vsock: support sockmap

This patch adds sockmap support for vsock sockets. It is intended to be
usable by all transports, but only the virtio and loopback transports
are implemented.

SOCK_STREAM, SOCK_DGRAM, and SOCK_SEQPACKET are all supported.

Signed-off-by: Bobby Eshleman <[e-mail]>
Acked-by: Michael S. Tsirkin <[e-mail]>
Reviewed-by: Stefano Garzarella <[e-mail]>
Signed-off-by: David S. Miller <[e-mail]>

drivers/vhost/vsock.c

···
 		.notify_send_post_enqueue = virtio_transport_notify_send_post_enqueue,
 		.notify_buffer_size       = virtio_transport_notify_buffer_size,
 
+		.read_skb = virtio_transport_read_skb,
 	},
 
 	.send_pkt = vhost_transport_send_pkt,

include/linux/virtio_vsock.h

···
 void virtio_transport_put_credit(struct virtio_vsock_sock *vvs, u32 credit);
 void virtio_transport_deliver_tap_pkt(struct sk_buff *skb);
 int virtio_transport_purge_skbs(void *vsk, struct sk_buff_head *list);
+int virtio_transport_read_skb(struct vsock_sock *vsk, skb_read_actor_t read_actor);
 #endif /* _LINUX_VIRTIO_VSOCK_H */

include/net/af_vsock.h

···
 	void *trans;
 };
 
+s64 vsock_connectible_has_data(struct vsock_sock *vsk);
 s64 vsock_stream_has_data(struct vsock_sock *vsk);
 s64 vsock_stream_has_space(struct vsock_sock *vsk);
 struct sock *vsock_create_connected(struct sock *parent);
···
 
 	/* Addressing. */
 	u32 (*get_local_cid)(void);
+
+	/* Read a single skb */
+	int (*read_skb)(struct vsock_sock *, skb_read_actor_t);
 };
 
 /**** CORE ****/
···
 int vsock_add_tap(struct vsock_tap *vt);
 int vsock_remove_tap(struct vsock_tap *vt);
 void vsock_deliver_tap(struct sk_buff *build_skb(void *opaque), void *opaque);
+int vsock_connectible_recvmsg(struct socket *sock, struct msghdr *msg, size_t len,
+			      int flags);
+int vsock_dgram_recvmsg(struct socket *sock, struct msghdr *msg,
+			size_t len, int flags);
+
+#ifdef CONFIG_BPF_SYSCALL
+extern struct proto vsock_proto;
+int vsock_bpf_update_proto(struct sock *sk, struct sk_psock *psock, bool restore);
+void __init vsock_bpf_build_proto(void);
+#else
+static inline void __init vsock_bpf_build_proto(void)
+{}
+#endif
 
 #endif /* __AF_VSOCK_H__ */

net/vmw_vsock/Makefile

···
 obj-$(CONFIG_VSOCKETS_LOOPBACK) += vsock_loopback.o
 
 vsock-y += af_vsock.o af_vsock_tap.o vsock_addr.o
+vsock-$(CONFIG_BPF_SYSCALL) += vsock_bpf.o
 
 vsock_diag-y += diag.o
 

net/vmw_vsock/af_vsock.c

···
 static int vsock_queue_rcv_skb(struct sock *sk, struct sk_buff *skb);
 
 /* Protocol family. */
-static struct proto vsock_proto = {
+struct proto vsock_proto = {
 	.name = "AF_VSOCK",
 	.owner = THIS_MODULE,
 	.obj_size = sizeof(struct vsock_sock),
+#ifdef CONFIG_BPF_SYSCALL
+	.psock_update_sk_prot = vsock_bpf_update_proto,
+#endif
 };
 
 /* The default peer timeout indicates how long we will wait for a peer response
···
 }
 EXPORT_SYMBOL_GPL(vsock_stream_has_data);
 
-static s64 vsock_connectible_has_data(struct vsock_sock *vsk)
+s64 vsock_connectible_has_data(struct vsock_sock *vsk)
 {
 	struct sock *sk = sk_vsock(vsk);
 
