Merge branch 'vsock'

Andy King says:

====================
In an effort to improve the out-of-the-box experience with Linux kernels for
VMware users, VMware is working on readying the VM Sockets (VSOCK, formerly
VMCI Sockets) (vsock) kernel module for inclusion in the Linux kernel. The
purpose of this post is to acquire feedback on the vsock kernel module.

Unlike previous submissions, where the new socket family was entirely reliant
on VMware's VMCI PCI device (and thus VMware's hypervisor), VM Sockets is now
completely[1] separated out into two parts, each in its own module:

o Core socket code, which is transport-neutral and invokes transport
  callbacks to communicate with the hypervisor.  This is vsock.ko.
o A VMCI transport, which communicates over VMCI with the VMware hypervisor.
  This is vmw_vsock_vmci_transport.ko, and it registers with the core module
  as a transport.

This should provide a path to introducing additional transports, for example
virtio, with the ultimate goal being to make this new socket family
hypervisor-neutral.

[1] If Gerd tries it and determines this to be false (still), I'll ship him
    a keg of beer.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

include/linux/socket.h

@@ -178,7 +178,8 @@ struct ucred {
 #define AF_CAIF		37	/* CAIF sockets			*/
 #define AF_ALG		38	/* Algorithm sockets		*/
 #define AF_NFC		39	/* NFC sockets			*/
-#define AF_MAX		40	/* For now.. */
+#define AF_VSOCK	40	/* vSockets			*/
+#define AF_MAX		41	/* For now.. */
 
 /* Protocol families, same as address families. */
 #define PF_UNSPEC	AF_UNSPEC
@@ -221,6 +222,7 @@ struct ucred {
 #define PF_CAIF		AF_CAIF
 #define PF_ALG		AF_ALG
 #define PF_NFC		AF_NFC
+#define PF_VSOCK	AF_VSOCK
 #define PF_MAX		AF_MAX
 
 /* Maximum queue length specifiable by listen.  */

include/uapi/linux/vm_sockets.h

@@ -0,0 +1,171 @@
+/*
+ * VMware vSockets Driver
+ *
+ * Copyright (C) 2007-2013 VMware, Inc. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation version 2 and no later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ */
+
+#ifndef _VM_SOCKETS_H_
+#define _VM_SOCKETS_H_
+
+#if !defined(__KERNEL__)
+#include <sys/socket.h>
+#endif
+
+/* Option name for STREAM socket buffer size.  Use as the option name in
+ * setsockopt(3) or getsockopt(3) to set or get an unsigned long long that
+ * specifies the size of the buffer underlying a vSockets STREAM socket.
+ * Value is clamped to the MIN and MAX.
+ */
+
+#define SO_VM_SOCKETS_BUFFER_SIZE 0
+
+/* Option name for STREAM socket minimum buffer size.  Use as the option name
+ * in setsockopt(3) or getsockopt(3) to set or get an unsigned long long that
+ * specifies the minimum size allowed for the buffer underlying a vSockets
+ * STREAM socket.
+ */
+
+#define SO_VM_SOCKETS_BUFFER_MIN_SIZE 1
+
+/* Option name for STREAM socket maximum buffer size.  Use as the option name
+ * in setsockopt(3) or getsockopt(3) to set or get an unsigned long long
+ * that specifies the maximum size allowed for the buffer underlying a
+ * vSockets STREAM socket.
+ */
+
+#define SO_VM_SOCKETS_BUFFER_MAX_SIZE 2
+
+/* Option name for socket peer's host-specific VM ID.  Use as the option name
+ * in getsockopt(3) to get a host-specific identifier for the peer endpoint's
+ * VM.  The identifier is a signed integer.
+ * Only available for hypervisor endpoints.
+ */
+
+#define SO_VM_SOCKETS_PEER_HOST_VM_ID 3
+
+/* Option name for socket's service label.  Use as the option name in
+ * setsockopt(3) or getsockopt(3) to set or get the service label for a socket.
+ * The service label is a C-style NUL-terminated string.  Only available for
+ * hypervisor endpoints.
+ */
+
+#define SO_VM_SOCKETS_SERVICE_LABEL 4
+
+/* Option name for determining if a socket is trusted.  Use as the option name
+ * in getsockopt(3) to determine if a socket is trusted.  The value is a
+ * signed integer.
+ */
+
+#define SO_VM_SOCKETS_TRUSTED 5
+
+/* Option name for STREAM socket connection timeout.  Use as the option name
+ * in setsockopt(3) or getsockopt(3) to set or get the connection
+ * timeout for a STREAM socket.
+ */
+
+#define SO_VM_SOCKETS_CONNECT_TIMEOUT 6
+
+/* Option name for using non-blocking send/receive.  Use as the option name
+ * for setsockopt(3) or getsockopt(3) to set or get the non-blocking
+ * transmit/receive flag for a STREAM socket.  This flag determines whether
+ * send() and recv() can be called in non-blocking contexts for the given
+ * socket.  The value is a signed integer.
+ *
+ * This option is only relevant to kernel endpoints, where descheduling the
+ * thread of execution is not allowed, for example, while holding a spinlock.
+ * It is not to be confused with conventional non-blocking socket operations.
+ *
+ * Only available for hypervisor endpoints.
+ */
+
+#define SO_VM_SOCKETS_NONBLOCK_TXRX 7
+
+/* The vSocket equivalent of INADDR_ANY.  This works for the svm_cid field of
+ * sockaddr_vm and indicates the context ID of the current endpoint.
+ */
+
+#define VMADDR_CID_ANY -1U
+
+/* Bind to any available port.  Works for the svm_port field of
+ * sockaddr_vm.
+ */
+
+#define VMADDR_PORT_ANY -1U
+
+/* Use this as the destination CID in an address when referring to the
+ * hypervisor.  VMCI relies on it being 0, but this would be useful for other
+ * transports too.
+ */
+
+#define VMADDR_CID_HYPERVISOR 0
+
+/* This CID is specific to VMCI and can be considered reserved (even VMCI
+ * doesn't use it anymore, it's a legacy value from an older release).
+ */
+
+#define VMADDR_CID_RESERVED 1
+
+/* Use this as the destination CID in an address when referring to the host
+ * (any process other than the hypervisor).  VMCI relies on it being 2, but
+ * this would be useful for other transports too.
+ */
+
+#define VMADDR_CID_HOST 2
+
+/* Invalid vSockets version. */
+
+#define VM_SOCKETS_INVALID_VERSION -1U
+
+/* The epoch (first) component of the vSockets version.  A single byte
+ * representing the epoch component of the vSockets version.
+ */
+
+#define VM_SOCKETS_VERSION_EPOCH(_v) (((_v) & 0xFF000000) >> 24)
+
+/* The major (second) component of the vSockets version.   A single byte
+ * representing the major component of the vSockets version.  Typically
+ * changes for every major release of a product.
+ */
+
+#define VM_SOCKETS_VERSION_MAJOR(_v) (((_v) & 0x00FF0000) >> 16)
+
+/* The minor (third) component of the vSockets version.  Two bytes representing
+ * the minor component of the vSockets version.
+ */
+
+#define VM_SOCKETS_VERSION_MINOR(_v) (((_v) & 0x0000FFFF))
+
+/* Address structure for vSockets.   The address family should be set to
+ * whatever vmci_sock_get_af_value_fd() returns.  The structure members should
+ * all align on their natural boundaries without resorting to compiler packing
+ * directives.  The total size of this structure should be exactly the same as
+ * that of struct sockaddr.
+ */
+
+struct sockaddr_vm {
+	sa_family_t svm_family;
+	unsigned short svm_reserved1;
+	unsigned int svm_port;
+	unsigned int svm_cid;
+	unsigned char svm_zero[sizeof(struct sockaddr) -
+			       sizeof(sa_family_t) -
+			       sizeof(unsigned short) -
+			       sizeof(unsigned int) - sizeof(unsigned int)];
+};
+
+#define IOCTL_VM_SOCKETS_GET_LOCAL_CID		_IO(7, 0xb9)
+
+#if defined(__KERNEL__)
+int vm_sockets_get_local_cid(void);
+#endif
+
+#endif

net/Kconfig

@@ -217,6 +217,7 @@ source "net/dcb/Kconfig"
 source "net/dns_resolver/Kconfig"
 source "net/batman-adv/Kconfig"
 source "net/openvswitch/Kconfig"
+source "net/vmw_vsock/Kconfig"
 
 config RPS
 	boolean

net/Makefile

@@ -69,3 +69,4 @@ obj-$(CONFIG_CEPH_LIB)		+= ceph/
 obj-$(CONFIG_BATMAN_ADV)	+= batman-adv/
 obj-$(CONFIG_NFC)		+= nfc/
 obj-$(CONFIG_OPENVSWITCH)	+= openvswitch/
+obj-$(CONFIG_VSOCKETS)	+= vmw_vsock/

net/vmw_vsock/Kconfig

@@ -0,0 +1,28 @@
+#
+# Vsock protocol
+#
+
+config VSOCKETS
+	tristate "Virtual Socket protocol"
+	help
+	  Virtual Socket Protocol is a socket protocol similar to TCP/IP
+	  allowing comunication between Virtual Machines and hypervisor
+	  or host.
+
+	  You should also select one or more hypervisor-specific transports
+	  below.
+
+	  To compile this driver as a module, choose M here: the module
+	  will be called vsock. If unsure, say N.
+
+config VMWARE_VMCI_VSOCKETS
+	tristate "VMware VMCI transport for Virtual Sockets"
+	depends on VSOCKETS && VMWARE_VMCI
+	help
+	  This module implements a VMCI transport for Virtual Sockets.
+
+	  Enable this transport if your Virtual Machine runs on a VMware
+	  hypervisor.
+
+	  To compile this driver as a module, choose M here: the module
+	  will be called vmw_vsock_vmci_transport. If unsure, say N.

