[PATCH v5] nfsd: validate sockaddr length per family in listener_set

[PATCH v5] nfsd: validate sockaddr length per family in listener_set

[Jeff Layton]

nfsd_sock_nl_policy declares NFSD_A_SOCK_ADDR as a bare NLA_BINARY
attribute with no minimum length. A CAP_NET_ADMIN caller can send a
16-byte NFSD_A_SOCK_ADDR with sa_family=AF_INET6, causing a 12-byte
OOB read across three consumers (rpc_cmp_addr_port, svc_find_listener,
kernel_bind).

nfsd_nl_listener_set_doit() also parsed and validated each listener
entry inline in two separate loops, interleaved with mutating the
running listener configuration. The validation was duplicated, used an
open-coded "nla_len < sizeof(struct sockaddr)" check that was too short
for AF_INET6, and handled a malformed entry inconsistently depending on
which loop noticed it.

Add an nfsd_nl_validate_listeners() helper that walks the entire list
once and confirms each entry parses, carries both an address and a
transport name, and is long enough for its address family
(sizeof(struct sockaddr_in) for AF_INET, sizeof(struct sockaddr_in6)
for AF_INET6, -EAFNOSUPPORT otherwise). Call it before taking
nfsd_mutex or creating the serv, so a malformed request fails cleanly
with no side effects.

Since every entry is known valid by the time the two existing loops
run, drop the redundant presence and per-family length checks from
both, leaving only the nla_parse_nested() call needed to extract the
data.

Fixes: 16a471177496 ("NFSD: add listener-{set,get} netlink command")
Assisted-by: Claude:claude-opus-4-8
Signed-off-by: Jeff Layton <jlayton@kernel.org>
---
Changes in v5:
- Vet the entire listener list before making any changes
- Link to v4: https://lore.kernel.org/r/20260615-nfsd-testing-v4-1-517fce448c85@kernel.org

Changes in v4:
- Drop policy floor and do full inline validation
- Link to v3: https://lore.kernel.org/r/20260615-nfsd-testing-v3-1-e9b515e17e54@kernel.org
---
 fs/nfsd/nfsctl.c | 77 +++++++++++++++++++++++++++++++++++++++++++++++---------
 1 file changed, 65 insertions(+), 12 deletions(-)

diff --git a/fs/nfsd/nfsctl.c b/fs/nfsd/nfsctl.c
index f1ecbb13f642..b34adebec43a 100644
--- a/fs/nfsd/nfsctl.c
+++ b/fs/nfsd/nfsctl.c
@@ -1971,6 +1971,60 @@ int nfsd_nl_version_get_doit(struct sk_buff *skb, struct genl_info *info)
 	return err;
 }
 
+/**
+ * nfsd_nl_validate_listeners - sanity-check the listener list from userland
+ * @info: netlink metadata and command arguments
+ *
+ * Walk every NFSD_A_SERVER_SOCK_ADDR attribute and confirm that each entry
+ * is well-formed: it parses against the policy, carries both an address and
+ * a transport name, and the address is long enough for its family. Doing
+ * this up front lets the callers below assume every entry is valid and
+ * guarantees we make no changes when the request is malformed.
+ *
+ * Return 0 if every entry is valid, or a negative errno otherwise.
+ */
+static int nfsd_nl_validate_listeners(struct genl_info *info)
+{
+	const struct nlattr *attr;
+	int rem;
+
+	nlmsg_for_each_attr_type(attr, NFSD_A_SERVER_SOCK_ADDR, info->nlhdr,
+				 GENL_HDRLEN, rem) {
+		struct nlattr *tb[NFSD_A_SOCK_MAX + 1];
+		struct sockaddr *sa;
+		int err;
+
+		err = nla_parse_nested(tb, NFSD_A_SOCK_MAX, attr,
+				       nfsd_sock_nl_policy, info->extack);
+		if (err < 0)
+			return err;
+
+		if (!tb[NFSD_A_SOCK_ADDR] || !tb[NFSD_A_SOCK_TRANSPORT_NAME])
+			return -EINVAL;
+
+		sa = nla_data(tb[NFSD_A_SOCK_ADDR]);
+		if (nla_len(tb[NFSD_A_SOCK_ADDR]) < sizeof(sa->sa_family))
+			return -EINVAL;
+
+		switch (sa->sa_family) {
+		case AF_INET:
+			if (nla_len(tb[NFSD_A_SOCK_ADDR]) <
+			    sizeof(struct sockaddr_in))
+				return -EINVAL;
+			break;
+		case AF_INET6:
+			if (nla_len(tb[NFSD_A_SOCK_ADDR]) <
+			    sizeof(struct sockaddr_in6))
+				return -EINVAL;
+			break;
+		default:
+			return -EAFNOSUPPORT;
+		}
+	}
+
+	return 0;
+}
+
 /**
  * nfsd_nl_listener_set_doit - set the nfs running sockets
  * @skb: reply buffer
@@ -1989,6 +2043,15 @@ int nfsd_nl_listener_set_doit(struct sk_buff *skb, struct genl_info *info)
 	bool delete = false;
 	int err, rem;
 
