1 // SPDX-License-Identifier: GPL-2.0-only
3 * VMware vSockets Driver
5 * Copyright (C) 2007-2013 VMware, Inc. All rights reserved.
8 /* Implementation notes:
10 * - There are two kinds of sockets: those created by user action (such as
11 * calling socket(2)) and those created by incoming connection request packets.
13 * - There are two "global" tables, one for bound sockets (sockets that have
14 * specified an address that they are responsible for) and one for connected
15 * sockets (sockets that have established a connection with another socket).
16 * These tables are "global" in that all sockets on the system are placed
17 * within them. - Note, though, that the bound table contains an extra entry
18 * for a list of unbound sockets and SOCK_DGRAM sockets will always remain in
19 * that list. The bound table is used solely for lookup of sockets when packets
20 * are received and that's not necessary for SOCK_DGRAM sockets since we create
21 * a datagram handle for each and need not perform a lookup. Keeping SOCK_DGRAM
22 * sockets out of the bound hash buckets will reduce the chance of collisions
23 * when looking for SOCK_STREAM sockets and prevents us from having to check the
24 * socket type in the hash table lookups.
26 * - Sockets created by user action will either be "client" sockets that
27 * initiate a connection or "server" sockets that listen for connections; we do
28 * not support simultaneous connects (two "client" sockets connecting).
30 * - "Server" sockets are referred to as listener sockets throughout this
31 * implementation because they are in the TCP_LISTEN state. When a
32 * connection request is received (the second kind of socket mentioned above),
33 * we create a new socket and refer to it as a pending socket. These pending
34 * sockets are placed on the pending connection list of the listener socket.
35 * When future packets are received for the address the listener socket is
36 * bound to, we check if the source of the packet is from one that has an
37 * existing pending connection. If it does, we process the packet for the
38 * pending socket. When that socket reaches the connected state, it is removed
39 * from the listener socket's pending list and enqueued in the listener
40 * socket's accept queue. Callers of accept(2) will accept connected sockets
41 * from the listener socket's accept queue. If the socket cannot be accepted
42 * for some reason then it is marked rejected. Once the connection is
43 * accepted, it is owned by the user process and the responsibility for cleanup
44 * falls with that user process.
46 * - It is possible that these pending sockets will never reach the connected
47 * state; in fact, we may never receive another packet after the connection
48 * request. Because of this, we must schedule a cleanup function to run in the
49 * future, after some amount of time passes where a connection should have been
50 * established. This function ensures that the socket is off all lists so it
51 * cannot be retrieved, then drops all references to the socket so it is cleaned
52 * up (sock_put() -> sk_free() -> our sk_destruct implementation). Note this
53 * function will also cleanup rejected sockets, those that reach the connected
54 * state but leave it before they have been accepted.
56 * - Lock ordering for pending or accept queue sockets is:
58 * lock_sock(listener);
59 * lock_sock_nested(pending, SINGLE_DEPTH_NESTING);
61 * Using explicit nested locking keeps lockdep happy since normally only one
62 * lock of a given class may be taken at a time.
64 * - Sockets created by user action will be cleaned up when the user process
65 * calls close(2), causing our release implementation to be called. Our release
66 * implementation will perform some cleanup then drop the last reference so our
67 * sk_destruct implementation is invoked. Our sk_destruct implementation will
68 * perform additional cleanup that's common for both types of sockets.
70 * - A socket's reference count is what ensures that the structure won't be
71 * freed. Each entry in a list (such as the "global" bound and connected tables
72 * and the listener socket's pending list and connected queue) ensures a
73 * reference. When we defer work until process context and pass a socket as our
74 * argument, we must ensure the reference count is increased to ensure the
75 * socket isn't freed before the function is run; the deferred function will
76 * then drop the reference.
78 * - sk->sk_state uses the TCP state constants because they are widely used by
79 * other address families and exposed to userspace tools like ss(8):
81 * TCP_CLOSE - unconnected
82 * TCP_SYN_SENT - connecting
83 * TCP_ESTABLISHED - connected
84 * TCP_CLOSING - disconnecting
85 * TCP_LISTEN - listening
88 #include <linux/types.h>
89 #include <linux/bitops.h>
90 #include <linux/cred.h>
91 #include <linux/init.h>
93 #include <linux/kernel.h>
94 #include <linux/sched/signal.h>
95 #include <linux/kmod.h>
96 #include <linux/list.h>
97 #include <linux/miscdevice.h>
98 #include <linux/module.h>
99 #include <linux/mutex.h>
100 #include <linux/net.h>
101 #include <linux/poll.h>
102 #include <linux/random.h>
103 #include <linux/skbuff.h>
104 #include <linux/smp.h>
105 #include <linux/socket.h>
106 #include <linux/stddef.h>
107 #include <linux/unistd.h>
108 #include <linux/wait.h>
109 #include <linux/workqueue.h>
110 #include <net/sock.h>
111 #include <net/af_vsock.h>
113 static int __vsock_bind(struct sock *sk, struct sockaddr_vm *addr);
114 static void vsock_sk_destruct(struct sock *sk);
115 static int vsock_queue_rcv_skb(struct sock *sk, struct sk_buff *skb);
117 /* Protocol family. */
118 static struct proto vsock_proto = {
120 .owner = THIS_MODULE,
121 .obj_size = sizeof(struct vsock_sock),
124 /* The default peer timeout indicates how long we will wait for a peer response
125 * to a control message.
127 #define VSOCK_DEFAULT_CONNECT_TIMEOUT (2 * HZ)
129 #define VSOCK_DEFAULT_BUFFER_SIZE (1024 * 256)
130 #define VSOCK_DEFAULT_BUFFER_MAX_SIZE (1024 * 256)
131 #define VSOCK_DEFAULT_BUFFER_MIN_SIZE 128
133 /* Transport used for host->guest communication */
134 static const struct vsock_transport *transport_h2g;
135 /* Transport used for guest->host communication */
136 static const struct vsock_transport *transport_g2h;
137 /* Transport used for DGRAM communication */
138 static const struct vsock_transport *transport_dgram;
139 static DEFINE_MUTEX(vsock_register_mutex);
143 /* Each bound VSocket is stored in the bind hash table and each connected
144 * VSocket is stored in the connected hash table.
146 * Unbound sockets are all put on the same list attached to the end of the hash
147 * table (vsock_unbound_sockets). Bound sockets are added to the hash table in
148 * the bucket that their local address hashes to (vsock_bound_sockets(addr)
149 * represents the list that addr hashes to).
151 * Specifically, we initialize the vsock_bind_table array to a size of
152 * VSOCK_HASH_SIZE + 1 so that vsock_bind_table[0] through
153 * vsock_bind_table[VSOCK_HASH_SIZE - 1] are for bound sockets and
154 * vsock_bind_table[VSOCK_HASH_SIZE] is for unbound sockets. The hash function
155 * mods with VSOCK_HASH_SIZE to ensure this.