net/vmw_vsock/Makefile

@@ -0,0 +1,7 @@
+obj-$(CONFIG_VSOCKETS) += vsock.o
+obj-$(CONFIG_VMWARE_VMCI_VSOCKETS) += vmw_vsock_vmci_transport.o
+
+vsock-y += af_vsock.o vsock_addr.o
+
+vmw_vsock_vmci_transport-y += vmci_transport.o vmci_transport_notify.o \
+	vmci_transport_notify_qstate.o

net/vmw_vsock/af_vsock.c

@@ -0,0 +1,2015 @@
+/*
+ * VMware vSockets Driver
+ *
+ * Copyright (C) 2007-2013 VMware, Inc. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation version 2 and no later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ */
+
+/* Implementation notes:
+ *
+ * - There are two kinds of sockets: those created by user action (such as
+ * calling socket(2)) and those created by incoming connection request packets.
+ *
+ * - There are two "global" tables, one for bound sockets (sockets that have
+ * specified an address that they are responsible for) and one for connected
+ * sockets (sockets that have established a connection with another socket).
+ * These tables are "global" in that all sockets on the system are placed
+ * within them. - Note, though, that the bound table contains an extra entry
+ * for a list of unbound sockets and SOCK_DGRAM sockets will always remain in
+ * that list. The bound table is used solely for lookup of sockets when packets
+ * are received and that's not necessary for SOCK_DGRAM sockets since we create
+ * a datagram handle for each and need not perform a lookup.  Keeping SOCK_DGRAM
+ * sockets out of the bound hash buckets will reduce the chance of collisions
+ * when looking for SOCK_STREAM sockets and prevents us from having to check the
+ * socket type in the hash table lookups.
+ *
+ * - Sockets created by user action will either be "client" sockets that
+ * initiate a connection or "server" sockets that listen for connections; we do
+ * not support simultaneous connects (two "client" sockets connecting).
+ *
+ * - "Server" sockets are referred to as listener sockets throughout this
+ * implementation because they are in the SS_LISTEN state.  When a connection
+ * request is received (the second kind of socket mentioned above), we create a
+ * new socket and refer to it as a pending socket.  These pending sockets are
+ * placed on the pending connection list of the listener socket.  When future
+ * packets are received for the address the listener socket is bound to, we
+ * check if the source of the packet is from one that has an existing pending
+ * connection.  If it does, we process the packet for the pending socket.  When
+ * that socket reaches the connected state, it is removed from the listener
+ * socket's pending list and enqueued in the listener socket's accept queue.
+ * Callers of accept(2) will accept connected sockets from the listener socket's
+ * accept queue.  If the socket cannot be accepted for some reason then it is
+ * marked rejected.  Once the connection is accepted, it is owned by the user
+ * process and the responsibility for cleanup falls with that user process.
+ *
+ * - It is possible that these pending sockets will never reach the connected
+ * state; in fact, we may never receive another packet after the connection
+ * request.  Because of this, we must schedule a cleanup function to run in the
+ * future, after some amount of time passes where a connection should have been
+ * established.  This function ensures that the socket is off all lists so it
+ * cannot be retrieved, then drops all references to the socket so it is cleaned
+ * up (sock_put() -> sk_free() -> our sk_destruct implementation).  Note this
+ * function will also cleanup rejected sockets, those that reach the connected
+ * state but leave it before they have been accepted.
+ *
+ * - Sockets created by user action will be cleaned up when the user process
+ * calls close(2), causing our release implementation to be called. Our release
+ * implementation will perform some cleanup then drop the last reference so our
+ * sk_destruct implementation is invoked.  Our sk_destruct implementation will
+ * perform additional cleanup that's common for both types of sockets.
+ *
+ * - A socket's reference count is what ensures that the structure won't be
+ * freed.  Each entry in a list (such as the "global" bound and connected tables
+ * and the listener socket's pending list and connected queue) ensures a
+ * reference.  When we defer work until process context and pass a socket as our
+ * argument, we must ensure the reference count is increased to ensure the
+ * socket isn't freed before the function is run; the deferred function will
+ * then drop the reference.
+ */
+
+#include <linux/types.h>
+
+#define EXPORT_SYMTAB
+#include <linux/bitops.h>
+#include <linux/cred.h>
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/kernel.h>
+#include <linux/kmod.h>
+#include <linux/list.h>
+#include <linux/miscdevice.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/net.h>
+#include <linux/poll.h>
+#include <linux/skbuff.h>
+#include <linux/smp.h>
+#include <linux/socket.h>
+#include <linux/stddef.h>
+#include <linux/unistd.h>
+#include <linux/wait.h>
+#include <linux/workqueue.h>
+#include <net/sock.h>
+
+#include "af_vsock.h"
+#include "vsock_version.h"
+
+static int __vsock_bind(struct sock *sk, struct sockaddr_vm *addr);
+static void vsock_sk_destruct(struct sock *sk);
+static int vsock_queue_rcv_skb(struct sock *sk, struct sk_buff *skb);
+
+/* Protocol family. */
+static struct proto vsock_proto = {
+	.name = "AF_VSOCK",
+	.owner = THIS_MODULE,
+	.obj_size = sizeof(struct vsock_sock),
+};
+
+/* The default peer timeout indicates how long we will wait for a peer response
+ * to a control message.
+ */
+#define VSOCK_DEFAULT_CONNECT_TIMEOUT (2 * HZ)
+
+#define SS_LISTEN 255
+
+static const struct vsock_transport *transport;
+static DEFINE_MUTEX(vsock_register_mutex);
+
+/**** EXPORTS ****/
+
+/* Get the ID of the local context.  This is transport dependent. */
+
+int vm_sockets_get_local_cid(void)
+{
+	return transport->get_local_cid();
+}
+EXPORT_SYMBOL_GPL(vm_sockets_get_local_cid);
+
+/**** UTILS ****/
+
+/* Each bound VSocket is stored in the bind hash table and each connected
+ * VSocket is stored in the connected hash table.
+ *
+ * Unbound sockets are all put on the same list attached to the end of the hash
+ * table (vsock_unbound_sockets).  Bound sockets are added to the hash table in
+ * the bucket that their local address hashes to (vsock_bound_sockets(addr)
+ * represents the list that addr hashes to).
+ *
+ * Specifically, we initialize the vsock_bind_table array to a size of
+ * VSOCK_HASH_SIZE + 1 so that vsock_bind_table[0] through
+ * vsock_bind_table[VSOCK_HASH_SIZE - 1] are for bound sockets and
+ * vsock_bind_table[VSOCK_HASH_SIZE] is for unbound sockets.  The hash function
+ * mods with VSOCK_HASH_SIZE - 1 to ensure this.
+ */
+#define VSOCK_HASH_SIZE         251
+#define MAX_PORT_RETRIES        24
+
+#define VSOCK_HASH(addr)        ((addr)->svm_port % (VSOCK_HASH_SIZE - 1))
+#define vsock_bound_sockets(addr) (&vsock_bind_table[VSOCK_HASH(addr)])
+#define vsock_unbound_sockets     (&vsock_bind_table[VSOCK_HASH_SIZE])
+
+/* XXX This can probably be implemented in a better way. */
+#define VSOCK_CONN_HASH(src, dst)				\
+	(((src)->svm_cid ^ (dst)->svm_port) % (VSOCK_HASH_SIZE - 1))
+#define vsock_connected_sockets(src, dst)		\
+	(&vsock_connected_table[VSOCK_CONN_HASH(src, dst)])
+#define vsock_connected_sockets_vsk(vsk)				\
+	vsock_connected_sockets(&(vsk)->remote_addr, &(vsk)->local_addr)
+
+static struct list_head vsock_bind_table[VSOCK_HASH_SIZE + 1];
+static struct list_head vsock_connected_table[VSOCK_HASH_SIZE];
+static DEFINE_SPINLOCK(vsock_table_lock);
+
+static __init void vsock_init_tables(void)
+{
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(vsock_bind_table); i++)
+		INIT_LIST_HEAD(&vsock_bind_table[i]);
+
+	for (i = 0; i < ARRAY_SIZE(vsock_connected_table); i++)
+		INIT_LIST_HEAD(&vsock_connected_table[i]);
+}
+
+static void __vsock_insert_bound(struct list_head *list,
+				 struct vsock_sock *vsk)
+{
+	sock_hold(&vsk->sk);
+	list_add(&vsk->bound_table, list);
+}
+
+static void __vsock_insert_connected(struct list_head *list,
+				     struct vsock_sock *vsk)
+{
+	sock_hold(&vsk->sk);
+	list_add(&vsk->connected_table, list);
+}
+
+static void __vsock_remove_bound(struct vsock_sock *vsk)
+{
+	list_del_init(&vsk->bound_table);
+	sock_put(&vsk->sk);
+}
+
+static void __vsock_remove_connected(struct vsock_sock *vsk)
