grabbing the mutex, therefore

  03c1f1ef1584 ("netfilter: Cleanup nft_net->module_list from nf_tables_exit_net()")

only needs to release the pending modules for registration.

Cc: stable@vger.kernel.org
Fixes: c0391b6ab810 ("netfilter: nf_tables: missing validation from the abort path")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
 net/netfilter/nf_tables_api.c | 15 ++++++++++-----
 1 file changed, 10 insertions(+), 5 deletions(-)

diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index fd86f2720c9e..ffcd3213c335 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -10455,10 +10455,11 @@ static int __nf_tables_abort(struct net *net, enum nfnl_abort_action action)
 	struct nft_trans *trans, *next;
 	LIST_HEAD(set_update_list);
 	struct nft_trans_elem *te;
+	int err = 0;
 
 	if (action == NFNL_ABORT_VALIDATE &&
 	    nf_tables_validate(net) < 0)
-		return -EAGAIN;
+		err = -EAGAIN;
 
 	list_for_each_entry_safe_reverse(trans, next, &nft_net->commit_list,
 					 list) {
@@ -10655,7 +10656,7 @@ static int __nf_tables_abort(struct net *net, enum nfnl_abort_action action)
 	else
 		nf_tables_module_autoload_cleanup(net);
 
-	return 0;
+	return err;
 }
 
 static int nf_tables_abort(struct net *net, struct sk_buff *skb,
@@ -10668,6 +10669,9 @@ static int nf_tables_abort(struct net *net, struct sk_buff *skb,
 	gc_seq = nft_gc_seq_begin(nft_net);
 	ret = __nf_tables_abort(net, action);
 	nft_gc_seq_end(nft_net, gc_seq);
+
+	WARN_ON_ONCE(!list_empty(&nft_net->commit_list));
+
 	mutex_unlock(&nft_net->commit_mutex);
 
 	return ret;
@@ -11473,9 +11477,10 @@ static void __net_exit nf_tables_exit_net(struct net *net)
 
 	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);
+	WARN_ON_ONCE(!list_empty(&nft_net->commit_list));
+
+	if (!list_empty(&nft_net->module_list))
+		nf_tables_module_autoload_cleanup(net);
 
 	__nft_release_tables(net);
 
-- 
2.30.2

Re: [PATCH net 1/6] netfilter: nf_tables: release batch on table validation from abort path

[patchwork-bot+netdevbpf]

Hello:

This series was applied to netdev/net.git (main)
by Pablo Neira Ayuso <pablo@netfilter.org>:

On Thu,  4 Apr 2024 12:43:29 +0200 you wrote:
> Unlike early commit path stage which triggers a call to abort, an
> explicit release of the batch is required on abort, otherwise mutex is
> released and commit_list remains in place.
> 
> Add WARN_ON_ONCE to ensure commit_list is empty from the abort path
> before releasing the mutex.
> 
> [...]

Here is the summary with links:
  - [net,1/6] netfilter: nf_tables: release batch on table validation from abort path
    https://git.kernel.org/netdev/net/c/a45e6889575c
  - [net,2/6] netfilter: nf_tables: release mutex after nft_gc_seq_end from abort path
    https://git.kernel.org/netdev/net/c/0d459e2ffb54
  - [net,3/6] netfilter: nf_tables: flush pending destroy work before exit_net release
    https://git.kernel.org/netdev/net/c/24cea9677025
  - [net,4/6] netfilter: nf_tables: reject new basechain after table flag update
    https://git.kernel.org/netdev/net/c/994209ddf4f4
  - [net,5/6] netfilter: nf_tables: Fix potential data-race in __nft_flowtable_type_get()
    https://git.kernel.org/netdev/net/c/24225011d81b
  - [net,6/6] netfilter: nf_tables: discard table flag update with pending basechain deletion
    https://git.kernel.org/netdev/net/c/1bc83a019bbe

You are awesome, thank you!
-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html

[PATCH net 2/6] netfilter: nf_tables: release mutex after nft_gc_seq_end from abort path

[Pablo Neira Ayuso]

The commit mutex should not be released during the critical section
between nft_gc_seq_begin() and nft_gc_seq_end(), otherwise, async GC
worker could collect expired objects and get the released commit lock
within the same GC sequence.

nf_tables_module_autoload() temporarily releases the mutex to load
module dependencies, then it goes back to replay the transaction again.
Move it at the end of the abort phase after nft_gc_seq_end() is called.