157 #define MAX_PORT_RETRIES 24
159 #define VSOCK_HASH(addr) ((addr)->svm_port % VSOCK_HASH_SIZE)
160 #define vsock_bound_sockets(addr) (&vsock_bind_table[VSOCK_HASH(addr)])
161 #define vsock_unbound_sockets (&vsock_bind_table[VSOCK_HASH_SIZE])
163 /* XXX This can probably be implemented in a better way. */
164 #define VSOCK_CONN_HASH(src, dst) \
165 (((src)->svm_cid ^ (dst)->svm_port) % VSOCK_HASH_SIZE)
166 #define vsock_connected_sockets(src, dst) \
167 (&vsock_connected_table[VSOCK_CONN_HASH(src, dst)])
168 #define vsock_connected_sockets_vsk(vsk) \
169 vsock_connected_sockets(&(vsk)->remote_addr, &(vsk)->local_addr)
171 struct list_head vsock_bind_table[VSOCK_HASH_SIZE + 1];
172 EXPORT_SYMBOL_GPL(vsock_bind_table);
173 struct list_head vsock_connected_table[VSOCK_HASH_SIZE];
174 EXPORT_SYMBOL_GPL(vsock_connected_table);
175 DEFINE_SPINLOCK(vsock_table_lock);
176 EXPORT_SYMBOL_GPL(vsock_table_lock);
178 /* Autobind this socket to the local address if necessary. */
179 static int vsock_auto_bind(struct vsock_sock *vsk)
181 struct sock *sk = sk_vsock(vsk);
182 struct sockaddr_vm local_addr;
184 if (vsock_addr_bound(&vsk->local_addr))
186 vsock_addr_init(&local_addr, VMADDR_CID_ANY, VMADDR_PORT_ANY);
187 return __vsock_bind(sk, &local_addr);
190 static void vsock_init_tables(void)
194 for (i = 0; i < ARRAY_SIZE(vsock_bind_table); i++)
195 INIT_LIST_HEAD(&vsock_bind_table[i]);
197 for (i = 0; i < ARRAY_SIZE(vsock_connected_table); i++)
198 INIT_LIST_HEAD(&vsock_connected_table[i]);
201 static void __vsock_insert_bound(struct list_head *list,
202 struct vsock_sock *vsk)
205 list_add(&vsk->bound_table, list);
208 static void __vsock_insert_connected(struct list_head *list,
209 struct vsock_sock *vsk)
212 list_add(&vsk->connected_table, list);
215 static void __vsock_remove_bound(struct vsock_sock *vsk)
217 list_del_init(&vsk->bound_table);
221 static void __vsock_remove_connected(struct vsock_sock *vsk)
223 list_del_init(&vsk->connected_table);
227 static struct sock *__vsock_find_bound_socket(struct sockaddr_vm *addr)
229 struct vsock_sock *vsk;
231 list_for_each_entry(vsk, vsock_bound_sockets(addr), bound_table) {
232 if (vsock_addr_equals_addr(addr, &vsk->local_addr))
233 return sk_vsock(vsk);
235 if (addr->svm_port == vsk->local_addr.svm_port &&
236 (vsk->local_addr.svm_cid == VMADDR_CID_ANY ||
237 addr->svm_cid == VMADDR_CID_ANY))
238 return sk_vsock(vsk);
244 static struct sock *__vsock_find_connected_socket(struct sockaddr_vm *src,
245 struct sockaddr_vm *dst)
247 struct vsock_sock *vsk;
249 list_for_each_entry(vsk, vsock_connected_sockets(src, dst),
251 if (vsock_addr_equals_addr(src, &vsk->remote_addr) &&
252 dst->svm_port == vsk->local_addr.svm_port) {
253 return sk_vsock(vsk);
260 static void vsock_insert_unbound(struct vsock_sock *vsk)
262 spin_lock_bh(&vsock_table_lock);
263 __vsock_insert_bound(vsock_unbound_sockets, vsk);
264 spin_unlock_bh(&vsock_table_lock);
267 void vsock_insert_connected(struct vsock_sock *vsk)
269 struct list_head *list = vsock_connected_sockets(
270 &vsk->remote_addr, &vsk->local_addr);
272 spin_lock_bh(&vsock_table_lock);
273 __vsock_insert_connected(list, vsk);
274 spin_unlock_bh(&vsock_table_lock);
276 EXPORT_SYMBOL_GPL(vsock_insert_connected);
278 void vsock_remove_bound(struct vsock_sock *vsk)
280 spin_lock_bh(&vsock_table_lock);
281 if (__vsock_in_bound_table(vsk))
282 __vsock_remove_bound(vsk);
283 spin_unlock_bh(&vsock_table_lock);
285 EXPORT_SYMBOL_GPL(vsock_remove_bound);
287 void vsock_remove_connected(struct vsock_sock *vsk)
289 spin_lock_bh(&vsock_table_lock);
290 if (__vsock_in_connected_table(vsk))
291 __vsock_remove_connected(vsk);
292 spin_unlock_bh(&vsock_table_lock);
294 EXPORT_SYMBOL_GPL(vsock_remove_connected);
296 struct sock *vsock_find_bound_socket(struct sockaddr_vm *addr)
300 spin_lock_bh(&vsock_table_lock);
301 sk = __vsock_find_bound_socket(addr);
305 spin_unlock_bh(&vsock_table_lock);
309 EXPORT_SYMBOL_GPL(vsock_find_bound_socket);
311 struct sock *vsock_find_connected_socket(struct sockaddr_vm *src,
312 struct sockaddr_vm *dst)
316 spin_lock_bh(&vsock_table_lock);
317 sk = __vsock_find_connected_socket(src, dst);
321 spin_unlock_bh(&vsock_table_lock);
325 EXPORT_SYMBOL_GPL(vsock_find_connected_socket);
327 void vsock_remove_sock(struct vsock_sock *vsk)
329 vsock_remove_bound(vsk);
330 vsock_remove_connected(vsk);
332 EXPORT_SYMBOL_GPL(vsock_remove_sock);
334 void vsock_for_each_connected_socket(void (*fn)(struct sock *sk))
338 spin_lock_bh(&vsock_table_lock);
340 for (i = 0; i < ARRAY_SIZE(vsock_connected_table); i++) {
341 struct vsock_sock *vsk;
342 list_for_each_entry(vsk, &vsock_connected_table[i],
347 spin_unlock_bh(&vsock_table_lock);
349 EXPORT_SYMBOL_GPL(vsock_for_each_connected_socket);
351 void vsock_add_pending(struct sock *listener, struct sock *pending)
353 struct vsock_sock *vlistener;
354 struct vsock_sock *vpending;
356 vlistener = vsock_sk(listener);
357 vpending = vsock_sk(pending);
361 list_add_tail(&vpending->pending_links, &vlistener->pending_links);
363 EXPORT_SYMBOL_GPL(vsock_add_pending);
365 void vsock_remove_pending(struct sock *listener, struct sock *pending)
367 struct vsock_sock *vpending = vsock_sk(pending);
369 list_del_init(&vpending->pending_links);
373 EXPORT_SYMBOL_GPL(vsock_remove_pending);
375 void vsock_enqueue_accept(struct sock *listener, struct sock *connected)
377 struct vsock_sock *vlistener;
378 struct vsock_sock *vconnected;
380 vlistener = vsock_sk(listener);
381 vconnected = vsock_sk(connected);
383 sock_hold(connected);
385 list_add_tail(&vconnected->accept_queue, &vlistener->accept_queue);
387 EXPORT_SYMBOL_GPL(vsock_enqueue_accept);
389 static void vsock_deassign_transport(struct vsock_sock *vsk)
394 vsk->transport->destruct(vsk);
395 module_put(vsk->transport->module);
396 vsk->transport = NULL;
399 /* Assign a transport to a socket and call the .init transport callback.