+{
+	list_del_init(&vsk->connected_table);
+	sock_put(&vsk->sk);
+}
+
+static struct sock *__vsock_find_bound_socket(struct sockaddr_vm *addr)
+{
+	struct vsock_sock *vsk;
+
+	list_for_each_entry(vsk, vsock_bound_sockets(addr), bound_table)
+		if (vsock_addr_equals_addr_any(addr, &vsk->local_addr))
+			return sk_vsock(vsk);
+
+	return NULL;
+}
+
+static struct sock *__vsock_find_connected_socket(struct sockaddr_vm *src,
+						  struct sockaddr_vm *dst)
+{
+	struct vsock_sock *vsk;
+
+	list_for_each_entry(vsk, vsock_connected_sockets(src, dst),
+			    connected_table) {
+		if (vsock_addr_equals_addr(src, &vsk->remote_addr)
+		    && vsock_addr_equals_addr(dst, &vsk->local_addr)) {
+			return sk_vsock(vsk);
+		}
+	}
+
+	return NULL;
+}
+
+static bool __vsock_in_bound_table(struct vsock_sock *vsk)
+{
+	return !list_empty(&vsk->bound_table);
+}
+
+static bool __vsock_in_connected_table(struct vsock_sock *vsk)
+{
+	return !list_empty(&vsk->connected_table);
+}
+
+static void vsock_insert_unbound(struct vsock_sock *vsk)
+{
+	spin_lock_bh(&vsock_table_lock);
+	__vsock_insert_bound(vsock_unbound_sockets, vsk);
+	spin_unlock_bh(&vsock_table_lock);
+}
+
+void vsock_insert_connected(struct vsock_sock *vsk)
+{
+	struct list_head *list = vsock_connected_sockets(
+		&vsk->remote_addr, &vsk->local_addr);
+
+	spin_lock_bh(&vsock_table_lock);
+	__vsock_insert_connected(list, vsk);
+	spin_unlock_bh(&vsock_table_lock);
+}
+EXPORT_SYMBOL_GPL(vsock_insert_connected);
+
+void vsock_remove_bound(struct vsock_sock *vsk)
+{
+	spin_lock_bh(&vsock_table_lock);
+	__vsock_remove_bound(vsk);
+	spin_unlock_bh(&vsock_table_lock);
+}
+EXPORT_SYMBOL_GPL(vsock_remove_bound);
+
+void vsock_remove_connected(struct vsock_sock *vsk)
+{
+	spin_lock_bh(&vsock_table_lock);
+	__vsock_remove_connected(vsk);
+	spin_unlock_bh(&vsock_table_lock);
+}
+EXPORT_SYMBOL_GPL(vsock_remove_connected);
+
+struct sock *vsock_find_bound_socket(struct sockaddr_vm *addr)
+{
+	struct sock *sk;
+
+	spin_lock_bh(&vsock_table_lock);
+	sk = __vsock_find_bound_socket(addr);
+	if (sk)
+		sock_hold(sk);
+
+	spin_unlock_bh(&vsock_table_lock);
+
+	return sk;
+}
+EXPORT_SYMBOL_GPL(vsock_find_bound_socket);
+
+struct sock *vsock_find_connected_socket(struct sockaddr_vm *src,
+					 struct sockaddr_vm *dst)
+{
+	struct sock *sk;
+
+	spin_lock_bh(&vsock_table_lock);
+	sk = __vsock_find_connected_socket(src, dst);
+	if (sk)
+		sock_hold(sk);
+
+	spin_unlock_bh(&vsock_table_lock);
+
+	return sk;
+}
+EXPORT_SYMBOL_GPL(vsock_find_connected_socket);
+
+static bool vsock_in_bound_table(struct vsock_sock *vsk)
+{
+	bool ret;
+
+	spin_lock_bh(&vsock_table_lock);
+	ret = __vsock_in_bound_table(vsk);
+	spin_unlock_bh(&vsock_table_lock);
+
+	return ret;
+}
+
+static bool vsock_in_connected_table(struct vsock_sock *vsk)
+{
+	bool ret;
+
+	spin_lock_bh(&vsock_table_lock);
+	ret = __vsock_in_connected_table(vsk);
+	spin_unlock_bh(&vsock_table_lock);
+
+	return ret;
+}
+
+void vsock_for_each_connected_socket(void (*fn)(struct sock *sk))
+{
+	int i;
+
+	spin_lock_bh(&vsock_table_lock);
+
+	for (i = 0; i < ARRAY_SIZE(vsock_connected_table); i++) {
+		struct vsock_sock *vsk;
+		list_for_each_entry(vsk, &vsock_connected_table[i],
+				    connected_table);
+			fn(sk_vsock(vsk));
+	}
+
+	spin_unlock_bh(&vsock_table_lock);
+}
+EXPORT_SYMBOL_GPL(vsock_for_each_connected_socket);
+
+void vsock_add_pending(struct sock *listener, struct sock *pending)
+{
+	struct vsock_sock *vlistener;
+	struct vsock_sock *vpending;
+
+	vlistener = vsock_sk(listener);
+	vpending = vsock_sk(pending);
+
+	sock_hold(pending);
+	sock_hold(listener);
+	list_add_tail(&vpending->pending_links, &vlistener->pending_links);
+}
+EXPORT_SYMBOL_GPL(vsock_add_pending);
+
+void vsock_remove_pending(struct sock *listener, struct sock *pending)
+{
+	struct vsock_sock *vpending = vsock_sk(pending);
+
+	list_del_init(&vpending->pending_links);
+	sock_put(listener);
+	sock_put(pending);
+}
+EXPORT_SYMBOL_GPL(vsock_remove_pending);
+
+void vsock_enqueue_accept(struct sock *listener, struct sock *connected)
+{
+	struct vsock_sock *vlistener;
+	struct vsock_sock *vconnected;
+
+	vlistener = vsock_sk(listener);
+	vconnected = vsock_sk(connected);
+
+	sock_hold(connected);
+	sock_hold(listener);
+	list_add_tail(&vconnected->accept_queue, &vlistener->accept_queue);
+}
+EXPORT_SYMBOL_GPL(vsock_enqueue_accept);
+
+static struct sock *vsock_dequeue_accept(struct sock *listener)
+{
+	struct vsock_sock *vlistener;
+	struct vsock_sock *vconnected;
+
+	vlistener = vsock_sk(listener);
+
+	if (list_empty(&vlistener->accept_queue))
+		return NULL;
+
+	vconnected = list_entry(vlistener->accept_queue.next,
+				struct vsock_sock, accept_queue);
+
+	list_del_init(&vconnected->accept_queue);
+	sock_put(listener);
+	/* The caller will need a reference on the connected socket so we let
+	 * it call sock_put().
+	 */
+
+	return sk_vsock(vconnected);
+}
+
+static bool vsock_is_accept_queue_empty(struct sock *sk)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+	return list_empty(&vsk->accept_queue);
+}
+
+static bool vsock_is_pending(struct sock *sk)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+	return !list_empty(&vsk->pending_links);
+}
+
+static int vsock_send_shutdown(struct sock *sk, int mode)
+{
+	return transport->shutdown(vsock_sk(sk), mode);
+}
+
+void vsock_pending_work(struct work_struct *work)
+{
+	struct sock *sk;
+	struct sock *listener;
+	struct vsock_sock *vsk;
+	bool cleanup;
+
+	vsk = container_of(work, struct vsock_sock, dwork.work);
+	sk = sk_vsock(vsk);
+	listener = vsk->listener;
+	cleanup = true;
+
+	lock_sock(listener);
+	lock_sock(sk);
+
+	if (vsock_is_pending(sk)) {
+		vsock_remove_pending(listener, sk);
+	} else if (!vsk->rejected) {
+		/* We are not on the pending list and accept() did not reject
+		 * us, so we must have been accepted by our user process.  We
+		 * just need to drop our references to the sockets and be on
+		 * our way.
+		 */
+		cleanup = false;
+		goto out;
+	}
+
+	listener->sk_ack_backlog--;
+
+	/* We need to remove ourself from the global connected sockets list so
+	 * incoming packets can't find this socket, and to reduce the reference
+	 * count.
+	 */
+	if (vsock_in_connected_table(vsk))
+		vsock_remove_connected(vsk);
+
+	sk->sk_state = SS_FREE;
+
+out:
+	release_sock(sk);
+	release_sock(listener);
+	if (cleanup)
+		sock_put(sk);
+
+	sock_put(sk);
+	sock_put(listener);
+}
+EXPORT_SYMBOL_GPL(vsock_pending_work);
+
+/**** SOCKET OPERATIONS ****/
+
+static int __vsock_bind_stream(struct vsock_sock *vsk,
+			       struct sockaddr_vm *addr)
+{
+	static u32 port = LAST_RESERVED_PORT + 1;
+	struct sockaddr_vm new_addr;
+
+	vsock_addr_init(&new_addr, addr->svm_cid, addr->svm_port);
+
+	if (addr->svm_port == VMADDR_PORT_ANY) {
+		bool found = false;
+		unsigned int i;
+
+		for (i = 0; i < MAX_PORT_RETRIES; i++) {
+			if (port <= LAST_RESERVED_PORT)
+				port = LAST_RESERVED_PORT + 1;
+
+			new_addr.svm_port = port++;
+
+			if (!__vsock_find_bound_socket(&new_addr)) {
+				found = true;
+				break;
+			}
+		}
+
+		if (!found)
+			return -EADDRNOTAVAIL;
+	} else {
+		/* If port is in reserved range, ensure caller
+		 * has necessary privileges.
+		 */
+		if (addr->svm_port <= LAST_RESERVED_PORT &&
+		    !capable(CAP_NET_BIND_SERVICE)) {
+			return -EACCES;
+		}
+
+		if (__vsock_find_bound_socket(&new_addr))
+			return -EADDRINUSE;
+	}
+
+	vsock_addr_init(&vsk->local_addr, new_addr.svm_cid, new_addr.svm_port);
+
+	/* Remove stream sockets from the unbound list and add them to the hash
+	 * table for easy lookup by its address.  The unbound list is simply an
+	 * extra entry at the end of the hash table, a trick used by AF_UNIX.
+	 */
+	__vsock_remove_bound(vsk);
+	__vsock_insert_bound(vsock_bound_sockets(&vsk->local_addr), vsk);
+
+	return 0;
+}
+
+static int __vsock_bind_dgram(struct vsock_sock *vsk,
+			      struct sockaddr_vm *addr)
+{
+	return transport->dgram_bind(vsk, addr);
+}
+
+static int __vsock_bind(struct sock *sk, struct sockaddr_vm *addr)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+	u32 cid;
+	int retval;
+
+	/* First ensure this socket isn't already bound. */
+	if (vsock_addr_bound(&vsk->local_addr))
+		return -EINVAL;
+
+	/* Now bind to the provided address or select appropriate values if
+	 * none are provided (VMADDR_CID_ANY and VMADDR_PORT_ANY).  Note that
+	 * like AF_INET prevents binding to a non-local IP address (in most
