From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail.netfilter.org (mail.netfilter.org [217.70.190.124]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id 33F074EBAD2 for ; Thu, 3 Sep 2026 16:23:01 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=217.70.190.124 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1788452583; cv=none; b=g3Z2oC5rrdwoyNmQa9ChgGSskuK+5g7CeY/zQkS0fhRWwdBS/xKSY4L6uXnqE9kZmLt66fIuvoTBVWy1LSlxWLnFIuZS4G2GLwJV7PMcl6eRqSkmeMGBdMnuRbOSfHy/B/Kvw7BDHJ2OAPnMPUFgaAwAIgieM+h5IB1VWWCa3U8= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1788452583; c=relaxed/simple; bh=SMtKCR//ak7YJrNqlxAzFyR5v4lNhUmPACGmSZvNKkE=; h=Date:From:To:Cc:Subject:Message-ID:References:MIME-Version: Content-Type:Content-Disposition:In-Reply-To; b=BBXNntKAk4bULt22LFV0CCF+kIYpkoEb6HE9EfLt5Ane+y11Y7TJSv8ay/VZiAqS2QkHuzsbTVUQq/uHqBU869KyHF4GaCbugfNxVzwUAJh6rTt+lm/SNDtxsl0lmfUuDKfrBLxKCsW9JLC11Ao/jmDtAkgRx/GOiYWuL84ObJQ= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=netfilter.org; spf=pass smtp.mailfrom=netfilter.org; dkim=pass (2048-bit key) header.d=netfilter.org header.i=@netfilter.org header.b=vgko5Agu; arc=none smtp.client-ip=217.70.190.124 Authentication-Results: smtp.subspace.kernel.org; dmarc=none (p=none dis=none) header.from=netfilter.org Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=netfilter.org Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=netfilter.org header.i=@netfilter.org header.b="vgko5Agu" DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=netfilter.org; s=2025; t=1788452578; bh=n8U8FrNBTCp/jltc8tpuFl02Rk7DZcc7Obj5g1UoshQ=; h=Date:From:To:Cc:Subject:References:In-Reply-To:From; b=vgko5AguW5bjnvFGwmPOa2S0FzOdKTMNICcmyNNZ95TzzRluIfk4UrZdgQrVtLbUp YloAHl5tgRJKGXAECmZ3mcOV95tU3h13MVuOrPD4tNuSDWeVCuZzJ3MY3bpfE8KYLe 0UVTj+O6TUVYWzhaDVusJ7OiQUPAzXzdIUUNcgJYuzdzqwUKuybYwWNCpImKSJfXYr jQdSPjAcAVYyGku5IDLkk2jfqxopM07uDEJ70aiNAW/+HftNxNNJqaeDrAJo5R0zzo HL5JhhvT44RP3+3oMUScpeQ8vwbDG47dfhry3yfoOZddBQmMsrELbc6jx9PH9PLI63 8WI2BqNAnwEVg== Received: from netfilter.org (mail-agni [217.70.190.124]) by mail.netfilter.org (Postfix) with UTF8SMTPSA id 49C59607C2; Thu, 3 Sep 2026 18:22:58 +0200 (CEST) Date: Thu, 3 Sep 2026 18:22:55 +0200 From: Pablo Neira Ayuso To: Phil Sutter Cc: netfilter-devel@vger.kernel.org Subject: Re: [nf PATCH v2] netfilter: nfnetlink: Fix for interrupted hook dumps Message-ID: References: <20260903143921.942724-1-phil@nwl.cc> Precedence: bulk X-Mailing-List: netfilter-devel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <20260903143921.942724-1-phil@nwl.cc> Hi Phil, Thanks for your patch, see comments below. On Thu, Sep 03, 2026 at 04:39:21PM +0200, Phil Sutter wrote: > Handling of concurrent hook changes with a dump in progress was > problematic in nfnl_hook_dump and entirely broken in nfnl_hook_dump_nat. > Address all issues in a single patch to please the review LLM. > > Introduce sequence numbers in struct netns_nf to replace pointer > value-based modification detection which may fail due to memory buffer > reuse. This also eliminates the need for most manual cb->seq updates and > excessive array index value checks. > > Since nat hook updates are protected by a different mutex than others, > they have to bump their own sequence number. nfnl_hook_dump_nat > therefore open-codes what nl_dump_check_consistent does but bumps > cb->seq instead of setting NLM_F_DUMP_INTR flag. > > Dumps of many nat hooks was entirely broken since the array index was > not (re)stored. Use cb->args[1] for that and make sure it is reset upon > completion as the same dump may loop over multiple arrays of nat hooks. > > In nfnl_hook_dump, don't signal NLM_F_DUMP_INTR if > nfnl_hook_entries_head returns error: This is a permanent condition > which does not change during a dump. It is already caught by > nfnl_hook_dump_start though, so should not happen anyway. > > In general, access ops array pointer values using READ_ONCE since they > are assigned to using WRITE_ONCE. > > Avoid