401 * Note: for stream socket this must be called when vsk->remote_addr is set
402 * (e.g. during the connect() or when a connection request on a listener
403 * socket is received).
404 * The vsk->remote_addr is used to decide which transport to use:
405 * - remote CID <= VMADDR_CID_HOST will use guest->host transport;
406 * - remote CID == local_cid (guest->host transport) will use guest->host
407 * transport for loopback (host->guest transports don't support loopback);
408 * - remote CID > VMADDR_CID_HOST will use host->guest transport;
410 int vsock_assign_transport(struct vsock_sock *vsk, struct vsock_sock *psk)
412 const struct vsock_transport *new_transport;
413 struct sock *sk = sk_vsock(vsk);
414 unsigned int remote_cid = vsk->remote_addr.svm_cid;
417 switch (sk->sk_type) {
419 new_transport = transport_dgram;
422 if (remote_cid <= VMADDR_CID_HOST ||
424 remote_cid == transport_g2h->get_local_cid()))
425 new_transport = transport_g2h;
427 new_transport = transport_h2g;
430 return -ESOCKTNOSUPPORT;
433 if (vsk->transport) {
434 if (vsk->transport == new_transport)
437 vsk->transport->release(vsk);
438 vsock_deassign_transport(vsk);
441 /* We increase the module refcnt to prevent the transport unloading
442 * while there are open sockets assigned to it.
444 if (!new_transport || !try_module_get(new_transport->module))
447 ret = new_transport->init(vsk, psk);
449 module_put(new_transport->module);
453 vsk->transport = new_transport;
457 EXPORT_SYMBOL_GPL(vsock_assign_transport);
459 bool vsock_find_cid(unsigned int cid)
461 if (transport_g2h && cid == transport_g2h->get_local_cid())
464 if (transport_h2g && cid == VMADDR_CID_HOST)
469 EXPORT_SYMBOL_GPL(vsock_find_cid);
471 static struct sock *vsock_dequeue_accept(struct sock *listener)
473 struct vsock_sock *vlistener;
474 struct vsock_sock *vconnected;
476 vlistener = vsock_sk(listener);
478 if (list_empty(&vlistener->accept_queue))
481 vconnected = list_entry(vlistener->accept_queue.next,
482 struct vsock_sock, accept_queue);
484 list_del_init(&vconnected->accept_queue);
486 /* The caller will need a reference on the connected socket so we let
487 * it call sock_put().
490 return sk_vsock(vconnected);
493 static bool vsock_is_accept_queue_empty(struct sock *sk)
495 struct vsock_sock *vsk = vsock_sk(sk);
496 return list_empty(&vsk->accept_queue);
499 static bool vsock_is_pending(struct sock *sk)
501 struct vsock_sock *vsk = vsock_sk(sk);
502 return !list_empty(&vsk->pending_links);
505 static int vsock_send_shutdown(struct sock *sk, int mode)
507 struct vsock_sock *vsk = vsock_sk(sk);
512 return vsk->transport->shutdown(vsk, mode);
515 static void vsock_pending_work(struct work_struct *work)
518 struct sock *listener;
519 struct vsock_sock *vsk;
522 vsk = container_of(work, struct vsock_sock, pending_work.work);
524 listener = vsk->listener;
528 lock_sock_nested(sk, SINGLE_DEPTH_NESTING);
530 if (vsock_is_pending(sk)) {
531 vsock_remove_pending(listener, sk);
533 sk_acceptq_removed(listener);
534 } else if (!vsk->rejected) {
535 /* We are not on the pending list and accept() did not reject
536 * us, so we must have been accepted by our user process. We
537 * just need to drop our references to the sockets and be on
544 /* We need to remove ourself from the global connected sockets list so
545 * incoming packets can't find this socket, and to reduce the reference
548 vsock_remove_connected(vsk);
550 sk->sk_state = TCP_CLOSE;
554 release_sock(listener);
562 /**** SOCKET OPERATIONS ****/
564 static int __vsock_bind_stream(struct vsock_sock *vsk,
565 struct sockaddr_vm *addr)
568 struct sockaddr_vm new_addr;
571 port = LAST_RESERVED_PORT + 1 +
572 prandom_u32_max(U32_MAX - LAST_RESERVED_PORT);
574 vsock_addr_init(&new_addr, addr->svm_cid, addr->svm_port);
576 if (addr->svm_port == VMADDR_PORT_ANY) {
580 for (i = 0; i < MAX_PORT_RETRIES; i++) {
581 if (port <= LAST_RESERVED_PORT)
582 port = LAST_RESERVED_PORT + 1;
584 new_addr.svm_port = port++;
586 if (!__vsock_find_bound_socket(&new_addr)) {
593 return -EADDRNOTAVAIL;
595 /* If port is in reserved range, ensure caller
596 * has necessary privileges.
598 if (addr->svm_port <= LAST_RESERVED_PORT &&
599 !capable(CAP_NET_BIND_SERVICE)) {
603 if (__vsock_find_bound_socket(&new_addr))
607 vsock_addr_init(&vsk->local_addr, new_addr.svm_cid, new_addr.svm_port);
609 /* Remove stream sockets from the unbound list and add them to the hash
610 * table for easy lookup by its address. The unbound list is simply an
611 * extra entry at the end of the hash table, a trick used by AF_UNIX.
613 __vsock_remove_bound(vsk);
614 __vsock_insert_bound(vsock_bound_sockets(&vsk->local_addr), vsk);
619 static int __vsock_bind_dgram(struct vsock_sock *vsk,
620 struct sockaddr_vm *addr)
622 return vsk->transport->dgram_bind(vsk, addr);
625 static int __vsock_bind(struct sock *sk, struct sockaddr_vm *addr)
627 struct vsock_sock *vsk = vsock_sk(sk);
630 /* First ensure this socket isn't already bound. */
631 if (vsock_addr_bound(&vsk->local_addr))
634 /* Now bind to the provided address or select appropriate values if
635 * none are provided (VMADDR_CID_ANY and VMADDR_PORT_ANY). Note that
636 * like AF_INET prevents binding to a non-local IP address (in most
637 * cases), we only allow binding to a local CID.
639 if (addr->svm_cid != VMADDR_CID_ANY && !vsock_find_cid(addr->svm_cid))
640 return -EADDRNOTAVAIL;
642 switch (sk->sk_socket->type) {
644 spin_lock_bh(&vsock_table_lock);
645 retval = __vsock_bind_stream(vsk, addr);
646 spin_unlock_bh(&vsock_table_lock);
650 retval = __vsock_bind_dgram(vsk, addr);
661 static void vsock_connect_timeout(struct work_struct *work);
663 static struct sock *__vsock_create(struct net *net,
671 struct vsock_sock *psk;
672 struct vsock_sock *vsk;
674 sk = sk_alloc(net, AF_VSOCK, priority, &vsock_proto, kern);
678 sock_init_data(sock, sk);
680 /* sk->sk_type is normally set in sock_init_data, but only if sock is
681 * non-NULL. We make sure that our sockets always have a type by
682 * setting it here if needed.