···
 	else
 		return vsock_stream_has_data(vsk);
 }
+EXPORT_SYMBOL_GPL(vsock_connectible_has_data);
 
 s64 vsock_stream_has_space(struct vsock_sock *vsk)
 {
···
 	return mask;
 }
 
+static int vsock_read_skb(struct sock *sk, skb_read_actor_t read_actor)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	return vsk->transport->read_skb(vsk, read_actor);
+}
+
 static int vsock_dgram_sendmsg(struct socket *sock, struct msghdr *msg,
 			       size_t len)
 {
···
 	memcpy(&vsk->remote_addr, remote_addr, sizeof(vsk->remote_addr));
 	sock->state = SS_CONNECTED;
 
+	/* sock map disallows redirection of non-TCP sockets with sk_state !=
+	 * TCP_ESTABLISHED (see sock_map_redirect_allowed()), so we set
+	 * TCP_ESTABLISHED here to allow redirection of connected vsock dgrams.
+	 *
+	 * This doesn't seem to be abnormal state for datagram sockets, as the
+	 * same approach can be see in other datagram socket types as well
+	 * (such as unix sockets).
+	 */
+	sk->sk_state = TCP_ESTABLISHED;
+
 out:
 	release_sock(sk);
 	return err;
 }
 
-static int vsock_dgram_recvmsg(struct socket *sock, struct msghdr *msg,
-			       size_t len, int flags)
+int vsock_dgram_recvmsg(struct socket *sock, struct msghdr *msg,
+			size_t len, int flags)
 {
-	struct vsock_sock *vsk = vsock_sk(sock->sk);
+#ifdef CONFIG_BPF_SYSCALL
+	const struct proto *prot;
+#endif
+	struct vsock_sock *vsk;
+	struct sock *sk;
+
+	sk = sock->sk;
+	vsk = vsock_sk(sk);
+
+#ifdef CONFIG_BPF_SYSCALL
+	prot = READ_ONCE(sk->sk_prot);
+	if (prot != &vsock_proto)
+		return prot->recvmsg(sk, msg, len, flags, NULL);
+#endif
 
 	return vsk->transport->dgram_dequeue(vsk, msg, len, flags);
 }
+EXPORT_SYMBOL_GPL(vsock_dgram_recvmsg);
 
 static const struct proto_ops vsock_dgram_ops = {
 	.family = PF_VSOCK,
···
 	.recvmsg = vsock_dgram_recvmsg,
 	.mmap = sock_no_mmap,
 	.sendpage = sock_no_sendpage,
+	.read_skb = vsock_read_skb,
 };
 
 static int vsock_transport_cancel_pkt(struct vsock_sock *vsk)
···
 	return err;
 }
 
-static int
+int
 vsock_connectible_recvmsg(struct socket *sock, struct msghdr *msg, size_t len,
 			  int flags)
 {
 	struct sock *sk;
 	struct vsock_sock *vsk;
 	const struct vsock_transport *transport;
+#ifdef CONFIG_BPF_SYSCALL
+	const struct proto *prot;
+#endif
 	int err;
 
 	sk = sock->sk;
···
 		goto out;
 	}
 
+#ifdef CONFIG_BPF_SYSCALL
+	prot = READ_ONCE(sk->sk_prot);
+	if (prot != &vsock_proto) {
+		release_sock(sk);
+		return prot->recvmsg(sk, msg, len, flags, NULL);
+	}
+#endif
+
 	if (sk->sk_type == SOCK_STREAM)
 		err = __vsock_stream_recvmsg(sk, msg, len, flags);
 	else
···
 	release_sock(sk);
 	return err;
 }
+EXPORT_SYMBOL_GPL(vsock_connectible_recvmsg);
 
 static int vsock_set_rcvlowat(struct sock *sk, int val)
 {
···
 	.mmap = sock_no_mmap,
 	.sendpage = sock_no_sendpage,
 	.set_rcvlowat = vsock_set_rcvlowat,
+	.read_skb = vsock_read_skb,
 };
 
 static const struct proto_ops vsock_seqpacket_ops = {
···
 	.recvmsg = vsock_connectible_recvmsg,
 	.mmap = sock_no_mmap,
 	.sendpage = sock_no_sendpage,
+	.read_skb = vsock_read_skb,
 };
 
 static int vsock_create(struct net *net, struct socket *sock,
···
 		       AF_VSOCK, err);
 		goto err_unregister_proto;
 	}
+
+	vsock_bpf_build_proto();
 
 	return 0;
 

net/vmw_vsock/virtio_transport.c

···
 		.notify_send_pre_enqueue  = virtio_transport_notify_send_pre_enqueue,
 		.notify_send_post_enqueue = virtio_transport_notify_send_post_enqueue,
 		.notify_buffer_size       = virtio_transport_notify_buffer_size,
+
+		.read_skb = virtio_transport_read_skb,
 	},
 