Cc: stable@vger.kernel.org
Fixes: 720344340fb9 ("netfilter: nf_tables: GC transaction race with abort path")
Reported-by: Kuan-Ting Chen <hexrabbit@devco.re>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
 net/netfilter/nf_tables_api.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)

diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index ffcd3213c335..0d432d0674e1 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -10651,11 +10651,6 @@ static int __nf_tables_abort(struct net *net, enum nfnl_abort_action action)
 		nf_tables_abort_release(trans);
 	}
 
-	if (action == NFNL_ABORT_AUTOLOAD)
-		nf_tables_module_autoload(net);
-	else
-		nf_tables_module_autoload_cleanup(net);
-
 	return err;
 }
 
@@ -10672,6 +10667,14 @@ static int nf_tables_abort(struct net *net, struct sk_buff *skb,
 
 	WARN_ON_ONCE(!list_empty(&nft_net->commit_list));
 
+	/* module autoload needs to happen after GC sequence update because it
+	 * temporarily releases and grabs mutex again.
+	 */
+	if (action == NFNL_ABORT_AUTOLOAD)
+		nf_tables_module_autoload(net);
+	else
+		nf_tables_module_autoload_cleanup(net);
+
 	mutex_unlock(&nft_net->commit_mutex);
 
 	return ret;
-- 
2.30.2

[PATCH net 3/6] netfilter: nf_tables: flush pending destroy work before exit_net release

[Pablo Neira Ayuso]

Similar to 2c9f0293280e ("netfilter: nf_tables: flush pending destroy
work before netlink notifier") to address a race between exit_net and
the destroy workqueue.

The trace below shows an element to be released via destroy workqueue
while exit_net path (triggered via module removal) has already released
the set that is used in such transaction.

[ 1360.547789] BUG: KASAN: slab-use-after-free in nf_tables_trans_destroy_work+0x3f5/0x590 [nf_tables]
[ 1360.547861] Read of size 8 at addr ffff888140500cc0 by task kworker/4:1/152465
[ 1360.547870] CPU: 4 PID: 152465 Comm: kworker/4:1 Not tainted 6.8.0+ #359
[ 1360.547882] Workqueue: events nf_tables_trans_destroy_work [nf_tables]
[ 1360.547984] Call Trace:
[ 1360.547991]  <TASK>
[ 1360.547998]  dump_stack_lvl+0x53/0x70
[ 1360.548014]  print_report+0xc4/0x610
[ 1360.548026]  ? __virt_addr_valid+0xba/0x160
[ 1360.548040]  ? __pfx__raw_spin_lock_irqsave+0x10/0x10
[ 1360.548054]  ? nf_tables_trans_destroy_work+0x3f5/0x590 [nf_tables]
[ 1360.548176]  kasan_report+0xae/0xe0
[ 1360.548189]  ? nf_tables_trans_destroy_work+0x3f5/0x590 [nf_tables]
[ 1360.548312]  nf_tables_trans_destroy_work+0x3f5/0x590 [nf_tables]
[ 1360.548447]  ? __pfx_nf_tables_trans_destroy_work+0x10/0x10 [nf_tables]
[ 1360.548577]  ? _raw_spin_unlock_irq+0x18/0x30
[ 1360.548591]  process_one_work+0x2f1/0x670
[ 1360.548610]  worker_thread+0x4d3/0x760
[ 1360.548627]  ? __pfx_worker_thread+0x10/0x10
[ 1360.548640]  kthread+0x16b/0x1b0
[ 1360.548653]  ? __pfx_kthread+0x10/0x10
[ 1360.548665]  ret_from_fork+0x2f/0x50
[ 1360.548679]  ? __pfx_kthread+0x10/0x10
[ 1360.548690]  ret_from_fork_asm+0x1a/0x30
[ 1360.548707]  </TASK>