+	 * cases), we only allow binding to the local CID.
+	 */
+	cid = transport->get_local_cid();
+	if (addr->svm_cid != cid && addr->svm_cid != VMADDR_CID_ANY)
+		return -EADDRNOTAVAIL;
+
+	switch (sk->sk_socket->type) {
+	case SOCK_STREAM:
+		spin_lock_bh(&vsock_table_lock);
+		retval = __vsock_bind_stream(vsk, addr);
+		spin_unlock_bh(&vsock_table_lock);
+		break;
+
+	case SOCK_DGRAM:
+		retval = __vsock_bind_dgram(vsk, addr);
+		break;
+
+	default:
+		retval = -EINVAL;
+		break;
+	}
+
+	return retval;
+}
+
+struct sock *__vsock_create(struct net *net,
+			    struct socket *sock,
+			    struct sock *parent,
+			    gfp_t priority,
+			    unsigned short type)
+{
+	struct sock *sk;
+	struct vsock_sock *psk;
+	struct vsock_sock *vsk;
+
+	sk = sk_alloc(net, AF_VSOCK, priority, &vsock_proto);
+	if (!sk)
+		return NULL;
+
+	sock_init_data(sock, sk);
+
+	/* sk->sk_type is normally set in sock_init_data, but only if sock is
+	 * non-NULL. We make sure that our sockets always have a type by
+	 * setting it here if needed.
+	 */
+	if (!sock)
+		sk->sk_type = type;
+
+	vsk = vsock_sk(sk);
+	vsock_addr_init(&vsk->local_addr, VMADDR_CID_ANY, VMADDR_PORT_ANY);
+	vsock_addr_init(&vsk->remote_addr, VMADDR_CID_ANY, VMADDR_PORT_ANY);
+
+	sk->sk_destruct = vsock_sk_destruct;
+	sk->sk_backlog_rcv = vsock_queue_rcv_skb;
+	sk->sk_state = 0;
+	sock_reset_flag(sk, SOCK_DONE);
+
+	INIT_LIST_HEAD(&vsk->bound_table);
+	INIT_LIST_HEAD(&vsk->connected_table);
+	vsk->listener = NULL;
+	INIT_LIST_HEAD(&vsk->pending_links);
+	INIT_LIST_HEAD(&vsk->accept_queue);
+	vsk->rejected = false;
+	vsk->sent_request = false;
+	vsk->ignore_connecting_rst = false;
+	vsk->peer_shutdown = 0;
+
+	psk = parent ? vsock_sk(parent) : NULL;
+	if (parent) {
+		vsk->trusted = psk->trusted;
+		vsk->owner = get_cred(psk->owner);
+		vsk->connect_timeout = psk->connect_timeout;
+	} else {
+		vsk->trusted = capable(CAP_NET_ADMIN);
+		vsk->owner = get_current_cred();
+		vsk->connect_timeout = VSOCK_DEFAULT_CONNECT_TIMEOUT;
+	}
+
+	if (transport->init(vsk, psk) < 0) {
+		sk_free(sk);
+		return NULL;
+	}
+
+	if (sock)
+		vsock_insert_unbound(vsk);
+
+	return sk;
+}
+EXPORT_SYMBOL_GPL(__vsock_create);
+
+static void __vsock_release(struct sock *sk)
+{
+	if (sk) {
+		struct sk_buff *skb;
+		struct sock *pending;
+		struct vsock_sock *vsk;
+
+		vsk = vsock_sk(sk);
+		pending = NULL;	/* Compiler warning. */
+
+		if (vsock_in_bound_table(vsk))
+			vsock_remove_bound(vsk);
+
+		if (vsock_in_connected_table(vsk))
+			vsock_remove_connected(vsk);
+
+		transport->release(vsk);
+
+		lock_sock(sk);
+		sock_orphan(sk);
+		sk->sk_shutdown = SHUTDOWN_MASK;
+
+		while ((skb = skb_dequeue(&sk->sk_receive_queue)))
+			kfree_skb(skb);
+
+		/* Clean up any sockets that never were accepted. */
+		while ((pending = vsock_dequeue_accept(sk)) != NULL) {
+			__vsock_release(pending);
+			sock_put(pending);
+		}
+
+		release_sock(sk);
+		sock_put(sk);
+	}
+}
+
+static void vsock_sk_destruct(struct sock *sk)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	transport->destruct(vsk);
+
+	/* When clearing these addresses, there's no need to set the family and
+	 * possibly register the address family with the kernel.
+	 */
+	vsock_addr_init(&vsk->local_addr, VMADDR_CID_ANY, VMADDR_PORT_ANY);
+	vsock_addr_init(&vsk->remote_addr, VMADDR_CID_ANY, VMADDR_PORT_ANY);
+
+	put_cred(vsk->owner);
+}
+
+static int vsock_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
+{
+	int err;
+
+	err = sock_queue_rcv_skb(sk, skb);
+	if (err)
+		kfree_skb(skb);
+
+	return err;
+}
+
+s64 vsock_stream_has_data(struct vsock_sock *vsk)
+{
+	return transport->stream_has_data(vsk);
+}
+EXPORT_SYMBOL_GPL(vsock_stream_has_data);
+
+s64 vsock_stream_has_space(struct vsock_sock *vsk)
+{
+	return transport->stream_has_space(vsk);
+}
+EXPORT_SYMBOL_GPL(vsock_stream_has_space);
+
+static int vsock_release(struct socket *sock)
+{
+	__vsock_release(sock->sk);
+	sock->sk = NULL;
+	sock->state = SS_FREE;
+
+	return 0;
+}
+
+static int
+vsock_bind(struct socket *sock, struct sockaddr *addr, int addr_len)
+{
+	int err;
+	struct sock *sk;
+	struct sockaddr_vm *vm_addr;
+
+	sk = sock->sk;
+
+	if (vsock_addr_cast(addr, addr_len, &vm_addr) != 0)
+		return -EINVAL;
+
+	lock_sock(sk);
+	err = __vsock_bind(sk, vm_addr);
+	release_sock(sk);
+
+	return err;
+}
+
+static int vsock_getname(struct socket *sock,
+			 struct sockaddr *addr, int *addr_len, int peer)
+{
+	int err;
+	struct sock *sk;
+	struct vsock_sock *vsk;
+	struct sockaddr_vm *vm_addr;
+
+	sk = sock->sk;
+	vsk = vsock_sk(sk);
+	err = 0;
+
+	lock_sock(sk);
+
+	if (peer) {
+		if (sock->state != SS_CONNECTED) {
+			err = -ENOTCONN;
+			goto out;
+		}
+		vm_addr = &vsk->remote_addr;
+	} else {
+		vm_addr = &vsk->local_addr;
+	}
+
+	if (!vm_addr) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	/* sys_getsockname() and sys_getpeername() pass us a
+	 * MAX_SOCK_ADDR-sized buffer and don't set addr_len.  Unfortunately
+	 * that macro is defined in socket.c instead of .h, so we hardcode its
+	 * value here.
+	 */
+	BUILD_BUG_ON(sizeof(*vm_addr) > 128);
+	memcpy(addr, vm_addr, sizeof(*vm_addr));
+	*addr_len = sizeof(*vm_addr);
+
+out:
+	release_sock(sk);
+	return err;
+}
+
+static int vsock_shutdown(struct socket *sock, int mode)
+{
+	int err;
+	struct sock *sk;
+
+	/* User level uses SHUT_RD (0) and SHUT_WR (1), but the kernel uses
+	 * RCV_SHUTDOWN (1) and SEND_SHUTDOWN (2), so we must increment mode
+	 * here like the other address families do.  Note also that the
+	 * increment makes SHUT_RDWR (2) into RCV_SHUTDOWN | SEND_SHUTDOWN (3),
+	 * which is what we want.
+	 */
+	mode++;
+
+	if ((mode & ~SHUTDOWN_MASK) || !mode)
+		return -EINVAL;
+
+	/* If this is a STREAM socket and it is not connected then bail out
+	 * immediately.  If it is a DGRAM socket then we must first kick the
+	 * socket so that it wakes up from any sleeping calls, for example
+	 * recv(), and then afterwards return the error.
+	 */
+
+	sk = sock->sk;
+	if (sock->state == SS_UNCONNECTED) {
+		err = -ENOTCONN;
+		if (sk->sk_type == SOCK_STREAM)
+			return err;
+	} else {
+		sock->state = SS_DISCONNECTING;
+		err = 0;
+	}
+
+	/* Receive and send shutdowns are treated alike. */
+	mode = mode & (RCV_SHUTDOWN | SEND_SHUTDOWN);
+	if (mode) {
+		lock_sock(sk);
+		sk->sk_shutdown |= mode;
+		sk->sk_state_change(sk);
+		release_sock(sk);
+
+		if (sk->sk_type == SOCK_STREAM) {
+			sock_reset_flag(sk, SOCK_DONE);
+			vsock_send_shutdown(sk, mode);
+		}
+	}
+
+	return err;
+}
+
+static unsigned int vsock_poll(struct file *file, struct socket *sock,
+			       poll_table *wait)
+{
+	struct sock *sk;
+	unsigned int mask;
+	struct vsock_sock *vsk;
+
+	sk = sock->sk;
+	vsk = vsock_sk(sk);
+
+	poll_wait(file, sk_sleep(sk), wait);
+	mask = 0;
+
+	if (sk->sk_err)
+		/* Signify that there has been an error on this socket. */
+		mask |= POLLERR;
+
+	/* INET sockets treat local write shutdown and peer write shutdown as a
+	 * case of POLLHUP set.
+	 */
+	if ((sk->sk_shutdown == SHUTDOWN_MASK) ||
+	    ((sk->sk_shutdown & SEND_SHUTDOWN) &&
+	     (vsk->peer_shutdown & SEND_SHUTDOWN))) {
+		mask |= POLLHUP;
+	}
+
+	if (sk->sk_shutdown & RCV_SHUTDOWN ||
+	    vsk->peer_shutdown & SEND_SHUTDOWN) {
+		mask |= POLLRDHUP;
+	}
+
+	if (sock->type == SOCK_DGRAM) {
+		/* For datagram sockets we can read if there is something in
+		 * the queue and write as long as the socket isn't shutdown for
+		 * sending.
+		 */
+		if (!skb_queue_empty(&sk->sk_receive_queue) ||
+		    (sk->sk_shutdown & RCV_SHUTDOWN)) {
+			mask |= POLLIN | POLLRDNORM;
+		}
+
+		if (!(sk->sk_shutdown & SEND_SHUTDOWN))
+			mask |= POLLOUT | POLLWRNORM | POLLWRBAND;
+
+	} else if (sock->type == SOCK_STREAM) {
+		lock_sock(sk);
+
+		/* Listening sockets that have connections in their accept
+		 * queue can be read.
+		 */
+		if (sk->sk_state == SS_LISTEN
+		    && !vsock_is_accept_queue_empty(sk))
+			mask |= POLLIN | POLLRDNORM;
+
+		/* If there is something in the queue then we can read. */
+		if (transport->stream_is_active(vsk) &&
+		    !(sk->sk_shutdown & RCV_SHUTDOWN)) {
+			bool data_ready_now = false;
+			int ret = transport->notify_poll_in(
+					vsk, 1, &data_ready_now);
+			if (ret < 0) {
+				mask |= POLLERR;
+			} else {
+				if (data_ready_now)
+					mask |= POLLIN | POLLRDNORM;
+
+			}
+		}
+
+		/* Sockets whose connections have been closed, reset, or
+		 * terminated should also be considered read, and we check the
+		 * shutdown flag for that.