+	/*
+	 * Validate the entire listener list before making any changes, so a
+	 * malformed request fails cleanly without creating a serv or touching
+	 * the existing listeners.
+	 */
+	err = nfsd_nl_validate_listeners(info);
+	if (err)
+		return err;
+
 	mutex_lock(&nfsd_mutex);
 
 	err = nfsd_create_serv(net);
@@ -2015,16 +2078,11 @@ int nfsd_nl_listener_set_doit(struct sk_buff *skb, struct genl_info *info)
 		const char *xcl_name;
 		struct sockaddr *sa;
 
+		/* validated up front in nfsd_nl_validate_listeners() */
 		if (nla_parse_nested(tb, NFSD_A_SOCK_MAX, attr,
 				     nfsd_sock_nl_policy, info->extack) < 0)
 			continue;
 
-		if (!tb[NFSD_A_SOCK_ADDR] || !tb[NFSD_A_SOCK_TRANSPORT_NAME])
-			continue;
-
-		if (nla_len(tb[NFSD_A_SOCK_ADDR]) < sizeof(*sa))
-			continue;
-
 		xcl_name = nla_data(tb[NFSD_A_SOCK_TRANSPORT_NAME]);
 		sa = nla_data(tb[NFSD_A_SOCK_ADDR]);
 
@@ -2076,16 +2134,11 @@ int nfsd_nl_listener_set_doit(struct sk_buff *skb, struct genl_info *info)
 		struct sockaddr *sa;
 		int ret;
 
+		/* validated up front in nfsd_nl_validate_listeners() */
 		if (nla_parse_nested(tb, NFSD_A_SOCK_MAX, attr,
 				     nfsd_sock_nl_policy, info->extack) < 0)
 			continue;
 
-		if (!tb[NFSD_A_SOCK_ADDR] || !tb[NFSD_A_SOCK_TRANSPORT_NAME])
-			continue;
-
-		if (nla_len(tb[NFSD_A_SOCK_ADDR]) < sizeof(*sa))
-			continue;
-
 		xcl_name = nla_data(tb[NFSD_A_SOCK_TRANSPORT_NAME]);
 		sa = nla_data(tb[NFSD_A_SOCK_ADDR]);
 

---
base-commit: 332e2f4f37b213f231be1ab5ddc17e2052383b60
change-id: 20260608-nfsd-testing-688a82433c50

Best regards,
-- 
Jeff Layton <jlayton@kernel.org>

Re: [PATCH v5] nfsd: validate sockaddr length per family in listener_set

[Chuck Lever]

On Mon, 15 Jun 2026 14:31:25 -0400, Jeff Layton wrote:
> nfsd_sock_nl_policy declares NFSD_A_SOCK_ADDR as a bare NLA_BINARY
> attribute with no minimum length. A CAP_NET_ADMIN caller can send a
> 16-byte NFSD_A_SOCK_ADDR with sa_family=AF_INET6, causing a 12-byte
> OOB read across three consumers (rpc_cmp_addr_port, svc_find_listener,
> kernel_bind).
> 
> nfsd_nl_listener_set_doit() also parsed and validated each listener
> entry inline in two separate loops, interleaved with mutating the
> running listener configuration. The validation was duplicated, used an
> open-coded "nla_len < sizeof(struct sockaddr)" check that was too short
> for AF_INET6, and handled a malformed entry inconsistently depending on
> which loop noticed it.
> 
> [...]

Applied to nfsd-testing, thanks!

[1/1] nfsd: validate sockaddr length per family in listener_set
      commit: 1f102d9ff00620b845546051ad1ee57976f2db88

--
Chuck Lever