688 vsock_addr_init(&vsk->local_addr, VMADDR_CID_ANY, VMADDR_PORT_ANY);
689 vsock_addr_init(&vsk->remote_addr, VMADDR_CID_ANY, VMADDR_PORT_ANY);
691 sk->sk_destruct = vsock_sk_destruct;
692 sk->sk_backlog_rcv = vsock_queue_rcv_skb;
693 sock_reset_flag(sk, SOCK_DONE);
695 INIT_LIST_HEAD(&vsk->bound_table);
696 INIT_LIST_HEAD(&vsk->connected_table);
697 vsk->listener = NULL;
698 INIT_LIST_HEAD(&vsk->pending_links);
699 INIT_LIST_HEAD(&vsk->accept_queue);
700 vsk->rejected = false;
701 vsk->sent_request = false;
702 vsk->ignore_connecting_rst = false;
703 vsk->peer_shutdown = 0;
704 INIT_DELAYED_WORK(&vsk->connect_work, vsock_connect_timeout);
705 INIT_DELAYED_WORK(&vsk->pending_work, vsock_pending_work);
707 psk = parent ? vsock_sk(parent) : NULL;
709 vsk->trusted = psk->trusted;
710 vsk->owner = get_cred(psk->owner);
711 vsk->connect_timeout = psk->connect_timeout;
712 vsk->buffer_size = psk->buffer_size;
713 vsk->buffer_min_size = psk->buffer_min_size;
714 vsk->buffer_max_size = psk->buffer_max_size;
716 vsk->trusted = capable(CAP_NET_ADMIN);
717 vsk->owner = get_current_cred();
718 vsk->connect_timeout = VSOCK_DEFAULT_CONNECT_TIMEOUT;
719 vsk->buffer_size = VSOCK_DEFAULT_BUFFER_SIZE;
720 vsk->buffer_min_size = VSOCK_DEFAULT_BUFFER_MIN_SIZE;
721 vsk->buffer_max_size = VSOCK_DEFAULT_BUFFER_MAX_SIZE;
727 static void __vsock_release(struct sock *sk, int level)
730 struct sock *pending;
731 struct vsock_sock *vsk;
734 pending = NULL; /* Compiler warning. */
736 /* The release call is supposed to use lock_sock_nested()
737 * rather than lock_sock(), if a sock lock should be acquired.
740 vsk->transport->release(vsk);
741 else if (sk->sk_type == SOCK_STREAM)
742 vsock_remove_sock(vsk);
744 /* When "level" is SINGLE_DEPTH_NESTING, use the nested
745 * version to avoid the warning "possible recursive locking
746 * detected". When "level" is 0, lock_sock_nested(sk, level)
747 * is the same as lock_sock(sk).
749 lock_sock_nested(sk, level);
751 sk->sk_shutdown = SHUTDOWN_MASK;
753 skb_queue_purge(&sk->sk_receive_queue);
755 /* Clean up any sockets that never were accepted. */
756 while ((pending = vsock_dequeue_accept(sk)) != NULL) {
757 __vsock_release(pending, SINGLE_DEPTH_NESTING);
766 static void vsock_sk_destruct(struct sock *sk)
768 struct vsock_sock *vsk = vsock_sk(sk);
770 vsock_deassign_transport(vsk);
772 /* When clearing these addresses, there's no need to set the family and
773 * possibly register the address family with the kernel.
775 vsock_addr_init(&vsk->local_addr, VMADDR_CID_ANY, VMADDR_PORT_ANY);
776 vsock_addr_init(&vsk->remote_addr, VMADDR_CID_ANY, VMADDR_PORT_ANY);
778 put_cred(vsk->owner);
781 static int vsock_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
785 err = sock_queue_rcv_skb(sk, skb);
792 struct sock *vsock_create_connected(struct sock *parent)
794 return __vsock_create(sock_net(parent), NULL, parent, GFP_KERNEL,
797 EXPORT_SYMBOL_GPL(vsock_create_connected);
799 s64 vsock_stream_has_data(struct vsock_sock *vsk)
801 return vsk->transport->stream_has_data(vsk);
803 EXPORT_SYMBOL_GPL(vsock_stream_has_data);
805 s64 vsock_stream_has_space(struct vsock_sock *vsk)
807 return vsk->transport->stream_has_space(vsk);
809 EXPORT_SYMBOL_GPL(vsock_stream_has_space);
811 static int vsock_release(struct socket *sock)
813 __vsock_release(sock->sk, 0);
815 sock->state = SS_FREE;
821 vsock_bind(struct socket *sock, struct sockaddr *addr, int addr_len)
825 struct sockaddr_vm *vm_addr;
829 if (vsock_addr_cast(addr, addr_len, &vm_addr) != 0)
833 err = __vsock_bind(sk, vm_addr);
839 static int vsock_getname(struct socket *sock,
840 struct sockaddr *addr, int peer)
844 struct vsock_sock *vsk;
845 struct sockaddr_vm *vm_addr;
854 if (sock->state != SS_CONNECTED) {
858 vm_addr = &vsk->remote_addr;
860 vm_addr = &vsk->local_addr;
868 /* sys_getsockname() and sys_getpeername() pass us a
869 * MAX_SOCK_ADDR-sized buffer and don't set addr_len. Unfortunately
870 * that macro is defined in socket.c instead of .h, so we hardcode its
873 BUILD_BUG_ON(sizeof(*vm_addr) > 128);
874 memcpy(addr, vm_addr, sizeof(*vm_addr));
875 err = sizeof(*vm_addr);
882 static int vsock_shutdown(struct socket *sock, int mode)
887 /* User level uses SHUT_RD (0) and SHUT_WR (1), but the kernel uses
888 * RCV_SHUTDOWN (1) and SEND_SHUTDOWN (2), so we must increment mode
889 * here like the other address families do. Note also that the
890 * increment makes SHUT_RDWR (2) into RCV_SHUTDOWN | SEND_SHUTDOWN (3),
891 * which is what we want.
895 if ((mode & ~SHUTDOWN_MASK) || !mode)
898 /* If this is a STREAM socket and it is not connected then bail out
899 * immediately. If it is a DGRAM socket then we must first kick the
900 * socket so that it wakes up from any sleeping calls, for example
901 * recv(), and then afterwards return the error.
905 if (sock->state == SS_UNCONNECTED) {
907 if (sk->sk_type == SOCK_STREAM)
910 sock->state = SS_DISCONNECTING;
914 /* Receive and send shutdowns are treated alike. */
915 mode = mode & (RCV_SHUTDOWN | SEND_SHUTDOWN);
918 sk->sk_shutdown |= mode;
919 sk->sk_state_change(sk);
922 if (sk->sk_type == SOCK_STREAM) {
923 sock_reset_flag(sk, SOCK_DONE);
924 vsock_send_shutdown(sk, mode);
931 static __poll_t vsock_poll(struct file *file, struct socket *sock,
936 struct vsock_sock *vsk;
941 poll_wait(file, sk_sleep(sk), wait);
945 /* Signify that there has been an error on this socket. */
948 /* INET sockets treat local write shutdown and peer write shutdown as a
949 * case of EPOLLHUP set.