+		 */
+		if (sk->sk_shutdown & RCV_SHUTDOWN ||
+		    vsk->peer_shutdown & SEND_SHUTDOWN) {
+			mask |= POLLIN | POLLRDNORM;
+		}
+
+		/* Connected sockets that can produce data can be written. */
+		if (sk->sk_state == SS_CONNECTED) {
+			if (!(sk->sk_shutdown & SEND_SHUTDOWN)) {
+				bool space_avail_now = false;
+				int ret = transport->notify_poll_out(
+						vsk, 1, &space_avail_now);
+				if (ret < 0) {
+					mask |= POLLERR;
+				} else {
+					if (space_avail_now)
+						/* Remove POLLWRBAND since INET
+						 * sockets are not setting it.
+						 */
+						mask |= POLLOUT | POLLWRNORM;
+
+				}
+			}
+		}
+
+		/* Simulate INET socket poll behaviors, which sets
+		 * POLLOUT|POLLWRNORM when peer is closed and nothing to read,
+		 * but local send is not shutdown.
+		 */
+		if (sk->sk_state == SS_UNCONNECTED) {
+			if (!(sk->sk_shutdown & SEND_SHUTDOWN))
+				mask |= POLLOUT | POLLWRNORM;
+
+		}
+
+		release_sock(sk);
+	}
+
+	return mask;
+}
+
+static int vsock_dgram_sendmsg(struct kiocb *kiocb, struct socket *sock,
+			       struct msghdr *msg, size_t len)
+{
+	int err;
+	struct sock *sk;
+	struct vsock_sock *vsk;
+	struct sockaddr_vm *remote_addr;
+
+	if (msg->msg_flags & MSG_OOB)
+		return -EOPNOTSUPP;
+
+	/* For now, MSG_DONTWAIT is always assumed... */
+	err = 0;
+	sk = sock->sk;
+	vsk = vsock_sk(sk);
+
+	lock_sock(sk);
+
+	if (!vsock_addr_bound(&vsk->local_addr)) {
+		struct sockaddr_vm local_addr;
+
+		vsock_addr_init(&local_addr, VMADDR_CID_ANY, VMADDR_PORT_ANY);
+		err = __vsock_bind(sk, &local_addr);
+		if (err != 0)
+			goto out;
+
+	}
+
+	/* If the provided message contains an address, use that.  Otherwise
+	 * fall back on the socket's remote handle (if it has been connected).
+	 */
+	if (msg->msg_name &&
+	    vsock_addr_cast(msg->msg_name, msg->msg_namelen,
+			    &remote_addr) == 0) {
+		/* Ensure this address is of the right type and is a valid
+		 * destination.
+		 */
+
+		if (remote_addr->svm_cid == VMADDR_CID_ANY)
+			remote_addr->svm_cid = transport->get_local_cid();
+
+		if (!vsock_addr_bound(remote_addr)) {
+			err = -EINVAL;
+			goto out;
+		}
+	} else if (sock->state == SS_CONNECTED) {
+		remote_addr = &vsk->remote_addr;
+
+		if (remote_addr->svm_cid == VMADDR_CID_ANY)
+			remote_addr->svm_cid = transport->get_local_cid();
+
+		/* XXX Should connect() or this function ensure remote_addr is
+		 * bound?
+		 */
+		if (!vsock_addr_bound(&vsk->remote_addr)) {
+			err = -EINVAL;
+			goto out;
+		}
+	} else {
+		err = -EINVAL;
+		goto out;
+	}
+
+	if (!transport->dgram_allow(remote_addr->svm_cid,
+				    remote_addr->svm_port)) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	err = transport->dgram_enqueue(vsk, remote_addr, msg->msg_iov, len);
+
+out:
+	release_sock(sk);
+	return err;
+}
+
+static int vsock_dgram_connect(struct socket *sock,
+			       struct sockaddr *addr, int addr_len, int flags)
+{
+	int err;
+	struct sock *sk;
+	struct vsock_sock *vsk;
+	struct sockaddr_vm *remote_addr;
+
+	sk = sock->sk;
+	vsk = vsock_sk(sk);
+
+	err = vsock_addr_cast(addr, addr_len, &remote_addr);
+	if (err == -EAFNOSUPPORT && remote_addr->svm_family == AF_UNSPEC) {
+		lock_sock(sk);
+		vsock_addr_init(&vsk->remote_addr, VMADDR_CID_ANY,
+				VMADDR_PORT_ANY);
+		sock->state = SS_UNCONNECTED;
+		release_sock(sk);
+		return 0;
+	} else if (err != 0)
+		return -EINVAL;
+
+	lock_sock(sk);
+
+	if (!vsock_addr_bound(&vsk->local_addr)) {
+		struct sockaddr_vm local_addr;
+
+		vsock_addr_init(&local_addr, VMADDR_CID_ANY, VMADDR_PORT_ANY);
+		err = __vsock_bind(sk, &local_addr);
+		if (err != 0)
+			goto out;
+
+	}
+
+	if (!transport->dgram_allow(remote_addr->svm_cid,
+				    remote_addr->svm_port)) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	memcpy(&vsk->remote_addr, remote_addr, sizeof(vsk->remote_addr));
+	sock->state = SS_CONNECTED;
+
+out:
+	release_sock(sk);
+	return err;
+}
+
+static int vsock_dgram_recvmsg(struct kiocb *kiocb, struct socket *sock,
+			       struct msghdr *msg, size_t len, int flags)
+{
+	return transport->dgram_dequeue(kiocb, vsock_sk(sock->sk), msg, len,
+					flags);
+}
+
+static const struct proto_ops vsock_dgram_ops = {
+	.family = PF_VSOCK,
+	.owner = THIS_MODULE,
+	.release = vsock_release,
+	.bind = vsock_bind,
+	.connect = vsock_dgram_connect,
+	.socketpair = sock_no_socketpair,
+	.accept = sock_no_accept,
+	.getname = vsock_getname,
+	.poll = vsock_poll,
+	.ioctl = sock_no_ioctl,
+	.listen = sock_no_listen,
+	.shutdown = vsock_shutdown,
+	.setsockopt = sock_no_setsockopt,
+	.getsockopt = sock_no_getsockopt,
+	.sendmsg = vsock_dgram_sendmsg,
+	.recvmsg = vsock_dgram_recvmsg,
+	.mmap = sock_no_mmap,
+	.sendpage = sock_no_sendpage,
+};
+
+static void vsock_connect_timeout(struct work_struct *work)
+{
+	struct sock *sk;
+	struct vsock_sock *vsk;
+
+	vsk = container_of(work, struct vsock_sock, dwork.work);
+	sk = sk_vsock(vsk);
+
+	lock_sock(sk);
+	if (sk->sk_state == SS_CONNECTING &&
+	    (sk->sk_shutdown != SHUTDOWN_MASK)) {
+		sk->sk_state = SS_UNCONNECTED;
+		sk->sk_err = ETIMEDOUT;
+		sk->sk_error_report(sk);
+	}
+	release_sock(sk);
+
+	sock_put(sk);
+}
+
+static int vsock_stream_connect(struct socket *sock, struct sockaddr *addr,
+				int addr_len, int flags)
+{
+	int err;
+	struct sock *sk;
+	struct vsock_sock *vsk;
+	struct sockaddr_vm *remote_addr;
+	long timeout;
+	DEFINE_WAIT(wait);
+
+	err = 0;
+	sk = sock->sk;
+	vsk = vsock_sk(sk);
+
+	lock_sock(sk);
+
+	/* XXX AF_UNSPEC should make us disconnect like AF_INET. */
+	switch (sock->state) {
+	case SS_CONNECTED:
+		err = -EISCONN;
+		goto out;
+	case SS_DISCONNECTING:
+		err = -EINVAL;
+		goto out;
+	case SS_CONNECTING:
+		/* This continues on so we can move sock into the SS_CONNECTED
+		 * state once the connection has completed (at which point err
+		 * will be set to zero also).  Otherwise, we will either wait
+		 * for the connection or return -EALREADY should this be a
+		 * non-blocking call.
+		 */
+		err = -EALREADY;
+		break;
+	default:
+		if ((sk->sk_state == SS_LISTEN) ||
+		    vsock_addr_cast(addr, addr_len, &remote_addr) != 0) {
+			err = -EINVAL;
+			goto out;
+		}
+
+		/* The hypervisor and well-known contexts do not have socket
+		 * endpoints.
+		 */
+		if (!transport->stream_allow(remote_addr->svm_cid,
+					     remote_addr->svm_port)) {
+			err = -ENETUNREACH;
+			goto out;
+		}
+
+		/* Set the remote address that we are connecting to. */
+		memcpy(&vsk->remote_addr, remote_addr,
+		       sizeof(vsk->remote_addr));
+
+		/* Autobind this socket to the local address if necessary. */
+		if (!vsock_addr_bound(&vsk->local_addr)) {
+			struct sockaddr_vm local_addr;
+
+			vsock_addr_init(&local_addr, VMADDR_CID_ANY,
+					VMADDR_PORT_ANY);
+			err = __vsock_bind(sk, &local_addr);
+			if (err != 0)
+				goto out;
+
+		}
+
+		sk->sk_state = SS_CONNECTING;
+
+		err = transport->connect(vsk);
+		if (err < 0)
+			goto out;
+
+		/* Mark sock as connecting and set the error code to in
+		 * progress in case this is a non-blocking connect.
+		 */
+		sock->state = SS_CONNECTING;
+		err = -EINPROGRESS;
+	}
+
+	/* The receive path will handle all communication until we are able to
+	 * enter the connected state.  Here we wait for the connection to be
+	 * completed or a notification of an error.
+	 */
+	timeout = vsk->connect_timeout;
+	prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
+
+	while (sk->sk_state != SS_CONNECTED && sk->sk_err == 0) {
+		if (flags & O_NONBLOCK) {
+			/* If we're not going to block, we schedule a timeout
+			 * function to generate a timeout on the connection
+			 * attempt, in case the peer doesn't respond in a
+			 * timely manner. We hold on to the socket until the
+			 * timeout fires.
+			 */
+			sock_hold(sk);
+			INIT_DELAYED_WORK(&vsk->dwork,
+					  vsock_connect_timeout);
+			schedule_delayed_work(&vsk->dwork, timeout);
+
+			/* Skip ahead to preserve error code set above. */
+			goto out_wait;
+		}
+
+		release_sock(sk);
+		timeout = schedule_timeout(timeout);
+		lock_sock(sk);
+
+		if (signal_pending(current)) {
+			err = sock_intr_errno(timeout);
+			goto out_wait_error;
+		} else if (timeout == 0) {
+			err = -ETIMEDOUT;
+			goto out_wait_error;
+		}
+
+		prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
+	}
+
+	if (sk->sk_err) {
+		err = -sk->sk_err;
+		goto out_wait_error;
+	} else
+		err = 0;
+
+out_wait:
+	finish_wait(sk_sleep(sk), &wait);
+out:
+	release_sock(sk);
+	return err;
+
