patches.lists.linux.dev archive mirror
 help / color / mirror / Atom feed
From: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To: stable@vger.kernel.org, netfilter-devel@vger.kernel.org
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
	patches@lists.linux.dev,
	syzbot+b26935466701e56cfdc2@syzkaller.appspotmail.com,
	Florian Westphal <fw@strlen.de>,
	Pablo Neira Ayuso <pablo@netfilter.org>
Subject: [PATCH 5.15 58/59] netfilter: nf_tables: do not defer rule destruction via call_rcu
Date: Tue, 20 May 2025 15:50:49 +0200	[thread overview]
Message-ID: <20250520125756.144291818@linuxfoundation.org> (raw)
In-Reply-To: <20250520125753.836407405@linuxfoundation.org>

5.15-stable review patch.  If anyone has any objections, please let me know.

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

From: Florian Westphal <fw@strlen.de>

commit b04df3da1b5c6f6dc7cdccc37941740c078c4043 upstream.

nf_tables_chain_destroy can sleep, it can't be used from call_rcu
callbacks.

Moreover, nf_tables_rule_release() is only safe for error unwinding,
while transaction mutex is held and the to-be-desroyed rule was not
exposed to either dataplane or dumps, as it deactives+frees without
the required synchronize_rcu() in-between.

nft_rule_expr_deactivate() callbacks will change ->use counters
of other chains/sets, see e.g. nft_lookup .deactivate callback, these
must be serialized via transaction mutex.

Also add a few lockdep asserts to make this more explicit.

Calling synchronize_rcu() isn't ideal, but fixing this without is hard
and way more intrusive.  As-is, we can get:

WARNING: .. net/netfilter/nf_tables_api.c:5515 nft_set_destroy+0x..
Workqueue: events nf_tables_trans_destroy_work
RIP: 0010:nft_set_destroy+0x3fe/0x5c0
Call Trace:
 <TASK>
 nf_tables_trans_destroy_work+0x6b7/0xad0
 process_one_work+0x64a/0xce0
 worker_thread+0x613/0x10d0

In case the synchronize_rcu becomes an issue, we can explore alternatives.

One way would be to allocate nft_trans_rule objects + one nft_trans_chain
object, deactivate the rules + the chain and then defer the freeing to the
nft destroy workqueue.  We'd still need to keep the synchronize_rcu path as
a fallback to handle -ENOMEM corner cases though.

Reported-by: syzbot+b26935466701e56cfdc2@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/all/67478d92.050a0220.253251.0062.GAE@google.com/T/
Fixes: c03d278fdf35 ("netfilter: nf_tables: wait for rcu grace period on net_device removal")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
---
 include/net/netfilter/nf_tables.h |    3 ---
 net/netfilter/nf_tables_api.c     |   32 +++++++++++++++-----------------
 2 files changed, 15 insertions(+), 20 deletions(-)

--- a/include/net/netfilter/nf_tables.h
+++ b/include/net/netfilter/nf_tables.h
@@ -1028,7 +1028,6 @@ struct nft_chain {
 	char				*name;
 	u16				udlen;
 	u8				*udata;
-	struct rcu_head			rcu_head;
 
 	/* Only used during control plane commit phase: */
 	struct nft_rule			**rules_next;
@@ -1171,7 +1170,6 @@ static inline void nft_use_inc_restore(u
  *	@sets: sets in the table
  *	@objects: stateful objects in the table
  *	@flowtables: flow tables in the table
- *	@net: netnamespace this table belongs to
  *	@hgenerator: handle generator state
  *	@handle: table handle
  *	@use: number of chain references to this table
@@ -1187,7 +1185,6 @@ struct nft_table {
 	struct list_head		sets;
 	struct list_head		objects;
 	struct list_head		flowtables;
-	possible_net_t			net;
 	u64				hgenerator;
 	u64				handle;
 	u32				use;
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -1360,7 +1360,6 @@ static int nf_tables_newtable(struct sk_
 	INIT_LIST_HEAD(&table->sets);
 	INIT_LIST_HEAD(&table->objects);
 	INIT_LIST_HEAD(&table->flowtables);
-	write_pnet(&table->net, net);
 	table->family = family;
 	table->flags = flags;
 	table->handle = ++nft_net->table_handle;
@@ -3441,8 +3440,11 @@ void nf_tables_rule_destroy(const struct
 	kfree(rule);
 }
 