951 if ((sk->sk_shutdown == SHUTDOWN_MASK) ||
952 ((sk->sk_shutdown & SEND_SHUTDOWN) &&
953 (vsk->peer_shutdown & SEND_SHUTDOWN))) {
957 if (sk->sk_shutdown & RCV_SHUTDOWN ||
958 vsk->peer_shutdown & SEND_SHUTDOWN) {
962 if (sock->type == SOCK_DGRAM) {
963 /* For datagram sockets we can read if there is something in
964 * the queue and write as long as the socket isn't shutdown for
967 if (!skb_queue_empty_lockless(&sk->sk_receive_queue) ||
968 (sk->sk_shutdown & RCV_SHUTDOWN)) {
969 mask |= EPOLLIN | EPOLLRDNORM;
972 if (!(sk->sk_shutdown & SEND_SHUTDOWN))
973 mask |= EPOLLOUT | EPOLLWRNORM | EPOLLWRBAND;
975 } else if (sock->type == SOCK_STREAM) {
976 const struct vsock_transport *transport = vsk->transport;
979 /* Listening sockets that have connections in their accept
982 if (sk->sk_state == TCP_LISTEN
983 && !vsock_is_accept_queue_empty(sk))
984 mask |= EPOLLIN | EPOLLRDNORM;
986 /* If there is something in the queue then we can read. */
987 if (transport && transport->stream_is_active(vsk) &&
988 !(sk->sk_shutdown & RCV_SHUTDOWN)) {
989 bool data_ready_now = false;
990 int ret = transport->notify_poll_in(
991 vsk, 1, &data_ready_now);
996 mask |= EPOLLIN | EPOLLRDNORM;
1001 /* Sockets whose connections have been closed, reset, or
1002 * terminated should also be considered read, and we check the
1003 * shutdown flag for that.
1005 if (sk->sk_shutdown & RCV_SHUTDOWN ||
1006 vsk->peer_shutdown & SEND_SHUTDOWN) {
1007 mask |= EPOLLIN | EPOLLRDNORM;
1010 /* Connected sockets that can produce data can be written. */
1011 if (sk->sk_state == TCP_ESTABLISHED) {
1012 if (!(sk->sk_shutdown & SEND_SHUTDOWN)) {
1013 bool space_avail_now = false;
1014 int ret = transport->notify_poll_out(
1015 vsk, 1, &space_avail_now);
1019 if (space_avail_now)
1020 /* Remove EPOLLWRBAND since INET
1021 * sockets are not setting it.
1023 mask |= EPOLLOUT | EPOLLWRNORM;
1029 /* Simulate INET socket poll behaviors, which sets
1030 * EPOLLOUT|EPOLLWRNORM when peer is closed and nothing to read,
1031 * but local send is not shutdown.
1033 if (sk->sk_state == TCP_CLOSE || sk->sk_state == TCP_CLOSING) {
1034 if (!(sk->sk_shutdown & SEND_SHUTDOWN))
1035 mask |= EPOLLOUT | EPOLLWRNORM;
1045 static int vsock_dgram_sendmsg(struct socket *sock, struct msghdr *msg,
1050 struct vsock_sock *vsk;
1051 struct sockaddr_vm *remote_addr;
1052 const struct vsock_transport *transport;
1054 if (msg->msg_flags & MSG_OOB)
1057 /* For now, MSG_DONTWAIT is always assumed... */
1061 transport = vsk->transport;
1065 err = vsock_auto_bind(vsk);
1070 /* If the provided message contains an address, use that. Otherwise
1071 * fall back on the socket's remote handle (if it has been connected).
1073 if (msg->msg_name &&
1074 vsock_addr_cast(msg->msg_name, msg->msg_namelen,
1075 &remote_addr) == 0) {
1076 /* Ensure this address is of the right type and is a valid
1080 if (remote_addr->svm_cid == VMADDR_CID_ANY)
1081 remote_addr->svm_cid = transport->get_local_cid();
1083 if (!vsock_addr_bound(remote_addr)) {
1087 } else if (sock->state == SS_CONNECTED) {
1088 remote_addr = &vsk->remote_addr;
1090 if (remote_addr->svm_cid == VMADDR_CID_ANY)
1091 remote_addr->svm_cid = transport->get_local_cid();
1093 /* XXX Should connect() or this function ensure remote_addr is
1096 if (!vsock_addr_bound(&vsk->remote_addr)) {
1105 if (!transport->dgram_allow(remote_addr->svm_cid,
1106 remote_addr->svm_port)) {
1111 err = transport->dgram_enqueue(vsk, remote_addr, msg, len);
1118 static int vsock_dgram_connect(struct socket *sock,
1119 struct sockaddr *addr, int addr_len, int flags)
1123 struct vsock_sock *vsk;
1124 struct sockaddr_vm *remote_addr;
1129 err = vsock_addr_cast(addr, addr_len, &remote_addr);
1130 if (err == -EAFNOSUPPORT && remote_addr->svm_family == AF_UNSPEC) {
1132 vsock_addr_init(&vsk->remote_addr, VMADDR_CID_ANY,
1134 sock->state = SS_UNCONNECTED;
1137 } else if (err != 0)
1142 err = vsock_auto_bind(vsk);
1146 if (!vsk->transport->dgram_allow(remote_addr->svm_cid,
1147 remote_addr->svm_port)) {
1152 memcpy(&vsk->remote_addr, remote_addr, sizeof(vsk->remote_addr));
1153 sock->state = SS_CONNECTED;
1160 static int vsock_dgram_recvmsg(struct socket *sock, struct msghdr *msg,
1161 size_t len, int flags)
1163 struct vsock_sock *vsk = vsock_sk(sock->sk);
1165 return vsk->transport->dgram_dequeue(vsk, msg, len, flags);
1168 static const struct proto_ops vsock_dgram_ops = {
1170 .owner = THIS_MODULE,
1171 .release = vsock_release,
1173 .connect = vsock_dgram_connect,
1174 .socketpair = sock_no_socketpair,
1175 .accept = sock_no_accept,
1176 .getname = vsock_getname,
1178 .ioctl = sock_no_ioctl,
1179 .listen = sock_no_listen,
1180 .shutdown = vsock_shutdown,
1181 .setsockopt = sock_no_setsockopt,
1182 .getsockopt = sock_no_getsockopt,
1183 .sendmsg = vsock_dgram_sendmsg,
1184 .recvmsg = vsock_dgram_recvmsg,
1185 .mmap = sock_no_mmap,
1186 .sendpage = sock_no_sendpage,
1189 static int vsock_transport_cancel_pkt(struct vsock_sock *vsk)
1191 const struct vsock_transport *transport = vsk->transport;
1193 if (!transport->cancel_pkt)
1196 return transport->cancel_pkt(vsk);
1199 static void vsock_connect_timeout(struct work_struct *work)
1202 struct vsock_sock *vsk;
1205 vsk = container_of(work, struct vsock_sock, connect_work.work);
1209 if (sk->sk_state == TCP_SYN_SENT &&
1210 (sk->sk_shutdown != SHUTDOWN_MASK)) {
1211 sk->sk_state = TCP_CLOSE;
1212 sk->sk_err = ETIMEDOUT;
1213 sk->sk_error_report(sk);
1218 vsock_transport_cancel_pkt(vsk);
1223 static int vsock_stream_connect(struct socket *sock, struct sockaddr *addr,
1224 int addr_len, int flags)
1228 struct vsock_sock *vsk;
1229 const struct vsock_transport *transport;
1230 struct sockaddr_vm *remote_addr;
1240 /* XXX AF_UNSPEC should make us disconnect like AF_INET. */
1241 switch (sock->state) {
1245 case SS_DISCONNECTING:
1249 /* This continues on so we can move sock into the SS_CONNECTED
1250 * state once the connection has completed (at which point err
1251 * will be set to zero also). Otherwise, we will either wait
1252 * for the connection or return -EALREADY should this be a
1253 * non-blocking call.