setting NLM_F_DUMP_INTR flag in garbage memory by calling > nl_dump_check_consistent only for non-empty skbs. If not a single > netlink message was created, nfnetlink code will take care of setting > the flag. > > Fixes: e2cf17d3774c ("netfilter: add new hook nfnl subsystem") > Fixes: b010e2a4a9ac ("netfilter: nfnetlink_hook: Dump nat type chains") > Signed-off-by: Phil Sutter > --- > Changes since v1: > - Reset cursor value and perform EINTR check if NAT priv->entries > becomes NULL during dump. > --- > include/net/netns/netfilter.h | 2 + > net/netfilter/core.c | 13 ++++++ > net/netfilter/nf_nat_core.c | 14 ++++++- > net/netfilter/nfnetlink_hook.c | 75 ++++++++++++++++++++-------------- > 4 files changed, 72 insertions(+), 32 deletions(-) > > diff --git a/include/net/netns/netfilter.h b/include/net/netns/netfilter.h > index a6a0bf4a247e..7fd78394d1e7 100644 > --- a/include/net/netns/netfilter.h > +++ b/include/net/netns/netfilter.h > @@ -33,5 +33,7 @@ struct netns_nf { > #if IS_ENABLED(CONFIG_NF_DEFRAG_IPV6) > unsigned int defrag_ipv6_users; > #endif > + unsigned int hook_base_seq; > + unsigned int nat_hook_base_seq; > }; > #endif > diff --git a/net/netfilter/core.c b/net/netfilter/core.c > index 675a1034b340..a284b3241211 100644 > --- a/net/netfilter/core.c > +++ b/net/netfilter/core.c > @@ -386,6 +386,15 @@ static void nf_static_key_dec(const struct nf_hook_ops *reg, int pf) > #endif > } > > +static void bump_hook_base_seq(struct net *net) > +{ > + unsigned int base_seq = READ_ONCE(net->nf.hook_base_seq); > + > + while (++base_seq == 0) > + ; > + smp_store_release(&net->nf.hook_base_seq, base_seq); > +} > + > static int __nf_register_net_hook(struct net *net, int pf, > const struct nf_hook_ops *reg) > { > @@ -428,6 +437,7 @@ static int __nf_register_net_hook(struct net *net, int pf, > new_hooks = nf_hook_entries_grow(p, reg); > > if (!IS_ERR(new_hooks)) { > + bump_hook_base_seq(net); > hooks_validate(new_hooks); > rcu_assign_pointer(*pp, new_hooks); I think you have to call bump_hook_base_seq(net) here after rcu_assign_pointer()? > } > @@ -506,6 +516,7 @@ static void __nf_unregister_net_hook(struct net *net, int pf, > net_dec_egress_queue(); > #endif > nf_static_key_dec(reg, pf); > + bump_hook_base_seq(net); > } else { > WARN_ONCE(1, "hook not found, pf %d num %d", pf, reg->hooknum); > } > @@ -784,6 +795,8 @@ static int __net_init netfilter_net_init(struct net *net) > return -ENOMEM; > } > #endif > + net->nf.hook_base_seq = 1; > + net->nf.nat_hook_base_seq = 1; > > return 0; > } > diff --git a/net/netfilter/nf_nat_core.c b/net/netfilter/nf_nat_core.c > index 8ac326e1eb5b..ff2cba8df886 100644 > --- a/net/netfilter/nf_nat_core.c > +++ b/net/netfilter/nf_nat_core.c > @@ -1160,6 +1160,15 @@ nfnetlink_parse_nat_setup(struct nf_conn *ct, > } > #endif > > +static void bump_nat_hook_base_seq(struct net *net) > +{ > + unsigned int base_seq = READ_ONCE(net->nf.nat_hook_base_seq); > + > + while (++base_seq == 0) > + ; > + smp_store_release(&net->nf.nat_hook_base_seq, base_seq); > +} > + > static struct nf_ct_helper_expectfn follow_master_nat = { > .name = "nat-follow-master", > .expectfn = nf_nat_follow_master, > @@ -1245,8 +1254,10 @@ int nf_nat_register_fn(struct net *net, u8 pf, const struct nf_hook_ops *ops, > } > > ret = nf_hook_entries_insert_raw(&priv->entries, ops); > - if (ret == 0) > + if (ret == 0) { > + bump_nat_hook_base_seq(net); > nat_proto_net->users++; Sashiko refers to another issue that is fixed here: https://patchwork.ozlabs.org/project/netfilter-devel/patch/20260902232856.1024959-1-pablo@netfilter.org/ This patch needs a rebase on top of that fix above. Maybe it is good to get the two patches in the same batch. I think fixing the ordering in which base_seq is bumped (after publishing the new hook) should be fine. A few more comments below. > + } > > mutex_unlock(&nf_nat_proto_mutex); > return ret; > @@ -1284,6 +1295,7 @@ void nf_nat_unregister_fn(struct net *net, u8 pf, const struct nf_hook_ops *ops, > goto