+/* can only be used if rule is no longer visible to dumps */
 static void nf_tables_rule_release(const struct nft_ctx *ctx, struct nft_rule *rule)
 {
+	lockdep_commit_lock_is_held(ctx->net);
+
 	nft_rule_expr_deactivate(ctx, rule, NFT_TRANS_RELEASE);
 	nf_tables_rule_destroy(ctx, rule);
 }
@@ -5178,6 +5180,8 @@ void nf_tables_deactivate_set(const stru
 			      struct nft_set_binding *binding,
 			      enum nft_trans_phase phase)
 {
+	lockdep_commit_lock_is_held(ctx->net);
+
 	switch (phase) {
 	case NFT_TRANS_PREPARE_ERROR:
 		nft_set_trans_unbind(ctx, set);
@@ -10440,19 +10444,6 @@ static void __nft_release_basechain_now(
 	nf_tables_chain_destroy(ctx->chain);
 }
 
-static void nft_release_basechain_rcu(struct rcu_head *head)
-{
-	struct nft_chain *chain = container_of(head, struct nft_chain, rcu_head);
-	struct nft_ctx ctx = {
-		.family	= chain->table->family,
-		.chain	= chain,
-		.net	= read_pnet(&chain->table->net),
-	};
-
-	__nft_release_basechain_now(&ctx);
-	put_net(ctx.net);
-}
-
 int __nft_release_basechain(struct nft_ctx *ctx)
 {
 	struct nft_rule *rule;
@@ -10467,11 +10458,18 @@ int __nft_release_basechain(struct nft_c
 	nft_chain_del(ctx->chain);
 	nft_use_dec(&ctx->table->use);
 
-	if (maybe_get_net(ctx->net))
-		call_rcu(&ctx->chain->rcu_head, nft_release_basechain_rcu);
-	else
+	if (!maybe_get_net(ctx->net)) {
 		__nft_release_basechain_now(ctx);
+		return 0;
+	}
+
+	/* wait for ruleset dumps to complete.  Owning chain is no longer in
+	 * lists, so new dumps can't find any of these rules anymore.
+	 */
+	synchronize_rcu();
 
+	__nft_release_basechain_now(ctx);
+	put_net(ctx->net);
 	return 0;
 }
 EXPORT_SYMBOL_GPL(__nft_release_basechain);



  parent reply	other threads:[~2025-05-20 13:54 UTC|newest]