1258 if ((sk->sk_state == TCP_LISTEN) ||
1259 vsock_addr_cast(addr, addr_len, &remote_addr) != 0) {
1264 /* Set the remote address that we are connecting to. */
1265 memcpy(&vsk->remote_addr, remote_addr,
1266 sizeof(vsk->remote_addr));
1268 err = vsock_assign_transport(vsk, NULL);
1272 transport = vsk->transport;
1274 /* The hypervisor and well-known contexts do not have socket
1278 !transport->stream_allow(remote_addr->svm_cid,
1279 remote_addr->svm_port)) {
1284 err = vsock_auto_bind(vsk);
1288 sk->sk_state = TCP_SYN_SENT;
1290 err = transport->connect(vsk);
1294 /* Mark sock as connecting and set the error code to in
1295 * progress in case this is a non-blocking connect.
1297 sock->state = SS_CONNECTING;
1301 /* The receive path will handle all communication until we are able to
1302 * enter the connected state. Here we wait for the connection to be
1303 * completed or a notification of an error.
1305 timeout = vsk->connect_timeout;
1306 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
1308 while (sk->sk_state != TCP_ESTABLISHED && sk->sk_err == 0) {
1309 if (flags & O_NONBLOCK) {
1310 /* If we're not going to block, we schedule a timeout
1311 * function to generate a timeout on the connection
1312 * attempt, in case the peer doesn't respond in a
1313 * timely manner. We hold on to the socket until the
1317 schedule_delayed_work(&vsk->connect_work, timeout);
1319 /* Skip ahead to preserve error code set above. */
1324 timeout = schedule_timeout(timeout);
1327 if (signal_pending(current)) {
1328 err = sock_intr_errno(timeout);
1329 sk->sk_state = TCP_CLOSE;
1330 sock->state = SS_UNCONNECTED;
1331 vsock_transport_cancel_pkt(vsk);
1333 } else if (timeout == 0) {
1335 sk->sk_state = TCP_CLOSE;
1336 sock->state = SS_UNCONNECTED;
1337 vsock_transport_cancel_pkt(vsk);
1341 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
1346 sk->sk_state = TCP_CLOSE;
1347 sock->state = SS_UNCONNECTED;
1353 finish_wait(sk_sleep(sk), &wait);
1359 static int vsock_accept(struct socket *sock, struct socket *newsock, int flags,
1362 struct sock *listener;
1364 struct sock *connected;
1365 struct vsock_sock *vconnected;
1370 listener = sock->sk;
1372 lock_sock(listener);
1374 if (sock->type != SOCK_STREAM) {
1379 if (listener->sk_state != TCP_LISTEN) {
1384 /* Wait for children sockets to appear; these are the new sockets
1385 * created upon connection establishment.
1387 timeout = sock_sndtimeo(listener, flags & O_NONBLOCK);
1388 prepare_to_wait(sk_sleep(listener), &wait, TASK_INTERRUPTIBLE);
1390 while ((connected = vsock_dequeue_accept(listener)) == NULL &&
1391 listener->sk_err == 0) {
1392 release_sock(listener);
1393 timeout = schedule_timeout(timeout);
1394 finish_wait(sk_sleep(listener), &wait);
1395 lock_sock(listener);
1397 if (signal_pending(current)) {
1398 err = sock_intr_errno(timeout);
1400 } else if (timeout == 0) {
1405 prepare_to_wait(sk_sleep(listener), &wait, TASK_INTERRUPTIBLE);
1407 finish_wait(sk_sleep(listener), &wait);
1409 if (listener->sk_err)
1410 err = -listener->sk_err;
1413 sk_acceptq_removed(listener);
1415 lock_sock_nested(connected, SINGLE_DEPTH_NESTING);
1416 vconnected = vsock_sk(connected);
1418 /* If the listener socket has received an error, then we should
1419 * reject this socket and return. Note that we simply mark the
1420 * socket rejected, drop our reference, and let the cleanup
1421 * function handle the cleanup; the fact that we found it in
1422 * the listener's accept queue guarantees that the cleanup
1423 * function hasn't run yet.
