From: Chengfeng Ye <nicoyip.dev@gmail.com>
To: Eric Dumazet <edumazet@google.com>,
Kuniyuki Iwashima <kuniyu@google.com>,
Paolo Abeni <pabeni@redhat.com>,
Willem de Bruijn <willemb@google.com>,
"David S. Miller" <davem@davemloft.net>,
Jakub Kicinski <kuba@kernel.org>, Simon Horman <horms@kernel.org>
Cc: netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
Chengfeng Ye <nicoyip.dev@gmail.com>,
stable@vger.kernel.org
Subject: [PATCH net v3] net: pin protocol module before inet socket allocation
Date: Fri, 4 Sep 2026 19:15:14 +0800 [thread overview]
Message-ID: <20260904111514.584264-1-nicoyip.dev@gmail.com> (raw)
inet_create() and inet6_create() look up the protocol under
rcu_read_lock(), then drop RCU before using the resulting proto.
sk_alloc() uses GFP_KERNEL, so RCU cannot be held across it.
A loadable protocol can be unregistered in that window.
inet_unregister_protosw() waits with synchronize_net() only for readers
still in the RCU section. After rcu_read_unlock(), module exit can run
proto_unregister(), destroy prot->slab, and free the module. inet_create()
then uses a dangling proto pointer:
CPU 0 inet_create CPU 1 l2tp_ip_exit
rcu_read_lock()
answer_prot = answer->prot
rcu_read_unlock() inet_unregister_protosw()
synchronize_net()
proto_unregister()
kmem_cache_destroy(slab)
WARN_ON(!answer_prot->slab)
sk_alloc() -> kmem_cache_alloc(stale)
This was reproduced with socket(AF_INET, SOCK_DGRAM, IPPROTO_L2TP)
racing delete_module("l2tp_ip"):
Oops: general protection fault, probably for non-canonical address
KASAN: maybe wild-memory-access in range
RIP: kmem_cache_alloc_noprof+0x63/0x370
Call Trace:
sk_prot_alloc+0x74/0x2c0
sk_alloc+0x2b/0x6c0
inet_create+0x2cd/0xd40
__sock_create+0x1c3/0x430
__sys_socket+0x116/0x1d0
A protosw can also be published while its owner is in
MODULE_STATE_COMING. try_module_get() accepts modules in that state, but
module initialization failure does not wait for such references before
freeing the module. A protocol initialization unwind can therefore destroy
the slab and module image despite the reference.
Cache the proto and its owner while still under RCU. Reject an owner in
MODULE_STATE_COMING, then pin it before leaving the RCU section. The RCU
section protects the rejection path against initialization failure, while
the module reference protects the allocation path against normal unload.
sk_prot_alloc() obtains the socket-lifetime reference. Drop the temporary
reference after sk_alloc(). In inet6_create(), use the cached proto rather
than dereferencing the protosw after leaving RCU.
Fixes: a79af59efd20 ("[NET]: Fix module reference counts for loadable protocol modules")
Cc: stable@vger.kernel.org
Signed-off-by: Chengfeng Ye <nicoyip.dev@gmail.com>
---
Changes in v3:
- Reject protocol owners in MODULE_STATE_COMING before taking a temporary
module reference.
- Cache the owner under RCU and use that pointer for module_put().
- Use the cached proto for the IPv6 backlog callback instead of
dereferencing the protosw after RCU unlock.
Changes in v2:
- Pin answer_prot->owner in inet_create()/inet6_create() under RCU instead
of reordering try_module_get() in sk_prot_alloc().