Thread overview: 71+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2025-05-20 13:49 [PATCH 5.15 00/59] 5.15.184-rc1 review Greg Kroah-Hartman
2025-05-20 13:49 ` [PATCH 5.15 01/59] platform/x86: asus-wmi: Fix wlan_ctrl_by_user detection Greg Kroah-Hartman
2025-05-20 13:49 ` [PATCH 5.15 02/59] tracing: probes: Fix a possible race in trace_probe_log APIs Greg Kroah-Hartman
2025-05-20 13:49 ` [PATCH 5.15 03/59] iio: adc: ad7768-1: Fix insufficient alignment of timestamp Greg Kroah-Hartman
2025-05-20 13:49 ` [PATCH 5.15 04/59] iio: chemical: sps30: use aligned_s64 for timestamp Greg Kroah-Hartman
2025-05-20 13:49 ` [PATCH 5.15 05/59] RDMA/rxe: Fix slab-use-after-free Read in rxe_queue_cleanup bug Greg Kroah-Hartman
2025-05-20 13:49 ` [PATCH 5.15 06/59] nfs: handle failure of nfs_get_lock_context in unlock path Greg Kroah-Hartman
2025-05-20 13:49 ` [PATCH 5.15 07/59] spi: loopback-test: Do not split 1024-byte hexdumps Greg Kroah-Hartman
2025-05-20 13:49 ` [PATCH 5.15 08/59] net_sched: Flush gso_skb list too during ->change() Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 09/59] net: cadence: macb: Fix a possible deadlock in macb_halt_tx Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 10/59] net: dsa: sja1105: discard incoming frames in BR_STATE_LISTENING Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 11/59] ALSA: sh: SND_AICA should depend on SH_DMA_API Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 12/59] qlcnic: fix memory leak in qlcnic_sriov_channel_cfg_cmd() Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 13/59] NFSv4/pnfs: Reset the layout state after a layoutreturn Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 14/59] x86,nospec: Simplify {JMP,CALL}_NOSPEC Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 15/59] x86/speculation: Simplify and make CALL_NOSPEC consistent Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 16/59] x86/speculation: Add a conditional CS prefix to CALL_NOSPEC Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 17/59] x86/speculation: Remove the extra #ifdef around CALL_NOSPEC Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 18/59] Documentation: x86/bugs/its: Add ITS documentation Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 19/59] x86/its: Enumerate Indirect Target Selection (ITS) bug Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 20/59] x86/its: Add support for ITS-safe indirect thunk Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 21/59] x86/alternative: Optimize returns patching Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 22/59] x86/alternatives: Remove faulty optimization Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 23/59] x86/its: Add support for ITS-safe return thunk Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 24/59] x86/its: Enable Indirect Target Selection mitigation Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 25/59] x86/its: Add "vmexit" option to skip mitigation on some CPUs Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 26/59] x86/its: Align RETs in BHB clear sequence to avoid thunking Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 27/59] x86/its: Use dynamic thunks for indirect branches Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 28/59] x86/its: Fix build errors when CONFIG_MODULES=n Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 29/59] x86/its: FineIBT-paranoid vs ITS Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 30/59] dmaengine: Revert "dmaengine: dmatest: Fix dmatest waiting less when interrupted" Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 31/59] btrfs: fix discard worker infinite loop after disabling discard Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 32/59] ACPI: PPTT: Fix processor subtable walk Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 33/59] ALSA: es1968: Add error handling for snd_pcm_hw_constraint_pow2() Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 34/59] ALSA: usb-audio: Add sample rate quirk for Audioengine D1 Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 35/59] ALSA: usb-audio: Add sample rate quirk for Microdia JP001 USB Camera Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 36/59] ftrace: Fix preemption accounting for stacktrace trigger command Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 37/59] ftrace: Fix preemption accounting for stacktrace filter command Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 38/59] tracing: samples: Initialize trace_array_printk() with the correct function Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 39/59] phy: Fix error handling in tegra_xusb_port_init Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 40/59] phy: renesas: rcar-gen3-usb2: Set timing registers only once Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 41/59] wifi: mt76: disable napi on driver removal Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 42/59] dmaengine: ti: k3-udma: Add missing locking Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 43/59] dmaengine: ti: k3-udma: Use cap_mask directly from dma_device structure instead of a local copy Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 44/59] dmaengine: idxd: fix memory leak in error handling path of idxd_setup_engines Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 45/59] dmaengine: idxd: fix memory leak in error handling path of idxd_setup_groups Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 46/59] block: fix direct io NOWAIT flag not work Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 47/59] clocksource/i8253: Use raw_spinlock_irqsave() in clockevent_i8253_disable() Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 48/59] usb: typec: ucsi: displayport: Fix deadlock Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 49/59] usb: typec: altmodes/displayport: create sysfs nodes as drivers default device attribute group Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 50/59] usb: typec: fix potential array underflow in ucsi_ccg_sync_control() Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 51/59] usb: typec: fix pm usage counter imbalance " Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 52/59] selftests/mm: compaction_test: support platform with huge mount of memory Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 53/59] sctp: add mutual exclusion in proc_sctp_do_udp_port() Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 54/59] btrfs: dont BUG_ON() when 0 reference count at btrfs_lookup_extent_info() Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 55/59] btrfs: do not clean up repair bio if submit fails Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 56/59] netfilter: nf_tables: pass nft_chain to destroy function, not nft_ctx Greg Kroah-Hartman
2025-05-20 13:50 ` [PATCH 5.15 57/59] netfilter: nf_tables: wait for rcu grace period on net_device removal Greg Kroah-Hartman
2025-05-20 13:50 ` Greg Kroah-Hartman [this message]
2025-05-20 13:50 ` [PATCH 5.15 59/59] ice: arfs: fix use-after-free when freeing @rx_cpu_rmap Greg Kroah-Hartman
2025-05-20 18:19 ` [PATCH 5.15 00/59] 5.15.184-rc1 review Florian Fainelli
2025-05-20 22:46 ` Shuah Khan
2025-05-21  1:53 ` Ron Economos
2025-05-21  3:16 ` Vijayendra Suman
2025-05-21  8:30 ` Jon Hunter
2025-05-21 12:39 ` Naresh Kamboju
2025-05-21 18:54 ` Mark Brown
2025-05-21 19:10 ` Alexandre Chartre
2025-05-21 21:25   ` Pawan Gupta
2025-05-22  5:09 ` Hardik Garg
2025-05-23  9:25 ` Guenter Roeck

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20250520125756.144291818@linuxfoundation.org \
    --to=gregkh@linuxfoundation.org \
    --cc=fw@strlen.de \
    --cc=netfilter-devel@vger.kernel.org \
    --cc=pablo@netfilter.org \
    --cc=patches@lists.linux.dev \
    --cc=stable@vger.kernel.org \
    --cc=syzbot+b26935466701e56cfdc2@syzkaller.appspotmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).