+out_wait_error:
+	sk->sk_state = SS_UNCONNECTED;
+	sock->state = SS_UNCONNECTED;
+	goto out_wait;
+}
+
+static int vsock_accept(struct socket *sock, struct socket *newsock, int flags)
+{
+	struct sock *listener;
+	int err;
+	struct sock *connected;
+	struct vsock_sock *vconnected;
+	long timeout;
+	DEFINE_WAIT(wait);
+
+	err = 0;
+	listener = sock->sk;
+
+	lock_sock(listener);
+
+	if (sock->type != SOCK_STREAM) {
+		err = -EOPNOTSUPP;
+		goto out;
+	}
+
+	if (listener->sk_state != SS_LISTEN) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	/* Wait for children sockets to appear; these are the new sockets
+	 * created upon connection establishment.
+	 */
+	timeout = sock_sndtimeo(listener, flags & O_NONBLOCK);
+	prepare_to_wait(sk_sleep(listener), &wait, TASK_INTERRUPTIBLE);
+
+	while ((connected = vsock_dequeue_accept(listener)) == NULL &&
+	       listener->sk_err == 0) {
+		release_sock(listener);
+		timeout = schedule_timeout(timeout);
+		lock_sock(listener);
+
+		if (signal_pending(current)) {
+			err = sock_intr_errno(timeout);
+			goto out_wait;
+		} else if (timeout == 0) {
+			err = -EAGAIN;
+			goto out_wait;
+		}
+
+		prepare_to_wait(sk_sleep(listener), &wait, TASK_INTERRUPTIBLE);
+	}
+
+	if (listener->sk_err)
+		err = -listener->sk_err;
+
+	if (connected) {
+		listener->sk_ack_backlog--;
+
+		lock_sock(connected);
+		vconnected = vsock_sk(connected);
+
+		/* If the listener socket has received an error, then we should
+		 * reject this socket and return.  Note that we simply mark the
+		 * socket rejected, drop our reference, and let the cleanup
+		 * function handle the cleanup; the fact that we found it in
+		 * the listener's accept queue guarantees that the cleanup
+		 * function hasn't run yet.
+		 */
+		if (err) {
+			vconnected->rejected = true;
+			release_sock(connected);
+			sock_put(connected);
+			goto out_wait;
+		}
+
+		newsock->state = SS_CONNECTED;
+		sock_graft(connected, newsock);
+		release_sock(connected);
+		sock_put(connected);
+	}
+
+out_wait:
+	finish_wait(sk_sleep(listener), &wait);
+out:
+	release_sock(listener);
+	return err;
+}
+
+static int vsock_listen(struct socket *sock, int backlog)
+{
+	int err;
+	struct sock *sk;
+	struct vsock_sock *vsk;
+
+	sk = sock->sk;
+
+	lock_sock(sk);
+
+	if (sock->type != SOCK_STREAM) {
+		err = -EOPNOTSUPP;
+		goto out;
+	}
+
+	if (sock->state != SS_UNCONNECTED) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	vsk = vsock_sk(sk);
+
+	if (!vsock_addr_bound(&vsk->local_addr)) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	sk->sk_max_ack_backlog = backlog;
+	sk->sk_state = SS_LISTEN;
+
+	err = 0;
+
+out:
+	release_sock(sk);
+	return err;
+}
+
+static int vsock_stream_setsockopt(struct socket *sock,
+				   int level,
+				   int optname,
+				   char __user *optval,
+				   unsigned int optlen)
+{
+	int err;
+	struct sock *sk;
+	struct vsock_sock *vsk;
+	u64 val;
+
+	if (level != AF_VSOCK)
+		return -ENOPROTOOPT;
+
+#define COPY_IN(_v)                                       \
+	do {						  \
+		if (optlen < sizeof(_v)) {		  \
+			err = -EINVAL;			  \
+			goto exit;			  \
+		}					  \
+		if (copy_from_user(&_v, optval, sizeof(_v)) != 0) {	\
+			err = -EFAULT;					\
+			goto exit;					\
+		}							\
+	} while (0)
+
+	err = 0;
+	sk = sock->sk;
+	vsk = vsock_sk(sk);
+
+	lock_sock(sk);
+
+	switch (optname) {
+	case SO_VM_SOCKETS_BUFFER_SIZE:
+		COPY_IN(val);
+		transport->set_buffer_size(vsk, val);
+		break;
+
+	case SO_VM_SOCKETS_BUFFER_MAX_SIZE:
+		COPY_IN(val);
+		transport->set_max_buffer_size(vsk, val);
+		break;
+
+	case SO_VM_SOCKETS_BUFFER_MIN_SIZE:
+		COPY_IN(val);
+		transport->set_min_buffer_size(vsk, val);
+		break;
+
+	case SO_VM_SOCKETS_CONNECT_TIMEOUT: {
+		struct timeval tv;
+		COPY_IN(tv);
+		if (tv.tv_sec >= 0 && tv.tv_usec < USEC_PER_SEC &&
+		    tv.tv_sec < (MAX_SCHEDULE_TIMEOUT / HZ - 1)) {
+			vsk->connect_timeout = tv.tv_sec * HZ +
+			    DIV_ROUND_UP(tv.tv_usec, (1000000 / HZ));
+			if (vsk->connect_timeout == 0)
+				vsk->connect_timeout =
+				    VSOCK_DEFAULT_CONNECT_TIMEOUT;
+
+		} else {
+			err = -ERANGE;
+		}
+		break;
+	}
+
+	default:
+		err = -ENOPROTOOPT;
+		break;
+	}
+
+#undef COPY_IN
+
+exit:
+	release_sock(sk);
+	return err;
+}
+
+static int vsock_stream_getsockopt(struct socket *sock,
+				   int level, int optname,
+				   char __user *optval,
+				   int __user *optlen)
+{
+	int err;
+	int len;
+	struct sock *sk;
+	struct vsock_sock *vsk;
+	u64 val;
+
+	if (level != AF_VSOCK)
+		return -ENOPROTOOPT;
+
+	err = get_user(len, optlen);
+	if (err != 0)
+		return err;
+
+#define COPY_OUT(_v)                            \
+	do {					\
+		if (len < sizeof(_v))		\
+			return -EINVAL;		\
+						\
+		len = sizeof(_v);		\
+		if (copy_to_user(optval, &_v, len) != 0)	\
+			return -EFAULT;				\
+								\
+	} while (0)
+
+	err = 0;
+	sk = sock->sk;
+	vsk = vsock_sk(sk);
+
+	switch (optname) {
+	case SO_VM_SOCKETS_BUFFER_SIZE:
+		val = transport->get_buffer_size(vsk);
+		COPY_OUT(val);
+		break;
+
+	case SO_VM_SOCKETS_BUFFER_MAX_SIZE:
+		val = transport->get_max_buffer_size(vsk);
+		COPY_OUT(val);
+		break;
+
+	case SO_VM_SOCKETS_BUFFER_MIN_SIZE:
+		val = transport->get_min_buffer_size(vsk);
+		COPY_OUT(val);
+		break;
+
+	case SO_VM_SOCKETS_CONNECT_TIMEOUT: {
+		struct timeval tv;
+		tv.tv_sec = vsk->connect_timeout / HZ;
+		tv.tv_usec =
+		    (vsk->connect_timeout -
+		     tv.tv_sec * HZ) * (1000000 / HZ);
+		COPY_OUT(tv);
+		break;
+	}
+	default:
+		return -ENOPROTOOPT;
+	}
+
+	err = put_user(len, optlen);
+	if (err != 0)
+		return -EFAULT;
+
+#undef COPY_OUT
+
+	return 0;
+}
+
+static int vsock_stream_sendmsg(struct kiocb *kiocb, struct socket *sock,
+				struct msghdr *msg, size_t len)
+{
+	struct sock *sk;
+	struct vsock_sock *vsk;
+	ssize_t total_written;
+	long timeout;
+	int err;
+	struct vsock_transport_send_notify_data send_data;
+
+	DEFINE_WAIT(wait);
+
+	sk = sock->sk;
+	vsk = vsock_sk(sk);
+	total_written = 0;
+	err = 0;
+
+	if (msg->msg_flags & MSG_OOB)
+		return -EOPNOTSUPP;
+
+	lock_sock(sk);
+
+	/* Callers should not provide a destination with stream sockets. */
+	if (msg->msg_namelen) {
+		err = sk->sk_state == SS_CONNECTED ? -EISCONN : -EOPNOTSUPP;
+		goto out;
+	}
+
+	/* Send data only if both sides are not shutdown in the direction. */
+	if (sk->sk_shutdown & SEND_SHUTDOWN ||
+	    vsk->peer_shutdown & RCV_SHUTDOWN) {
+		err = -EPIPE;
+		goto out;
+	}
+
+	if (sk->sk_state != SS_CONNECTED ||
+	    !vsock_addr_bound(&vsk->local_addr)) {
+		err = -ENOTCONN;
+		goto out;
+	}
+
+	if (!vsock_addr_bound(&vsk->remote_addr)) {
+		err = -EDESTADDRREQ;
+		goto out;
+	}
+
+	/* Wait for room in the produce queue to enqueue our user's data. */
+	timeout = sock_sndtimeo(sk, msg->msg_flags & MSG_DONTWAIT);
+
+	err = transport->notify_send_init(vsk, &send_data);
+	if (err < 0)
+		goto out;
+
+	prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
+
+	while (total_written < len) {
+		ssize_t written;
+
+		while (vsock_stream_has_space(vsk) == 0 &&
+		       sk->sk_err == 0 &&
+		       !(sk->sk_shutdown & SEND_SHUTDOWN) &&
+		       !(vsk->peer_shutdown & RCV_SHUTDOWN)) {
+
+			/* Don't wait for non-blocking sockets. */
+			if (timeout == 0) {
+				err = -EAGAIN;
+				goto out_wait;
+			}
+
+			err = transport->notify_send_pre_block(vsk, &send_data);
+			if (err < 0)
+				goto out_wait;
+
+			release_sock(sk);
+			timeout = schedule_timeout(timeout);
+			lock_sock(sk);
+			if (signal_pending(current)) {
+				err = sock_intr_errno(timeout);
+				goto out_wait;
+			} else if (timeout == 0) {
+				err = -EAGAIN;
+				goto out_wait;
+			}
+
+			prepare_to_wait(sk_sleep(sk), &wait,
+					TASK_INTERRUPTIBLE);
+		}
+
+		/* These checks occur both as part of and after the loop
+		 * conditional since we need to check before and after
+		 * sleeping.
+		 */
+		if (sk->sk_err) {
+			err = -sk->sk_err;
+			goto out_wait;
+		} else if ((sk->sk_shutdown & SEND_SHUTDOWN) ||
+			   (vsk->peer_shutdown & RCV_SHUTDOWN)) {
+			err = -EPIPE;
+			goto out_wait;
+		}
+
+		err = transport->notify_send_pre_enqueue(vsk, &send_data);
+		if (err < 0)
+			goto out_wait;
+
+		/* Note that enqueue will only write as many bytes as are free
+		 * in the produce queue, so we don't need to ensure len is
+		 * smaller than the queue size.  It is the caller's
+		 * responsibility to check how many bytes we were able to send.
+		 */
+
+		written = transport->stream_enqueue(
+				vsk, msg->msg_iov,