Link: https://lore.kernel.org/netdev/20260825172349.232794-1-nicoyip.dev@gmail.com/ [v2]
Link: https://lore.kernel.org/netdev/20260823171311.3857087-1-nicoyip.dev@gmail.com/ [v1]
---
net/ipv4/af_inet.c | 16 +++++++++++++---
net/ipv6/af_inet6.c | 18 ++++++++++++++----
2 files changed, 27 insertions(+), 7 deletions(-)
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index 32d006c1a8ee..6b87f0b6cc22 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -263,6 +263,7 @@ static int inet_create(struct net *net, struct socket *sock, int protocol,
struct inet_protosw *answer;
struct inet_sock *inet;
struct proto *answer_prot;
+ struct module *answer_owner;
unsigned char answer_flags;
int try_loading_module = 0;
int err;
@@ -322,9 +323,16 @@ static int inet_create(struct net *net, struct socket *sock, int protocol,
!ns_capable(net->user_ns, CAP_NET_RAW))
goto out_rcu_unlock;
- sock->ops = answer->ops;
answer_prot = answer->prot;
+ answer_owner = answer_prot->owner;
answer_flags = answer->flags;
+ if (answer_owner &&
+ (module_is_coming(answer_owner) ||
+ !try_module_get(answer_owner))) {
+ err = -EPROTONOSUPPORT;
+ goto out_rcu_unlock;
+ }
+ sock->ops = answer->ops;
rcu_read_unlock();
WARN_ON(!answer_prot->slab);
@@ -332,7 +340,7 @@ static int inet_create(struct net *net, struct socket *sock, int protocol,
err = -ENOMEM;
sk = sk_alloc(net, PF_INET, GFP_KERNEL, answer_prot, kern);
if (!sk)
- goto out;
+ goto out_module_put;
err = 0;
if (INET_PROTOSW_REUSE & answer_flags)
@@ -398,6 +406,8 @@ static int inet_create(struct net *net, struct socket *sock, int protocol,
if (err)
goto out_sk_release;
}
+out_module_put:
+ module_put(answer_owner);
out:
return err;
out_rcu_unlock:
@@ -406,7 +416,7 @@ static int inet_create(struct net *net, struct socket *sock, int protocol,
out_sk_release:
sk_common_release(sk);
sock->sk = NULL;
- goto out;
+ goto out_module_put;
}
diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
index 282912a11999..1df8b2e7e243 100644
--- a/net/ipv6/af_inet6.c
+++ b/net/ipv6/af_inet6.c
@@ -110,6 +110,7 @@ static int inet6_create(struct net *net, struct socket *sock, int protocol,
struct sock *sk;
struct inet_protosw *answer;
struct proto *answer_prot;
+ struct module *answer_owner;
unsigned char answer_flags;
int try_loading_module = 0;
int err;
@@ -167,9 +168,16 @@ static int inet6_create(struct net *net, struct socket *sock, int protocol,
!ns_capable(net->user_ns, CAP_NET_RAW))
goto out_rcu_unlock;
- sock->ops = answer->ops;
answer_prot = answer->prot;
+ answer_owner = answer_prot->owner;
answer_flags = answer->flags;
+ if (answer_owner &&
+ (module_is_coming(answer_owner) ||
+ !try_module_get(answer_owner))) {
+ err = -EPROTONOSUPPORT;
+ goto out_rcu_unlock;
+ }
+ sock->ops = answer->ops;
rcu_read_unlock();
WARN_ON(!answer_prot->slab);
@@ -177,7 +185,7 @@ static int inet6_create(struct net *net, struct socket *sock, int protocol,
err = -ENOBUFS;
sk = sk_alloc(net, PF_INET6, GFP_KERNEL, answer_prot, kern);
if (!sk)
- goto out;
+ goto out_module_put;
sock_init_data(sock, sk);
@@ -201,7 +209,7 @@ static int inet6_create(struct net *net, struct socket *sock, int protocol,
sk->sk_family = PF_INET6;
sk->sk_protocol = protocol;
- sk->sk_backlog_rcv = answer->prot->backlog_rcv;
+ sk->sk_backlog_rcv = answer_prot->backlog_rcv;
inet_sk(sk)->pinet6 = np = inet6_sk_generic(sk);
np->hop_limit = -1;
@@ -251,6 +259,8 @@ static int inet6_create(struct net *net, struct socket *sock, int protocol,
if (err)
goto out_sk_release;
}
+out_module_put:
+ module_put(answer_owner);
out:
return err;
out_rcu_unlock:
@@ -259,7 +269,7 @@ static int inet6_create(struct net *net, struct socket *sock, int protocol,
out_sk_release:
sk_common_release(sk);
sock->sk = NULL;
- goto out;
+ goto out_module_put;
}
int __inet6_bind(struct sock *sk, struct sockaddr_unsized *uaddr, int addr_len,
--
2.43.0
next reply other threads:[~2026-09-04 11:15 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-09-04 11:15 Chengfeng Ye [this message]
2026-09-08 14:16 ` netdev-bot+sashiko
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=20260904111514.584264-1-nicoyip.dev@gmail.com \
--to=nicoyip.dev@gmail.com \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=horms@kernel.org \
--cc=kuba@kernel.org \
--cc=kuniyu@google.com \
--cc=linux-kernel@vger.kernel.org \
--cc=netdev@vger.kernel.org \
--cc=pabeni@redhat.com \
--cc=stable@vger.kernel.org \
--cc=willemb@google.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
all inboxes | Powered by JetHome®