[ 1360.548719] Allocated by task 192061:
[ 1360.548726]  kasan_save_stack+0x20/0x40
[ 1360.548739]  kasan_save_track+0x14/0x30
[ 1360.548750]  __kasan_kmalloc+0x8f/0xa0
[ 1360.548760]  __kmalloc_node+0x1f1/0x450
[ 1360.548771]  nf_tables_newset+0x10c7/0x1b50 [nf_tables]
[ 1360.548883]  nfnetlink_rcv_batch+0xbc4/0xdc0 [nfnetlink]
[ 1360.548909]  nfnetlink_rcv+0x1a8/0x1e0 [nfnetlink]
[ 1360.548927]  netlink_unicast+0x367/0x4f0
[ 1360.548935]  netlink_sendmsg+0x34b/0x610
[ 1360.548944]  ____sys_sendmsg+0x4d4/0x510
[ 1360.548953]  ___sys_sendmsg+0xc9/0x120
[ 1360.548961]  __sys_sendmsg+0xbe/0x140
[ 1360.548971]  do_syscall_64+0x55/0x120
[ 1360.548982]  entry_SYSCALL_64_after_hwframe+0x55/0x5d

[ 1360.548994] Freed by task 192222:
[ 1360.548999]  kasan_save_stack+0x20/0x40
[ 1360.549009]  kasan_save_track+0x14/0x30
[ 1360.549019]  kasan_save_free_info+0x3b/0x60
[ 1360.549028]  poison_slab_object+0x100/0x180
[ 1360.549036]  __kasan_slab_free+0x14/0x30
[ 1360.549042]  kfree+0xb6/0x260
[ 1360.549049]  __nft_release_table+0x473/0x6a0 [nf_tables]
[ 1360.549131]  nf_tables_exit_net+0x170/0x240 [nf_tables]
[ 1360.549221]  ops_exit_list+0x50/0xa0
[ 1360.549229]  free_exit_list+0x101/0x140
[ 1360.549236]  unregister_pernet_operations+0x107/0x160
[ 1360.549245]  unregister_pernet_subsys+0x1c/0x30
[ 1360.549254]  nf_tables_module_exit+0x43/0x80 [nf_tables]
[ 1360.549345]  __do_sys_delete_module+0x253/0x370
[ 1360.549352]  do_syscall_64+0x55/0x120
[ 1360.549360]  entry_SYSCALL_64_after_hwframe+0x55/0x5d

(gdb) list *__nft_release_table+0x473
0x1e033 is in __nft_release_table (net/netfilter/nf_tables_api.c:11354).
11349           list_for_each_entry_safe(flowtable, nf, &table->flowtables, list) {
11350                   list_del(&flowtable->list);
11351                   nft_use_dec(&table->use);
11352                   nf_tables_flowtable_destroy(flowtable);
11353           }
11354           list_for_each_entry_safe(set, ns, &table->sets, list) {
11355                   list_del(&set->list);
11356                   nft_use_dec(&table->use);
11357                   if (set->flags & (NFT_SET_MAP | NFT_SET_OBJECT))
11358                           nft_map_deactivate(&ctx, set);
(gdb)

[ 1360.549372] Last potentially related work creation:
[ 1360.549376]  kasan_save_stack+0x20/0x40
[ 1360.549384]  __kasan_record_aux_stack+0x9b/0xb0
[ 1360.549392]  __queue_work+0x3fb/0x780
[ 1360.549399]  queue_work_on+0x4f/0x60
[ 1360.549407]  nft_rhash_remove+0x33b/0x340 [nf_tables]
[ 1360.549516]  nf_tables_commit+0x1c6a/0x2620 [nf_tables]
[ 1360.549625]  nfnetlink_rcv_batch+0x728/0xdc0 [nfnetlink]
[ 1360.549647]  nfnetlink_rcv+0x1a8/0x1e0 [nfnetlink]
[ 1360.549671]  netlink_unicast+0x367/0x4f0
[ 1360.549680]  netlink_sendmsg+0x34b/0x610
[ 1360.549690]  ____sys_sendmsg+0x4d4/0x510
[ 1360.549697]  ___sys_sendmsg+0xc9/0x120
[ 1360.549706]  __sys_sendmsg+0xbe/0x140
[ 1360.549715]  do_syscall_64+0x55/0x120
[ 1360.549725]  entry_SYSCALL_64_after_hwframe+0x55/0x5d