+				len - total_written);
+		if (written < 0) {
+			err = -ENOMEM;
+			goto out_wait;
+		}
+
+		total_written += written;
+
+		err = transport->notify_send_post_enqueue(
+				vsk, written, &send_data);
+		if (err < 0)
+			goto out_wait;
+
+	}
+
+out_wait:
+	if (total_written > 0)
+		err = total_written;
+	finish_wait(sk_sleep(sk), &wait);
+out:
+	release_sock(sk);
+	return err;
+}
+
+
+static int
+vsock_stream_recvmsg(struct kiocb *kiocb,
+		     struct socket *sock,
+		     struct msghdr *msg, size_t len, int flags)
+{
+	struct sock *sk;
+	struct vsock_sock *vsk;
+	int err;
+	size_t target;
+	ssize_t copied;
+	long timeout;
+	struct vsock_transport_recv_notify_data recv_data;
+
+	DEFINE_WAIT(wait);
+
+	sk = sock->sk;
+	vsk = vsock_sk(sk);
+	err = 0;
+
+	lock_sock(sk);
+
+	if (sk->sk_state != SS_CONNECTED) {
+		/* Recvmsg is supposed to return 0 if a peer performs an
+		 * orderly shutdown. Differentiate between that case and when a
+		 * peer has not connected or a local shutdown occured with the
+		 * SOCK_DONE flag.
+		 */
+		if (sock_flag(sk, SOCK_DONE))
+			err = 0;
+		else
+			err = -ENOTCONN;
+
+		goto out;
+	}
+
+	if (flags & MSG_OOB) {
+		err = -EOPNOTSUPP;
+		goto out;
+	}
+
+	/* We don't check peer_shutdown flag here since peer may actually shut
+	 * down, but there can be data in the queue that a local socket can
+	 * receive.
+	 */
+	if (sk->sk_shutdown & RCV_SHUTDOWN) {
+		err = 0;
+		goto out;
+	}
+
+	/* It is valid on Linux to pass in a zero-length receive buffer.  This
+	 * is not an error.  We may as well bail out now.
+	 */
+	if (!len) {
+		err = 0;
+		goto out;
+	}
+
+	/* We must not copy less than target bytes into the user's buffer
+	 * before returning successfully, so we wait for the consume queue to
+	 * have that much data to consume before dequeueing.  Note that this
+	 * makes it impossible to handle cases where target is greater than the
+	 * queue size.
+	 */
+	target = sock_rcvlowat(sk, flags & MSG_WAITALL, len);
+	if (target >= transport->stream_rcvhiwat(vsk)) {
+		err = -ENOMEM;
+		goto out;
+	}
+	timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT);
+	copied = 0;
+
+	err = transport->notify_recv_init(vsk, target, &recv_data);
+	if (err < 0)
+		goto out;
+
+	prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE);
+
+	while (1) {
+		s64 ready = vsock_stream_has_data(vsk);
+
+		if (ready < 0) {
+			/* Invalid queue pair content. XXX This should be
+			 * changed to a connection reset in a later change.
+			 */
+
+			err = -ENOMEM;
+			goto out_wait;
+		} else if (ready > 0) {
+			ssize_t read;
+
+			err = transport->notify_recv_pre_dequeue(
+					vsk, target, &recv_data);
+			if (err < 0)
+				break;
+
+			read = transport->stream_dequeue(
+					vsk, msg->msg_iov,
+					len - copied, flags);
+			if (read < 0) {
+				err = -ENOMEM;
+				break;
+			}
+
+			copied += read;
+
+			err = transport->notify_recv_post_dequeue(
+					vsk, target, read,
+					!(flags & MSG_PEEK), &recv_data);
+			if (err < 0)
+				goto out_wait;
+
+			if (read >= target || flags & MSG_PEEK)
+				break;
+
+			target -= read;
+		} else {
+			if (sk->sk_err != 0 || (sk->sk_shutdown & RCV_SHUTDOWN)
+			    || (vsk->peer_shutdown & SEND_SHUTDOWN)) {
+				break;
+			}
+			/* Don't wait for non-blocking sockets. */
+			if (timeout == 0) {
+				err = -EAGAIN;
+				break;
+			}
+
+			err = transport->notify_recv_pre_block(
+					vsk, target, &recv_data);
+			if (err < 0)
+				break;
+
+			release_sock(sk);
+			timeout = schedule_timeout(timeout);
+			lock_sock(sk);
+
+			if (signal_pending(current)) {
+				err = sock_intr_errno(timeout);
+				break;
+			} else if (timeout == 0) {
+				err = -EAGAIN;
+				break;
+			}
+
+			prepare_to_wait(sk_sleep(sk), &wait,
+					TASK_INTERRUPTIBLE);
+		}
+	}
+
+	if (sk->sk_err)
+		err = -sk->sk_err;
+	else if (sk->sk_shutdown & RCV_SHUTDOWN)
+		err = 0;
+
+	if (copied > 0) {
+		/* We only do these additional bookkeeping/notification steps
+		 * if we actually copied something out of the queue pair
+		 * instead of just peeking ahead.
+		 */
+
+		if (!(flags & MSG_PEEK)) {
+			/* If the other side has shutdown for sending and there
+			 * is nothing more to read, then modify the socket
+			 * state.
+			 */
+			if (vsk->peer_shutdown & SEND_SHUTDOWN) {
+				if (vsock_stream_has_data(vsk) <= 0) {
+					sk->sk_state = SS_UNCONNECTED;
+					sock_set_flag(sk, SOCK_DONE);
+					sk->sk_state_change(sk);
+				}
+			}
+		}
+		err = copied;
+	}
+
+out_wait:
+	finish_wait(sk_sleep(sk), &wait);
+out:
+	release_sock(sk);
+	return err;
+}
+
+static const struct proto_ops vsock_stream_ops = {
+	.family = PF_VSOCK,
+	.owner = THIS_MODULE,
+	.release = vsock_release,
+	.bind = vsock_bind,
+	.connect = vsock_stream_connect,
+	.socketpair = sock_no_socketpair,
+	.accept = vsock_accept,
+	.getname = vsock_getname,
+	.poll = vsock_poll,
+	.ioctl = sock_no_ioctl,
+	.listen = vsock_listen,
+	.shutdown = vsock_shutdown,
+	.setsockopt = vsock_stream_setsockopt,
+	.getsockopt = vsock_stream_getsockopt,
+	.sendmsg = vsock_stream_sendmsg,
+	.recvmsg = vsock_stream_recvmsg,
+	.mmap = sock_no_mmap,
+	.sendpage = sock_no_sendpage,
+};
+
+static int vsock_create(struct net *net, struct socket *sock,
+			int protocol, int kern)
+{
+	if (!sock)
+		return -EINVAL;
+
+	if (protocol)
+		return -EPROTONOSUPPORT;
+
+	switch (sock->type) {
+	case SOCK_DGRAM:
+		sock->ops = &vsock_dgram_ops;
+		break;
+	case SOCK_STREAM:
+		sock->ops = &vsock_stream_ops;
+		break;
+	default:
+		return -ESOCKTNOSUPPORT;
+	}
+
+	sock->state = SS_UNCONNECTED;
+
+	return __vsock_create(net, sock, NULL, GFP_KERNEL, 0) ? 0 : -ENOMEM;
+}
+
+static const struct net_proto_family vsock_family_ops = {
+	.family = AF_VSOCK,
+	.create = vsock_create,
+	.owner = THIS_MODULE,
+};
+
+static long vsock_dev_do_ioctl(struct file *filp,
+			       unsigned int cmd, void __user *ptr)
+{
+	u32 __user *p = ptr;
+	int retval = 0;
+
+	switch (cmd) {
+	case IOCTL_VM_SOCKETS_GET_LOCAL_CID:
+		if (put_user(transport->get_local_cid(), p) != 0)
+			retval = -EFAULT;
+		break;
+
+	default:
+		pr_err("Unknown ioctl %d\n", cmd);
+		retval = -EINVAL;
+	}
+
+	return retval;
+}
+
+static long vsock_dev_ioctl(struct file *filp,
+			    unsigned int cmd, unsigned long arg)
+{
+	return vsock_dev_do_ioctl(filp, cmd, (void __user *)arg);
+}
+
+#ifdef CONFIG_COMPAT
+static long vsock_dev_compat_ioctl(struct file *filp,
+				   unsigned int cmd, unsigned long arg)
+{
+	return vsock_dev_do_ioctl(filp, cmd, compat_ptr(arg));
+}
+#endif
+
+static const struct file_operations vsock_device_ops = {
+	.owner		= THIS_MODULE,
+	.unlocked_ioctl	= vsock_dev_ioctl,
+#ifdef CONFIG_COMPAT
+	.compat_ioctl	= vsock_dev_compat_ioctl,
+#endif
+	.open		= nonseekable_open,
+};
+
+static struct miscdevice vsock_device = {
+	.name		= "vsock",
+	.minor		= MISC_DYNAMIC_MINOR,
+	.fops		= &vsock_device_ops,
+};
+
+static int __vsock_core_init(void)
+{
+	int err;
+
+	vsock_init_tables();
+
+	err = misc_register(&vsock_device);
+	if (err) {
+		pr_err("Failed to register misc device\n");
+		return -ENOENT;
+	}
+
+	err = proto_register(&vsock_proto, 1);	/* we want our slab */
+	if (err) {
+		pr_err("Cannot register vsock protocol\n");
+		goto err_misc_deregister;
+	}
+
+	err = sock_register(&vsock_family_ops);
+	if (err) {
+		pr_err("could not register af_vsock (%d) address family: %d\n",
+		       AF_VSOCK, err);
+		goto err_unregister_proto;
+	}
+
+	return 0;
+
+err_unregister_proto:
+	proto_unregister(&vsock_proto);
+err_misc_deregister:
+	misc_deregister(&vsock_device);
+	return err;
+}
+
+int vsock_core_init(const struct vsock_transport *t)
+{
+	int retval = mutex_lock_interruptible(&vsock_register_mutex);
+	if (retval)
+		return retval;
+
+	if (transport) {
+		retval = -EBUSY;
+		goto out;
+	}
+
+	transport = t;
+	retval = __vsock_core_init();
+	if (retval)
+		transport = NULL;
+
+out:
+	mutex_unlock(&vsock_register_mutex);
+	return retval;
+}
+EXPORT_SYMBOL_GPL(vsock_core_init);
+
+void vsock_core_exit(void)
+{
+	mutex_lock(&vsock_register_mutex);
+
+	misc_deregister(&vsock_device);
+	sock_unregister(AF_VSOCK);
+	proto_unregister(&vsock_proto);
+
+	/* We do not want the assignment below re-ordered. */
+	mb();
+	transport = NULL;
+
+	mutex_unlock(&vsock_register_mutex);
+}
+EXPORT_SYMBOL_GPL(vsock_core_exit);
+
+MODULE_AUTHOR("VMware, Inc.");
+MODULE_DESCRIPTION("VMware Virtual Socket Family");
+MODULE_VERSION(VSOCK_DRIVER_VERSION_STRING);
+MODULE_LICENSE("GPL v2");