1426 vconnected->rejected = true;
1428 newsock->state = SS_CONNECTED;
1429 sock_graft(connected, newsock);
1432 release_sock(connected);
1433 sock_put(connected);
1437 release_sock(listener);
1441 static int vsock_listen(struct socket *sock, int backlog)
1445 struct vsock_sock *vsk;
1451 if (sock->type != SOCK_STREAM) {
1456 if (sock->state != SS_UNCONNECTED) {
1463 if (!vsock_addr_bound(&vsk->local_addr)) {
1468 sk->sk_max_ack_backlog = backlog;
1469 sk->sk_state = TCP_LISTEN;
1478 static void vsock_update_buffer_size(struct vsock_sock *vsk,
1479 const struct vsock_transport *transport,
1482 if (val > vsk->buffer_max_size)
1483 val = vsk->buffer_max_size;
1485 if (val < vsk->buffer_min_size)
1486 val = vsk->buffer_min_size;
1488 if (val != vsk->buffer_size &&
1489 transport && transport->notify_buffer_size)
1490 transport->notify_buffer_size(vsk, &val);
1492 vsk->buffer_size = val;
1495 static int vsock_stream_setsockopt(struct socket *sock,
1498 char __user *optval,
1499 unsigned int optlen)
1503 struct vsock_sock *vsk;
1504 const struct vsock_transport *transport;
1507 if (level != AF_VSOCK)
1508 return -ENOPROTOOPT;
1510 #define COPY_IN(_v) \
1512 if (optlen < sizeof(_v)) { \
1516 if (copy_from_user(&_v, optval, sizeof(_v)) != 0) { \
1525 transport = vsk->transport;
1530 case SO_VM_SOCKETS_BUFFER_SIZE:
1532 vsock_update_buffer_size(vsk, transport, val);
1535 case SO_VM_SOCKETS_BUFFER_MAX_SIZE:
1537 vsk->buffer_max_size = val;
1538 vsock_update_buffer_size(vsk, transport, vsk->buffer_size);
1541 case SO_VM_SOCKETS_BUFFER_MIN_SIZE:
1543 vsk->buffer_min_size = val;
1544 vsock_update_buffer_size(vsk, transport, vsk->buffer_size);
1547 case SO_VM_SOCKETS_CONNECT_TIMEOUT: {
1548 struct __kernel_old_timeval tv;
1550 if (tv.tv_sec >= 0 && tv.tv_usec < USEC_PER_SEC &&
1551 tv.tv_sec < (MAX_SCHEDULE_TIMEOUT / HZ - 1)) {
1552 vsk->connect_timeout = tv.tv_sec * HZ +
1553 DIV_ROUND_UP(tv.tv_usec, (1000000 / HZ));
1554 if (vsk->connect_timeout == 0)
1555 vsk->connect_timeout =
1556 VSOCK_DEFAULT_CONNECT_TIMEOUT;
1576 static int vsock_stream_getsockopt(struct socket *sock,
1577 int level, int optname,
1578 char __user *optval,
1584 struct vsock_sock *vsk;
1587 if (level != AF_VSOCK)
1588 return -ENOPROTOOPT;
1590 err = get_user(len, optlen);
1594 #define COPY_OUT(_v) \
1596 if (len < sizeof(_v)) \
1600 if (copy_to_user(optval, &_v, len) != 0) \
1610 case SO_VM_SOCKETS_BUFFER_SIZE:
1611 val = vsk->buffer_size;
1615 case SO_VM_SOCKETS_BUFFER_MAX_SIZE:
1616 val = vsk->buffer_max_size;
1620 case SO_VM_SOCKETS_BUFFER_MIN_SIZE:
1621 val = vsk->buffer_min_size;
1625 case SO_VM_SOCKETS_CONNECT_TIMEOUT: {
1626 struct __kernel_old_timeval tv;
1627 tv.tv_sec = vsk->connect_timeout / HZ;
1629 (vsk->connect_timeout -
1630 tv.tv_sec * HZ) * (1000000 / HZ);
1635 return -ENOPROTOOPT;
1638 err = put_user(len, optlen);
1647 static int vsock_stream_sendmsg(struct socket *sock, struct msghdr *msg,
1651 struct vsock_sock *vsk;
1652 const struct vsock_transport *transport;
1653 ssize_t total_written;
1656 struct vsock_transport_send_notify_data send_data;
1657 DEFINE_WAIT_FUNC(wait, woken_wake_function);
1661 transport = vsk->transport;
1665 if (msg->msg_flags & MSG_OOB)
1670 /* Callers should not provide a destination with stream sockets. */
1671 if (msg->msg_namelen) {
1672 err = sk->sk_state == TCP_ESTABLISHED ? -EISCONN : -EOPNOTSUPP;
1676 /* Send data only if both sides are not shutdown in the direction. */
1677 if (sk->sk_shutdown & SEND_SHUTDOWN ||
1678 vsk->peer_shutdown & RCV_SHUTDOWN) {
1683 if (!transport || sk->sk_state != TCP_ESTABLISHED ||
1684 !vsock_addr_bound(&vsk->local_addr)) {
1689 if (!vsock_addr_bound(&vsk->remote_addr)) {
1690 err = -EDESTADDRREQ;
1694 /* Wait for room in the produce queue to enqueue our user's data. */
1695 timeout = sock_sndtimeo(sk, msg->msg_flags & MSG_DONTWAIT);
1697 err = transport->notify_send_init(vsk, &send_data);
1701 while (total_written < len) {
1704 add_wait_queue(sk_sleep(sk), &wait);
1705 while (vsock_stream_has_space(vsk) == 0 &&
1707 !(sk->sk_shutdown & SEND_SHUTDOWN) &&
1708 !(vsk->peer_shutdown & RCV_SHUTDOWN)) {
1710 /* Don't wait for non-blocking sockets. */
1713 remove_wait_queue(sk_sleep(sk), &wait);
1717 err = transport->notify_send_pre_block(vsk, &send_data);
1719 remove_wait_queue(sk_sleep(sk), &wait);
1724 timeout = wait_woken(&wait, TASK_INTERRUPTIBLE, timeout);
1726 if (signal_pending(current)) {
1727 err = sock_intr_errno(timeout);
1728 remove_wait_queue(sk_sleep(sk), &wait);
1730 } else if (timeout == 0) {
1732 remove_wait_queue(sk_sleep(sk), &wait);
1736 remove_wait_queue(sk_sleep(sk), &wait);
1738 /* These checks occur both as part of and after the loop
1739 * conditional since we need to check before and after
1745 } else if ((sk->sk_shutdown & SEND_SHUTDOWN) ||
1746 (vsk->peer_shutdown & RCV_SHUTDOWN)) {
1751 err = transport->notify_send_pre_enqueue(vsk, &send_data);
1755 /* Note that enqueue will only write as many bytes as are free
1756 * in the produce queue, so we don't need to ensure len is
1757 * smaller than the queue size. It is the caller's
1758 * responsibility to check how many bytes we were able to send.
1761 written = transport->stream_enqueue(
1763 len - total_written);
1769 total_written += written;
1771 err = transport->notify_send_post_enqueue(
1772 vsk, written, &send_data);
1779 if (total_written > 0)
1780 err = total_written;
1788 vsock_stream_recvmsg(struct socket *sock, struct msghdr *msg, size_t len,
1792 struct vsock_sock *vsk;
1793 const struct vsock_transport *transport;
1798 struct vsock_transport_recv_notify_data recv_data;
1804 transport = vsk->transport;
1809 if (!transport || sk->sk_state != TCP_ESTABLISHED) {
1810 /* Recvmsg is supposed to return 0 if a peer performs an
1811 * orderly shutdown. Differentiate between that case and when a
1812 * peer has not connected or a local shutdown occured with the
1815 if (sock_flag(sk, SOCK_DONE))
1823 if (flags & MSG_OOB) {
1828 /* We don't check peer_shutdown flag here since peer may actually shut
1829 * down, but there can be data in the queue that a local socket can
1832 if (sk->sk_shutdown & RCV_SHUTDOWN) {
1837 /* It is valid on Linux to pass in a zero-length receive buffer. This
1838 * is not an error. We may as well bail out now.