Fixes: 0935d5588400 ("netfilter: nf_tables: asynchronous release")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
 net/netfilter/nf_tables_api.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 0d432d0674e1..17bf53cd0e84 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -11575,6 +11575,7 @@ static void __exit nf_tables_module_exit(void)
 	unregister_netdevice_notifier(&nf_tables_flowtable_notifier);
 	nft_chain_filter_fini();
 	nft_chain_route_fini();
+	nf_tables_trans_destroy_flush_work();
 	unregister_pernet_subsys(&nf_tables_net_ops);
 	cancel_work_sync(&trans_gc_work);
 	cancel_work_sync(&trans_destroy_work);
-- 
2.30.2

[PATCH net 4/6] netfilter: nf_tables: reject new basechain after table flag update

[Pablo Neira Ayuso]

When dormant flag is toggled, hooks are disabled in the commit phase by
iterating over current chains in table (existing and new).

The following configuration allows for an inconsistent state:

  add table x
  add chain x y { type filter hook input priority 0; }
  add table x { flags dormant; }
  add chain x w { type filter hook input priority 1; }

which triggers the following warning when trying to unregister chain w
which is already unregistered.

[  127.322252] WARNING: CPU: 7 PID: 1211 at net/netfilter/core.c:50                                                                     1 __nf_unregister_net_hook+0x21a/0x260
[...]
[  127.322519] Call Trace:
[  127.322521]  <TASK>
[  127.322524]  ? __warn+0x9f/0x1a0
[  127.322531]  ? __nf_unregister_net_hook+0x21a/0x260
[  127.322537]  ? report_bug+0x1b1/0x1e0
[  127.322545]  ? handle_bug+0x3c/0x70
[  127.322552]  ? exc_invalid_op+0x17/0x40
[  127.322556]  ? asm_exc_invalid_op+0x1a/0x20
[  127.322563]  ? kasan_save_free_info+0x3b/0x60
[  127.322570]  ? __nf_unregister_net_hook+0x6a/0x260
[  127.322577]  ? __nf_unregister_net_hook+0x21a/0x260
[  127.322583]  ? __nf_unregister_net_hook+0x6a/0x260
[  127.322590]  ? __nf_tables_unregister_hook+0x8a/0xe0 [nf_tables]
[  127.322655]  nft_table_disable+0x75/0xf0 [nf_tables]
[  127.322717]  nf_tables_commit+0x2571/0x2620 [nf_tables]

Fixes: 179d9ba5559a ("netfilter: nf_tables: fix table flag updates")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
 net/netfilter/nf_tables_api.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 17bf53cd0e84..1eb51bf24fc2 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -2449,6 +2449,9 @@ static int nf_tables_addchain(struct nft_ctx *ctx, u8 family, u8 genmask,
 		struct nft_stats __percpu *stats = NULL;
 		struct nft_chain_hook hook = {};
 
+		if (table->flags & __NFT_TABLE_F_UPDATE)
+			return -EINVAL;
+
 		if (flags & NFT_CHAIN_BINDING)
 			return -EOPNOTSUPP;
 
-- 
2.30.2

[PATCH net 5/6] netfilter: nf_tables: Fix potential data-race in __nft_flowtable_type_get()

[Pablo Neira Ayuso]

From: Ziyang Xuan <william.xuanziyang@huawei.com>

nft_unregister_flowtable_type() within nf_flow_inet_module_exit() can
concurrent with __nft_flowtable_type_get() within nf_tables_newflowtable().
And thhere is not any protection when iterate over nf_tables_flowtables
list in __nft_flowtable_type_get(). Therefore, there is pertential
data-race of nf_tables_flowtables list entry.

