On Mon, Jan 12, 2026 at 04:52:31PM -0800, Bobby Eshleman wrote:

On Mon, Jan 12, 2026 at 03:34:31PM -0800, Bobby Eshleman wrote:

quoted

On Sun, Jan 11, 2026 at 01:43:37AM -0500, Michael S. Tsirkin wrote:

quoted

On Tue, Dec 23, 2025 at 04:28:36PM -0800, Bobby Eshleman wrote:

quoted

From: Bobby Eshleman <redacted>

Add netns logic to vsock core. Additionally, modify transport hook
prototypes to be used by later transport-specific patches (e.g.,
*_seqpacket_allow()).

Namespaces are supported primarily by changing socket lookup functions
(e.g., vsock_find_connected_socket()) to take into account the socket
namespace and the namespace mode before considering a candidate socket a
"match".

This patch also introduces the sysctl /proc/sys/net/vsock/ns_mode to
report the mode and /proc/sys/net/vsock/child_ns_mode to set the mode
for new namespaces.

Add netns functionality (initialization, passing to transports, procfs,
etc...) to the af_vsock socket layer. Later patches that add netns
support to transports depend on this patch.

dgram_allow(), stream_allow(), and seqpacket_allow() callbacks are
modified to take a vsk in order to perform logic on namespace modes. In
future patches, the net will also be used for socket
lookups in these functions.

Signed-off-by: Bobby Eshleman <redacted>

...

quoted

 static int __vsock_bind_connectible(struct vsock_sock *vsk,
 				    struct sockaddr_vm *addr)
 {
+	struct net *net = sock_net(sk_vsock(vsk));
 	static u32 port;
 	struct sockaddr_vm new_addr;

Hmm this static port gives me pause. So some port number info leaks
between namespaces. I am not saying it's a big security issue
and yet ... people expect isolation.

Probably the easiest solution is making it per-ns, my quick rough draft
looks like this:

diff --git a/include/net/netns/vsock.h b/include/net/netns/vsock.h
index e2325e2d6ec5..b34d69a22fa8 100644
--- a/include/net/netns/vsock.h
+++ b/include/net/netns/vsock.h

@@ -11,6 +11,10 @@ enum vsock_net_mode {
 
 struct netns_vsock {
 	struct ctl_table_header *sysctl_hdr;
+
+	/* protected by the vsock_table_lock in af_vsock.c */
+	u32 port;
+
 	enum vsock_net_mode mode;
 	enum vsock_net_mode child_ns_mode;
 };

diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c
index 9d614e4a4fa5..cd2a47140134 100644
--- a/net/vmw_vsock/af_vsock.c
+++ b/net/vmw_vsock/af_vsock.c

@@ -748,11 +748,10 @@ static int __vsock_bind_connectible(struct vsock_sock *vsk,
 				    struct sockaddr_vm *addr)
 {
 	struct net *net = sock_net(sk_vsock(vsk));
-	static u32 port;
 	struct sockaddr_vm new_addr;
 
-	if (!port)
-		port = get_random_u32_above(LAST_RESERVED_PORT);
+	if (!net->vsock.port)
+		net->vsock.port = get_random_u32_above(LAST_RESERVED_PORT);
 
 	vsock_addr_init(&new_addr, addr->svm_cid, addr->svm_port);
 

@@ -761,11 +760,11 @@ static int __vsock_bind_connectible(struct vsock_sock *vsk,
 		unsigned int i;
 
 		for (i = 0; i < MAX_PORT_RETRIES; i++) {
-			if (port == VMADDR_PORT_ANY ||
-			    port <= LAST_RESERVED_PORT)
-				port = LAST_RESERVED_PORT + 1;
+			if (net->vsock.port == VMADDR_PORT_ANY ||
+			    net->vsock.port <= LAST_RESERVED_PORT)
+				net->vsock.port = LAST_RESERVED_PORT + 1;
 
-			new_addr.svm_port = port++;
+			new_addr.svm_port = net->vsock.port++;
 
 			if (!__vsock_find_bound_socket_net(&new_addr, net)) {
 				found = true;

Another option being to follow inet's path and use
siphash_4u32() the way that secure_ipv4_port_ephemeral() does...

Also acceptable.