1845 /* We must not copy less than target bytes into the user's buffer
1846 * before returning successfully, so we wait for the consume queue to
1847 * have that much data to consume before dequeueing. Note that this
1848 * makes it impossible to handle cases where target is greater than the
1851 target = sock_rcvlowat(sk, flags & MSG_WAITALL, len);
1852 if (target >= transport->stream_rcvhiwat(vsk)) {
1856 timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
1859 err = transport->notify_recv_init(vsk, target, &recv_data);
1867 prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
1868 ready = vsock_stream_has_data(vsk);
1871 if (sk->sk_err != 0 ||
1872 (sk->sk_shutdown & RCV_SHUTDOWN) ||
1873 (vsk->peer_shutdown & SEND_SHUTDOWN)) {
1874 finish_wait(sk_sleep(sk), &wait);
1877 /* Don't wait for non-blocking sockets. */
1880 finish_wait(sk_sleep(sk), &wait);
1884 err = transport->notify_recv_pre_block(
1885 vsk, target, &recv_data);
1887 finish_wait(sk_sleep(sk), &wait);
1891 timeout = schedule_timeout(timeout);
1894 if (signal_pending(current)) {
1895 err = sock_intr_errno(timeout);
1896 finish_wait(sk_sleep(sk), &wait);
1898 } else if (timeout == 0) {
1900 finish_wait(sk_sleep(sk), &wait);
1906 finish_wait(sk_sleep(sk), &wait);
1909 /* Invalid queue pair content. XXX This should
1910 * be changed to a connection reset in a later
1918 err = transport->notify_recv_pre_dequeue(
1919 vsk, target, &recv_data);
1923 read = transport->stream_dequeue(
1925 len - copied, flags);
1933 err = transport->notify_recv_post_dequeue(
1935 !(flags & MSG_PEEK), &recv_data);
1939 if (read >= target || flags & MSG_PEEK)
1948 else if (sk->sk_shutdown & RCV_SHUTDOWN)
1959 static const struct proto_ops vsock_stream_ops = {
1961 .owner = THIS_MODULE,
1962 .release = vsock_release,
1964 .connect = vsock_stream_connect,
1965 .socketpair = sock_no_socketpair,
1966 .accept = vsock_accept,
1967 .getname = vsock_getname,
1969 .ioctl = sock_no_ioctl,
1970 .listen = vsock_listen,
1971 .shutdown = vsock_shutdown,
1972 .setsockopt = vsock_stream_setsockopt,
1973 .getsockopt = vsock_stream_getsockopt,
1974 .sendmsg = vsock_stream_sendmsg,
1975 .recvmsg = vsock_stream_recvmsg,
1976 .mmap = sock_no_mmap,
1977 .sendpage = sock_no_sendpage,
1980 static int vsock_create(struct net *net, struct socket *sock,
1981 int protocol, int kern)
1983 struct vsock_sock *vsk;
1990 if (protocol && protocol != PF_VSOCK)
1991 return -EPROTONOSUPPORT;
1993 switch (sock->type) {
1995 sock->ops = &vsock_dgram_ops;
1998 sock->ops = &vsock_stream_ops;
2001 return -ESOCKTNOSUPPORT;
2004 sock->state = SS_UNCONNECTED;
2006 sk = __vsock_create(net, sock, NULL, GFP_KERNEL, 0, kern);
2012 if (sock->type == SOCK_DGRAM) {
2013 ret = vsock_assign_transport(vsk, NULL);
2020 vsock_insert_unbound(vsk);
2025 static const struct net_proto_family vsock_family_ops = {
2027 .create = vsock_create,
2028 .owner = THIS_MODULE,
2031 static long vsock_dev_do_ioctl(struct file *filp,
2032 unsigned int cmd, void __user *ptr)
2034 u32 __user *p = ptr;
2035 u32 cid = VMADDR_CID_ANY;
2039 case IOCTL_VM_SOCKETS_GET_LOCAL_CID:
2040 /* To be compatible with the VMCI behavior, we prioritize the
2041 * guest CID instead of well-know host CID (VMADDR_CID_HOST).
2044 cid = transport_g2h->get_local_cid();
2045 else if (transport_h2g)
2046 cid = transport_h2g->get_local_cid();
2048 if (put_user(cid, p) != 0)
2053 pr_err("Unknown ioctl %d\n", cmd);
2060 static long vsock_dev_ioctl(struct file *filp,
2061 unsigned int cmd, unsigned long arg)
2063 return vsock_dev_do_ioctl(filp, cmd, (void __user *)arg);
2066 #ifdef CONFIG_COMPAT
2067 static long vsock_dev_compat_ioctl(struct file *filp,
2068 unsigned int cmd, unsigned long arg)
2070 return vsock_dev_do_ioctl(filp, cmd, compat_ptr(arg));
2074 static const struct file_operations vsock_device_ops = {
2075 .owner = THIS_MODULE,
2076 .unlocked_ioctl = vsock_dev_ioctl,
2077 #ifdef CONFIG_COMPAT
2078 .compat_ioctl = vsock_dev_compat_ioctl,
2080 .open = nonseekable_open,
2083 static struct miscdevice vsock_device = {
2085 .fops = &vsock_device_ops,
2088 static int __init vsock_init(void)
2092 vsock_init_tables();
2094 vsock_proto.owner = THIS_MODULE;
2095 vsock_device.minor = MISC_DYNAMIC_MINOR;
2096 err = misc_register(&vsock_device);
2098 pr_err("Failed to register misc device\n");
2099 goto err_reset_transport;
2102 err = proto_register(&vsock_proto, 1); /* we want our slab */
2104 pr_err("Cannot register vsock protocol\n");
2105 goto err_deregister_misc;
2108 err = sock_register(&vsock_family_ops);
2110 pr_err("could not register af_vsock (%d) address family: %d\n",
2112 goto err_unregister_proto;
2117 err_unregister_proto:
2118 proto_unregister(&vsock_proto);
2119 err_deregister_misc:
2120 misc_deregister(&vsock_device);
2121 err_reset_transport:
2125 static void __exit vsock_exit(void)
2127 misc_deregister(&vsock_device);
2128 sock_unregister(AF_VSOCK);
2129 proto_unregister(&vsock_proto);
2132 const struct vsock_transport *vsock_core_get_transport(struct vsock_sock *vsk)
2134 return vsk->transport;
2136 EXPORT_SYMBOL_GPL(vsock_core_get_transport);
2138 int vsock_core_register(const struct vsock_transport *t, int features)
2140 const struct vsock_transport *t_h2g, *t_g2h, *t_dgram;
2141 int err = mutex_lock_interruptible(&vsock_register_mutex);
2146 t_h2g = transport_h2g;
2147 t_g2h = transport_g2h;
2148 t_dgram = transport_dgram;
2150 if (features & VSOCK_TRANSPORT_F_H2G) {
2158 if (features & VSOCK_TRANSPORT_F_G2H) {
2166 if (features & VSOCK_TRANSPORT_F_DGRAM) {
2174 transport_h2g = t_h2g;
2175 transport_g2h = t_g2h;
2176 transport_dgram = t_dgram;
2179 mutex_unlock(&vsock_register_mutex);
2182 EXPORT_SYMBOL_GPL(vsock_core_register);
2184 void vsock_core_unregister(const struct vsock_transport *t)
2186 mutex_lock(&vsock_register_mutex);
2188 if (transport_h2g == t)
2189 transport_h2g = NULL;
2191 if (transport_g2h == t)
2192 transport_g2h = NULL;
2194 if (transport_dgram == t)
2195 transport_dgram = NULL;
2197 mutex_unlock(&vsock_register_mutex);
2199 EXPORT_SYMBOL_GPL(vsock_core_unregister);
2201 module_init(vsock_init);
2202 module_exit(vsock_exit);
2204 MODULE_AUTHOR("VMware, Inc.");
2205 MODULE_DESCRIPTION("VMware Virtual Socket Family");
2206 MODULE_VERSION("1.0.2.0-k");
2207 MODULE_LICENSE("GPL v2");