unlock; > priv = nat_ops[hooknum].priv; > nf_hook_entries_delete_raw(&priv->entries, ops); > + bump_nat_hook_base_seq(net); > > if (nat_proto_net->users == 0) { > nf_unregister_net_hooks(net, nat_ops, ops_count); > diff --git a/net/netfilter/nfnetlink_hook.c b/net/netfilter/nfnetlink_hook.c > index 95005e9a6066..21a759168c38 100644 > --- a/net/netfilter/nfnetlink_hook.c > +++ b/net/netfilter/nfnetlink_hook.c > @@ -54,7 +54,6 @@ static int nf_netlink_dump_start_rcu(struct sock *nlsk, struct sk_buff *skb, > > struct nfnl_dump_hook_data { > char devname[IFNAMSIZ]; > - unsigned long headv; > u8 hook; > }; > > @@ -338,27 +337,48 @@ nfnl_hook_entries_head(u8 pf, unsigned int hook, struct net *net, const char *de > } > > static int nfnl_hook_dump_nat(struct sk_buff *nlskb, > - const struct nfnl_dump_hook_data *ctx, > - const struct nf_hook_ops *ops, > - int family, unsigned int seq) > + struct netlink_callback *cb, > + const struct nf_hook_ops *ops, int family) > { > struct nf_nat_lookup_hook_priv *priv = ops->priv; > - struct nf_hook_entries *e = rcu_dereference(priv->entries); > + struct nfnl_dump_hook_data *ctx = cb->data; > + struct net *net = sock_net(nlskb->sk); > struct nf_hook_ops **nat_ops; > - int i, err; > + unsigned int i = cb->args[1]; > + struct nf_hook_entries *e; > + unsigned int base_seq; > + int err = 0; > > + base_seq = smp_load_acquire(&net->nf.nat_hook_base_seq); > + > + e = rcu_dereference(priv->entries); > if (!e) > - return 0; > + goto out; > > nat_ops = nf_hook_entries_get_hook_ops(e); > > - for (i = 0; i < e->num_hook_entries; i++) { > - err = nfnl_hook_dump_one(nlskb, ctx, nat_ops[i], > - ops->priority, family, seq); > + for (; i < e->num_hook_entries; i++) { > + err = nfnl_hook_dump_one(nlskb, ctx, > + READ_ONCE(nat_ops[i]), > + ops->priority, family, > + cb->nlh->nlmsg_seq); > if (err) > - return err; > + break; > + > } > - return 0; > +out: > + if (!err) { > + i = 0; > + } Comestic: This curly braces can be removed for single statement. > + cb->args[1] = i; > + > + if (cb->args[2] && base_seq != cb->args[2]) { > + cb->seq++; > + err = -EINTR; > + } > + cb->args[2] = base_seq; > + > + return err; > } > > static int nfnl_hook_dump(struct sk_buff *nlskb, > @@ -373,35 +393,31 @@ static int nfnl_hook_dump(struct sk_buff *nlskb, > unsigned int i = cb->args[0]; > > rcu_read_lock(); > + cb->seq = smp_load_acquire(&net->nf.hook_base_seq); > > e = nfnl_hook_entries_head(family, ctx->hook, net, ctx->devname); > - if (!e) > + if (!e || IS_ERR(e)) > goto done; > > - if (IS_ERR(e)) { > - cb->seq++; > - goto done; > - } > - > - if ((unsigned long)e != ctx->headv || i >= e->num_hook_entries) > - cb->seq++; > - > ops = nf_hook_entries_get_hook_ops(e); > > for (; i < e->num_hook_entries; i++) { > - if (ops[i]->hook_ops_type == NF_HOOK_OP_NAT) > - err = nfnl_hook_dump_nat(nlskb, ctx, ops[i], family, > - cb->nlh->nlmsg_seq); > - else > - err = nfnl_hook_dump_one(nlskb, ctx, ops[i], > - ops[i]->priority, family, > + const struct nf_hook_ops *cur = READ_ONCE(ops[i]); > + > + if (cur->hook_ops_type == NF_HOOK_OP_NAT) > + err = nfnl_hook_dump_nat(nlskb, cb, cur, family); Missing curly brace for consistency with the else below (IIRC, if either if () or else is multistatement or multiline then curly braces are recommended. > + else { > + err = nfnl_hook_dump_one(nlskb, ctx, cur, > + cur->priority, family, > cb->nlh->nlmsg_seq); > + } > if (err) > break; > } > > done: > - nl_dump_check_consistent(cb, nlmsg_hdr(nlskb)); > + if (nlskb->len > 0) Is this new check needed? I think the consistency check can be performed inconditionally? > + nl_dump_check_consistent(cb, nlmsg_hdr(nlskb)); > rcu_read_unlock(); > cb->args[0] = i; > return nlskb->len; > @@ -442,10 +458,7 @@ static int nfnl_hook_dump_start(struct netlink_callback *cb) > return -ENOMEM; > > strscpy(ctx->devname, name, sizeof(ctx->devname)); > - ctx->headv = (unsigned long)head; > ctx->hook = hooknum; > - > - cb->seq = 1; > cb->data = ctx; > > return 0; > -- > 2.54.0 >