All of lore.kernel.org
 help / color / mirror / Atom feed
From: Pablo Neira Ayuso <pablo@netfilter.org>
To: netfilter-devel@vger.kernel.org
Cc: sashal@kernel.org, gregkh@linuxfoundation.org, stable@vger.kernel.org
Subject: [PATCH -stable,5.4 01/10] netfilter: nf_tables: fix nat hook table deletion
Date: Wed,  5 Jul 2023 18:54:14 +0200	[thread overview]
Message-ID: <20230705165423.50054-2-pablo@netfilter.org> (raw)
In-Reply-To: <20230705165423.50054-1-pablo@netfilter.org>

From: Florian Westphal <fw@strlen.de>

[ 1e9451cbda456a170518b2bfd643e2cb980880bf ]

sybot came up with following transaction:
 add table ip syz0
 add chain ip syz0 syz2 { type nat hook prerouting priority 0; policy accept; }
 add table ip syz0 { flags dormant; }
 delete chain ip syz0 syz2
 delete table ip syz0

which yields:
hook not found, pf 2 num 0
WARNING: CPU: 0 PID: 6775 at net/netfilter/core.c:413 __nf_unregister_net_hook+0x3e6/0x4a0 net/netfilter/core.c:413
[..]
 nft_unregister_basechain_hooks net/netfilter/nf_tables_api.c:206 [inline]
 nft_table_disable net/netfilter/nf_tables_api.c:835 [inline]
 nf_tables_table_disable net/netfilter/nf_tables_api.c:868 [inline]
 nf_tables_commit+0x32d3/0x4d70 net/netfilter/nf_tables_api.c:7550
 nfnetlink_rcv_batch net/netfilter/nfnetlink.c:486 [inline]
 nfnetlink_rcv_skb_batch net/netfilter/nfnetlink.c:544 [inline]
 nfnetlink_rcv+0x14a5/0x1e50 net/netfilter/nfnetlink.c:562
 netlink_unicast_kernel net/netlink/af_netlink.c:1303 [inline]

Problem is that when I added ability to override base hook registration
to make nat basechains register with the nat core instead of netfilter
core, I forgot to update nft_table_disable() to use that instead of
the 'raw' hook register interface.

In syzbot transaction, the basechain is of 'nat' type. Its registered
with the nat core.  The switch to 'dormant mode' attempts to delete from
netfilter core instead.

After updating nft_table_disable/enable to use the correct helper,
nft_(un)register_basechain_hooks can be folded into the only remaining
caller.

Because nft_trans_table_enable() won't do anything when the DORMANT flag
is set, remove the flag first, then re-add it in case re-enablement
fails, else this patch breaks sequence:

add table ip x { flags dormant; }
/* add base chains */
add table ip x

The last 'add' will remove the dormant flags, but won't have any other
effect -- base chains are not registered.
Then, next 'set dormant flag' will create another 'hook not found'
splat.

Reported-by: syzbot+2570f2c036e3da5db176@syzkaller.appspotmail.com
Fixes: 4e25ceb80b58 ("netfilter: nf_tables: allow chain type to override hook register")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
(cherry picked from commit 1e9451cbda456a170518b2bfd643e2cb980880bf)
---
 net/netfilter/nf_tables_api.c | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/net/netfilter/nf_tables_api.c b/net/netfilter/nf_tables_api.c
index 914fbd9ecef9..dfa1820ed032 100644
--- a/net/netfilter/nf_tables_api.c
+++ b/net/netfilter/nf_tables_api.c
@@ -770,7 +770,7 @@ static void nft_table_disable(struct net *net, struct nft_table *table, u32 cnt)
 		if (cnt && i++ == cnt)
 			break;
 
-		nf_unregister_net_hook(net, &nft_base_chain(chain)->ops);
+		nf_tables_unregister_hook(net, table, chain);
 	}
 }
 
@@ -785,7 +785,7 @@ static int nf_tables_table_enable(struct net *net, struct nft_table *table)
 		if (!nft_is_base_chain(chain))
 			continue;
 
-		err = nf_register_net_hook(net, &nft_base_chain(chain)->ops);
+		err = nf_tables_register_hook(net, table, chain);
 		if (err < 0)
 			goto err;
 
@@ -829,11 +829,12 @@ static int nf_tables_updtable(struct nft_ctx *ctx)
 		nft_trans_table_enable(trans) = false;
 	} else if (!(flags & NFT_TABLE_F_DORMANT) &&
 		   ctx->table->flags & NFT_TABLE_F_DORMANT) {
+		ctx->table->flags &= ~NFT_TABLE_F_DORMANT;
 		ret = nf_tables_table_enable(ctx->net, ctx->table);
-		if (ret >= 0) {
-			ctx->table->flags &= ~NFT_TABLE_F_DORMANT;
+		if (ret >= 0)
 			nft_trans_table_enable(trans) = true;
-		}
+		else
+			ctx->table->flags |= NFT_TABLE_F_DORMANT;
 	}
 	if (ret < 0)
 		goto err;
-- 
2.30.2


  reply	other threads:[~2023-07-05 16:54 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-07-05 16:54 [PATCH -stable,5.4 00/10] stable fixes for 5.4 Pablo Neira Ayuso
2023-07-05 16:54 ` Pablo Neira Ayuso [this message]
2023-07-05 16:54 ` [PATCH -stable,5.4 02/10] netfilter: nftables: add helper function to set the base sequence number Pablo Neira Ayuso
2023-07-05 16:54 ` [PATCH -stable,5.4 03/10] netfilter: add helper function to set up the nfnetlink header and use it Pablo Neira Ayuso
2023-07-05 16:54 ` [PATCH -stable,5.4 04/10] netfilter: nf_tables: use net_generic infra for transaction data Pablo Neira Ayuso
2023-07-05 16:54 ` [PATCH -stable,5.4 05/10] netfilter: nf_tables: add rescheduling points during loop detection walks Pablo Neira Ayuso
2023-07-05 16:54 ` [PATCH -stable,5.4 06/10] netfilter: nf_tables: incorrect error path handling with NFT_MSG_NEWRULE Pablo Neira Ayuso
2023-07-05 16:54 ` [PATCH -stable,5.4 07/10] netfilter: nf_tables: add NFT_TRANS_PREPARE_ERROR to deal with bound set/chain Pablo Neira Ayuso
2023-07-05 16:54 ` [PATCH -stable,5.4 08/10] netfilter: nf_tables: reject unbound anonymous set before commit phase Pablo Neira Ayuso
2023-07-05 16:54 ` [PATCH -stable,5.4 09/10] netfilter: nf_tables: unbind non-anonymous set if rule construction fails Pablo Neira Ayuso
2023-07-05 16:54 ` [PATCH -stable,5.4 10/10] netfilter: nf_tables: fix scheduling-while-atomic splat Pablo Neira Ayuso
2023-07-16 19:22 ` [PATCH -stable,5.4 00/10] stable fixes for 5.4 Greg KH

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=20230705165423.50054-2-pablo@netfilter.org \
    --to=pablo@netfilter.org \
    --cc=gregkh@linuxfoundation.org \
    --cc=netfilter-devel@vger.kernel.org \
    --cc=sashal@kernel.org \
    --cc=stable@vger.kernel.org \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.