 	.send_pkt = virtio_transport_send_pkt,

net/vmw_vsock/virtio_transport_common.c

···
 }
 EXPORT_SYMBOL_GPL(virtio_transport_purge_skbs);
 
+int virtio_transport_read_skb(struct vsock_sock *vsk, skb_read_actor_t recv_actor)
+{
+	struct virtio_vsock_sock *vvs = vsk->trans;
+	struct sock *sk = sk_vsock(vsk);
+	struct sk_buff *skb;
+	int off = 0;
+	int copied;
+	int err;
+
+	spin_lock_bh(&vvs->rx_lock);
+	/* Use __skb_recv_datagram() for race-free handling of the receive. It
+	 * works for types other than dgrams.
+	 */
+	skb = __skb_recv_datagram(sk, &vvs->rx_queue, MSG_DONTWAIT, &off, &err);
+	spin_unlock_bh(&vvs->rx_lock);
+
+	if (!skb)
+		return err;
+
+	copied = recv_actor(sk, skb);
+	kfree_skb(skb);
+	return copied;
+}
+EXPORT_SYMBOL_GPL(virtio_transport_read_skb);
+
 MODULE_LICENSE("GPL v2");
 MODULE_AUTHOR("Asias He");
 MODULE_DESCRIPTION("common code for virtio vsock");