Use list_for_each_entry_rcu() to iterate over nf_tables_flowtables list
in __nft_flowtable_type_get(), and use rcu_read_lock() in the caller
nft_flowtable_type_get() to protect the entire type query process.

Fixes: 3b49e2e94e6e ("netfilter: nf_tables: add flow table netlink frontend")
Signed-off-by: Ziyang Xuan <william.xuanziyang@huawei.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
 net/netfilter/nf_tables_api.c | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 1eb51bf24fc2..e02d0ae4f436 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -8296,11 +8296,12 @@ static int nft_flowtable_parse_hook(const struct nft_ctx *ctx,
 	return err;
 }
 
+/* call under rcu_read_lock */
 static const struct nf_flowtable_type *__nft_flowtable_type_get(u8 family)
 {
 	const struct nf_flowtable_type *type;
 
-	list_for_each_entry(type, &nf_tables_flowtables, list) {
+	list_for_each_entry_rcu(type, &nf_tables_flowtables, list) {
 		if (family == type->family)
 			return type;
 	}
@@ -8312,9 +8313,13 @@ nft_flowtable_type_get(struct net *net, u8 family)
 {
 	const struct nf_flowtable_type *type;
 
+	rcu_read_lock();
 	type = __nft_flowtable_type_get(family);
-	if (type != NULL && try_module_get(type->owner))
+	if (type != NULL && try_module_get(type->owner)) {
+		rcu_read_unlock();
 		return type;
+	}
+	rcu_read_unlock();
 
 	lockdep_nfnl_nft_mutex_not_held();
 #ifdef CONFIG_MODULES
-- 
2.30.2

Re: [PATCH net 5/6] netfilter: nf_tables: Fix potential data-race in __nft_flowtable_type_get()

[Pablo Neira Ayuso]

On Thu, Apr 04, 2024 at 12:43:33PM +0200, Pablo Neira Ayuso wrote:
> From: Ziyang Xuan <william.xuanziyang@huawei.com>
> 
> nft_unregister_flowtable_type() within nf_flow_inet_module_exit() can
> concurrent with __nft_flowtable_type_get() within nf_tables_newflowtable().
> And thhere is not any protection when iterate over nf_tables_flowtables
> list in __nft_flowtable_type_get(). Therefore, there is pertential
> data-race of nf_tables_flowtables list entry.
> 
> Use list_for_each_entry_rcu() to iterate over nf_tables_flowtables list
> in __nft_flowtable_type_get(), and use rcu_read_lock() in the caller
> nft_flowtable_type_get() to protect the entire type query process.

Applied to nf, thanks

[PATCH net 6/6] netfilter: nf_tables: discard table flag update with pending basechain deletion

[Pablo Neira Ayuso]

Hook unregistration is deferred to the commit phase, same occurs with
hook updates triggered by the table dormant flag. When both commands are
combined, this results in deleting a basechain while leaving its hook
still registered in the core.

Fixes: 179d9ba5559a ("netfilter: nf_tables: fix table flag updates")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
---
 net/netfilter/nf_tables_api.c | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index e02d0ae4f436..d89d77946719 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -1209,10 +1209,11 @@ static bool nft_table_pending_update(const struct nft_ctx *ctx)
 		return true;
 
 	list_for_each_entry(trans, &nft_net->commit_list, list) {
-		if ((trans->msg_type == NFT_MSG_NEWCHAIN ||
-		     trans->msg_type == NFT_MSG_DELCHAIN) &&
-		    trans->ctx.table == ctx->table &&
-		    nft_trans_chain_update(trans))
+		if (trans->ctx.table == ctx->table &&
+		    ((trans->msg_type == NFT_MSG_NEWCHAIN &&
+		      nft_trans_chain_update(trans)) ||
+		     (trans->msg_type == NFT_MSG_DELCHAIN &&
+		      nft_is_base_chain(trans->ctx.chain))))
 			return true;
 	}
 
-- 
2.30.2