net/vmw_vsock/vsock_bpf.c

···
+// SPDX-License-Identifier: GPL-2.0
+/* Copyright (c) 2022 Bobby Eshleman <[e-mail]>
+ *
+ * Based off of net/unix/unix_bpf.c
+ */
+
+#include <linux/bpf.h>
+#include <linux/module.h>
+#include <linux/skmsg.h>
+#include <linux/socket.h>
+#include <linux/wait.h>
+#include <net/af_vsock.h>
+#include <net/sock.h>
+
+#define vsock_sk_has_data(__sk, __psock)				\
+		({	!skb_queue_empty(&(__sk)->sk_receive_queue) ||	\
+			!skb_queue_empty(&(__psock)->ingress_skb) ||	\
+			!list_empty(&(__psock)->ingress_msg);		\
+		})
+
+static struct proto *vsock_prot_saved __read_mostly;
+static DEFINE_SPINLOCK(vsock_prot_lock);
+static struct proto vsock_bpf_prot;
+
+static bool vsock_has_data(struct sock *sk, struct sk_psock *psock)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+	s64 ret;
+
+	ret = vsock_connectible_has_data(vsk);
+	if (ret > 0)
+		return true;
+
+	return vsock_sk_has_data(sk, psock);
+}
+
+static bool vsock_msg_wait_data(struct sock *sk, struct sk_psock *psock, long timeo)
+{
+	bool ret;
+
+	DEFINE_WAIT_FUNC(wait, woken_wake_function);
+
+	if (sk->sk_shutdown & RCV_SHUTDOWN)
+		return true;
+
+	if (!timeo)
+		return false;
+
+	add_wait_queue(sk_sleep(sk), &wait);
+	sk_set_bit(SOCKWQ_ASYNC_WAITDATA, sk);
+	ret = vsock_has_data(sk, psock);
+	if (!ret) {
+		wait_woken(&wait, TASK_INTERRUPTIBLE, timeo);
+		ret = vsock_has_data(sk, psock);
+	}
+	sk_clear_bit(SOCKWQ_ASYNC_WAITDATA, sk);
+	remove_wait_queue(sk_sleep(sk), &wait);
+	return ret;
+}
+
+static int __vsock_recvmsg(struct sock *sk, struct msghdr *msg, size_t len, int flags)
+{
+	struct socket *sock = sk->sk_socket;
+	int err;
+
+	if (sk->sk_type == SOCK_STREAM || sk->sk_type == SOCK_SEQPACKET)
+		err = vsock_connectible_recvmsg(sock, msg, len, flags);
+	else if (sk->sk_type == SOCK_DGRAM)
+		err = vsock_dgram_recvmsg(sock, msg, len, flags);
+	else
+		err = -EPROTOTYPE;
+
+	return err;
+}
+
+static int vsock_bpf_recvmsg(struct sock *sk, struct msghdr *msg,
+			     size_t len, int flags, int *addr_len)
+{
+	struct sk_psock *psock;
+	int copied;
+
+	psock = sk_psock_get(sk);
+	if (unlikely(!psock))
+		return __vsock_recvmsg(sk, msg, len, flags);
+
+	lock_sock(sk);
+	if (vsock_has_data(sk, psock) && sk_psock_queue_empty(psock)) {
+		release_sock(sk);
+		sk_psock_put(sk, psock);
+		return __vsock_recvmsg(sk, msg, len, flags);
+	}
+
+	copied = sk_msg_recvmsg(sk, psock, msg, len, flags);
+	while (copied == 0) {
+		long timeo = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
+
+		if (!vsock_msg_wait_data(sk, psock, timeo)) {
+			copied = -EAGAIN;
+			break;
+		}
+
+		if (sk_psock_queue_empty(psock)) {
+			release_sock(sk);
+			sk_psock_put(sk, psock);
+			return __vsock_recvmsg(sk, msg, len, flags);
+		}
+
+		copied = sk_msg_recvmsg(sk, psock, msg, len, flags);
+	}
+
+	release_sock(sk);
+	sk_psock_put(sk, psock);
+
+	return copied;
+}
+
+/* Copy of original proto with updated sock_map methods */
+static struct proto vsock_bpf_prot = {
+	.close = sock_map_close,
+	.recvmsg = vsock_bpf_recvmsg,
+	.sock_is_readable = sk_msg_is_readable,
+	.unhash = sock_map_unhash,
+};
+
+static void vsock_bpf_rebuild_protos(struct proto *prot, const struct proto *base)
+{
+	*prot        = *base;
+	prot->close  = sock_map_close;
+	prot->recvmsg = vsock_bpf_recvmsg;
+	prot->sock_is_readable = sk_msg_is_readable;
+}
+
+static void vsock_bpf_check_needs_rebuild(struct proto *ops)
+{
+	/* Paired with the smp_store_release() below. */
+	if (unlikely(ops != smp_load_acquire(&vsock_prot_saved))) {
+		spin_lock_bh(&vsock_prot_lock);
+		if (likely(ops != vsock_prot_saved)) {
+			vsock_bpf_rebuild_protos(&vsock_bpf_prot, ops);
+			/* Make sure proto function pointers are updated before publishing the
+			 * pointer to the struct.
+			 */
+			smp_store_release(&vsock_prot_saved, ops);
+		}
+		spin_unlock_bh(&vsock_prot_lock);
+	}
+}
+
+int vsock_bpf_update_proto(struct sock *sk, struct sk_psock *psock, bool restore)
+{
+	struct vsock_sock *vsk;
+
+	if (restore) {
+		sk->sk_write_space = psock->saved_write_space;
+		sock_replace_proto(sk, psock->sk_proto);
+		return 0;
+	}
+
+	vsk = vsock_sk(sk);
+	if (!vsk->transport)
+		return -ENODEV;
+
+	if (!vsk->transport->read_skb)
+		return -EOPNOTSUPP;
+
+	vsock_bpf_check_needs_rebuild(psock->sk_proto);
+	sock_replace_proto(sk, &vsock_bpf_prot);
+	return 0;
+}
+
+void __init vsock_bpf_build_proto(void)
+{
+	vsock_bpf_rebuild_protos(&vsock_bpf_prot, &vsock_proto);
+}

net/vmw_vsock/vsock_loopback.c

···
 		.notify_send_pre_enqueue  = virtio_transport_notify_send_pre_enqueue,
 		.notify_send_post_enqueue = virtio_transport_notify_send_post_enqueue,
 		.notify_buffer_size       = virtio_transport_notify_buffer_size,
+
+		.read_skb = virtio_transport_read_skb,
 	},
 
 	.send_pkt = vsock_loopback_send_pkt,