net/vmw_vsock/af_vsock.h

@@ -0,0 +1,175 @@
+/*
+ * VMware vSockets Driver
+ *
+ * Copyright (C) 2007-2013 VMware, Inc. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation version 2 and no later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ */
+
+#ifndef __AF_VSOCK_H__
+#define __AF_VSOCK_H__
+
+#include <linux/kernel.h>
+#include <linux/workqueue.h>
+#include <linux/vm_sockets.h>
+
+#include "vsock_addr.h"
+
+#define LAST_RESERVED_PORT 1023
+
+#define vsock_sk(__sk)    ((struct vsock_sock *)__sk)
+#define sk_vsock(__vsk)   (&(__vsk)->sk)
+
+struct vsock_sock {
+	/* sk must be the first member. */
+	struct sock sk;
+	struct sockaddr_vm local_addr;
+	struct sockaddr_vm remote_addr;
+	/* Links for the global tables of bound and connected sockets. */
+	struct list_head bound_table;
+	struct list_head connected_table;
+	/* Accessed without the socket lock held. This means it can never be
+	 * modified outsided of socket create or destruct.
+	 */
+	bool trusted;
+	bool cached_peer_allow_dgram;	/* Dgram communication allowed to
+					 * cached peer?
+					 */
+	u32 cached_peer;  /* Context ID of last dgram destination check. */
+	const struct cred *owner;
+	/* Rest are SOCK_STREAM only. */
+	long connect_timeout;
+	/* Listening socket that this came from. */
+	struct sock *listener;
+	/* Used for pending list and accept queue during connection handshake.
+	 * The listening socket is the head for both lists.  Sockets created
+	 * for connection requests are placed in the pending list until they
+	 * are connected, at which point they are put in the accept queue list
+	 * so they can be accepted in accept().  If accept() cannot accept the
+	 * connection, it is marked as rejected so the cleanup function knows
+	 * to clean up the socket.
+	 */
+	struct list_head pending_links;
+	struct list_head accept_queue;
+	bool rejected;
+	struct delayed_work dwork;
+	u32 peer_shutdown;
+	bool sent_request;
+	bool ignore_connecting_rst;
+
+	/* Private to transport. */
+	void *trans;
+};
+
+s64 vsock_stream_has_data(struct vsock_sock *vsk);
+s64 vsock_stream_has_space(struct vsock_sock *vsk);
+void vsock_pending_work(struct work_struct *work);
+struct sock *__vsock_create(struct net *net,
+			    struct socket *sock,
+			    struct sock *parent,
+			    gfp_t priority, unsigned short type);
+
+/**** TRANSPORT ****/
+
+struct vsock_transport_recv_notify_data {
+	u64 data1; /* Transport-defined. */
+	u64 data2; /* Transport-defined. */
+	bool notify_on_block;
+};
+
+struct vsock_transport_send_notify_data {
+	u64 data1; /* Transport-defined. */
+	u64 data2; /* Transport-defined. */
+};
+
+struct vsock_transport {
+	/* Initialize/tear-down socket. */
+	int (*init)(struct vsock_sock *, struct vsock_sock *);
+	void (*destruct)(struct vsock_sock *);
+	void (*release)(struct vsock_sock *);
+
+	/* Connections. */
+	int (*connect)(struct vsock_sock *);
+
+	/* DGRAM. */
+	int (*dgram_bind)(struct vsock_sock *, struct sockaddr_vm *);
+	int (*dgram_dequeue)(struct kiocb *kiocb, struct vsock_sock *vsk,
+			     struct msghdr *msg, size_t len, int flags);
+	int (*dgram_enqueue)(struct vsock_sock *, struct sockaddr_vm *,
+			     struct iovec *, size_t len);
+	bool (*dgram_allow)(u32 cid, u32 port);
+
+	/* STREAM. */
+	/* TODO: stream_bind() */
+	ssize_t (*stream_dequeue)(struct vsock_sock *, struct iovec *,
+				  size_t len, int flags);
+	ssize_t (*stream_enqueue)(struct vsock_sock *, struct iovec *,
+				  size_t len);
+	s64 (*stream_has_data)(struct vsock_sock *);
+	s64 (*stream_has_space)(struct vsock_sock *);
+	u64 (*stream_rcvhiwat)(struct vsock_sock *);
+	bool (*stream_is_active)(struct vsock_sock *);
+	bool (*stream_allow)(u32 cid, u32 port);
+
+	/* Notification. */
+	int (*notify_poll_in)(struct vsock_sock *, size_t, bool *);
+	int (*notify_poll_out)(struct vsock_sock *, size_t, bool *);
+	int (*notify_recv_init)(struct vsock_sock *, size_t,
+		struct vsock_transport_recv_notify_data *);
+	int (*notify_recv_pre_block)(struct vsock_sock *, size_t,
+		struct vsock_transport_recv_notify_data *);
+	int (*notify_recv_pre_dequeue)(struct vsock_sock *, size_t,
+		struct vsock_transport_recv_notify_data *);
+	int (*notify_recv_post_dequeue)(struct vsock_sock *, size_t,
+		ssize_t, bool, struct vsock_transport_recv_notify_data *);
+	int (*notify_send_init)(struct vsock_sock *,
+		struct vsock_transport_send_notify_data *);
+	int (*notify_send_pre_block)(struct vsock_sock *,
+		struct vsock_transport_send_notify_data *);
+	int (*notify_send_pre_enqueue)(struct vsock_sock *,
+		struct vsock_transport_send_notify_data *);
+	int (*notify_send_post_enqueue)(struct vsock_sock *, ssize_t,
+		struct vsock_transport_send_notify_data *);
+
+	/* Shutdown. */
+	int (*shutdown)(struct vsock_sock *, int);
+
+	/* Buffer sizes. */
+	void (*set_buffer_size)(struct vsock_sock *, u64);
+	void (*set_min_buffer_size)(struct vsock_sock *, u64);
+	void (*set_max_buffer_size)(struct vsock_sock *, u64);
+	u64 (*get_buffer_size)(struct vsock_sock *);
+	u64 (*get_min_buffer_size)(struct vsock_sock *);
+	u64 (*get_max_buffer_size)(struct vsock_sock *);
+
+	/* Addressing. */
+	u32 (*get_local_cid)(void);
+};
+
+/**** CORE ****/
+
+int vsock_core_init(const struct vsock_transport *t);
+void vsock_core_exit(void);
+
+/**** UTILS ****/
+
+void vsock_release_pending(struct sock *pending);
+void vsock_add_pending(struct sock *listener, struct sock *pending);
+void vsock_remove_pending(struct sock *listener, struct sock *pending);
+void vsock_enqueue_accept(struct sock *listener, struct sock *connected);
+void vsock_insert_connected(struct vsock_sock *vsk);
+void vsock_remove_bound(struct vsock_sock *vsk);
+void vsock_remove_connected(struct vsock_sock *vsk);
+struct sock *vsock_find_bound_socket(struct sockaddr_vm *addr);
+struct sock *vsock_find_connected_socket(struct sockaddr_vm *src,
+					 struct sockaddr_vm *dst);
+void vsock_for_each_connected_socket(void (*fn)(struct sock *sk));
+
+#endif /* __AF_VSOCK_H__ */

net/vmw_vsock/vmci_transport.c

@@ -0,0 +1,2157 @@
+/*
+ * VMware vSockets Driver
+ *
+ * Copyright (C) 2007-2013 VMware, Inc. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation version 2 and no later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ */
+
+#include <linux/types.h>
+
+#define EXPORT_SYMTAB
+#include <linux/bitops.h>
+#include <linux/cred.h>
+#include <linux/init.h>
+#include <linux/io.h>
+#include <linux/kernel.h>
+#include <linux/kmod.h>
+#include <linux/list.h>
+#include <linux/miscdevice.h>
+#include <linux/module.h>
+#include <linux/mutex.h>
+#include <linux/net.h>
+#include <linux/poll.h>
+#include <linux/skbuff.h>
+#include <linux/smp.h>
+#include <linux/socket.h>
+#include <linux/stddef.h>
+#include <linux/unistd.h>
+#include <linux/wait.h>
+#include <linux/workqueue.h>
+#include <net/sock.h>
+
+#include "af_vsock.h"
+#include "vmci_transport_notify.h"
+
+static int vmci_transport_recv_dgram_cb(void *data, struct vmci_datagram *dg);
+static int vmci_transport_recv_stream_cb(void *data, struct vmci_datagram *dg);
+static void vmci_transport_peer_attach_cb(u32 sub_id,
+					  const struct vmci_event_data *ed,
+					  void *client_data);
+static void vmci_transport_peer_detach_cb(u32 sub_id,
+					  const struct vmci_event_data *ed,
+					  void *client_data);
+static void vmci_transport_recv_pkt_work(struct work_struct *work);
+static int vmci_transport_recv_listen(struct sock *sk,
+				      struct vmci_transport_packet *pkt);
+static int vmci_transport_recv_connecting_server(
+					struct sock *sk,
+					struct sock *pending,
+					struct vmci_transport_packet *pkt);
+static int vmci_transport_recv_connecting_client(
+					struct sock *sk,
+					struct vmci_transport_packet *pkt);
+static int vmci_transport_recv_connecting_client_negotiate(
+					struct sock *sk,
+					struct vmci_transport_packet *pkt);
+static int vmci_transport_recv_connecting_client_invalid(
+					struct sock *sk,
+					struct vmci_transport_packet *pkt);
+static int vmci_transport_recv_connected(struct sock *sk,
+					 struct vmci_transport_packet *pkt);
+static bool vmci_transport_old_proto_override(bool *old_pkt_proto);
+static u16 vmci_transport_new_proto_supported_versions(void);
+static bool vmci_transport_proto_to_notify_struct(struct sock *sk, u16 *proto,
+						  bool old_pkt_proto);
+
+struct vmci_transport_recv_pkt_info {
+	struct work_struct work;
+	struct sock *sk;
+	struct vmci_transport_packet pkt;
+};
+
+static struct vmci_handle vmci_transport_stream_handle = { VMCI_INVALID_ID,
+							   VMCI_INVALID_ID };
+static u32 vmci_transport_qp_resumed_sub_id = VMCI_INVALID_ID;
+
+static int PROTOCOL_OVERRIDE = -1;
+
+#define VMCI_TRANSPORT_DEFAULT_QP_SIZE_MIN   128
+#define VMCI_TRANSPORT_DEFAULT_QP_SIZE       262144
+#define VMCI_TRANSPORT_DEFAULT_QP_SIZE_MAX   262144
+
+/* The default peer timeout indicates how long we will wait for a peer response
+ * to a control message.
+ */
+#define VSOCK_DEFAULT_CONNECT_TIMEOUT (2 * HZ)
+
+#define SS_LISTEN 255
+
+/* Helper function to convert from a VMCI error code to a VSock error code. */
+
+static s32 vmci_transport_error_to_vsock_error(s32 vmci_error)
+{
+	int err;
+
+	switch (vmci_error) {
+	case VMCI_ERROR_NO_MEM:
+		err = ENOMEM;
+		break;
+	case VMCI_ERROR_DUPLICATE_ENTRY:
+	case VMCI_ERROR_ALREADY_EXISTS:
+		err = EADDRINUSE;
+		break;
+	case VMCI_ERROR_NO_ACCESS:
+		err = EPERM;
+		break;
+	case VMCI_ERROR_NO_RESOURCES:
+		err = ENOBUFS;
+		break;
+	case VMCI_ERROR_INVALID_RESOURCE:
+		err = EHOSTUNREACH;
+		break;
+	case VMCI_ERROR_INVALID_ARGS:
+	default:
+		err = EINVAL;
+	}
+
+	return err > 0 ? -err : err;
+}
+
+static inline void
+vmci_transport_packet_init(struct vmci_transport_packet *pkt,
+			   struct sockaddr_vm *src,
+			   struct sockaddr_vm *dst,
+			   u8 type,
+			   u64 size,
+			   u64 mode,
+			   struct vmci_transport_waiting_info *wait,
+			   u16 proto,
+			   struct vmci_handle handle)
+{
+	/* We register the stream control handler as an any cid handle so we
+	 * must always send from a source address of VMADDR_CID_ANY
+	 */
+	pkt->dg.src = vmci_make_handle(VMADDR_CID_ANY,
+				       VMCI_TRANSPORT_PACKET_RID);
+	pkt->dg.dst = vmci_make_handle(dst->svm_cid,
+				       VMCI_TRANSPORT_PACKET_RID);
+	pkt->dg.payload_size = sizeof(*pkt) - sizeof(pkt->dg);
+	pkt->version = VMCI_TRANSPORT_PACKET_VERSION;
+	pkt->type = type;
+	pkt->src_port = src->svm_port;
+	pkt->dst_port = dst->svm_port;
+	memset(&pkt->proto, 0, sizeof(pkt->proto));
+	memset(&pkt->_reserved2, 0, sizeof(pkt->_reserved2));
+
+	switch (pkt->type) {
+	case VMCI_TRANSPORT_PACKET_TYPE_INVALID:
+		pkt->u.size = 0;
+		break;
+
+	case VMCI_TRANSPORT_PACKET_TYPE_REQUEST:
+	case VMCI_TRANSPORT_PACKET_TYPE_NEGOTIATE:
+		pkt->u.size = size;
+		break;
+
+	case VMCI_TRANSPORT_PACKET_TYPE_OFFER:
+	case VMCI_TRANSPORT_PACKET_TYPE_ATTACH:
+		pkt->u.handle = handle;
+		break;
+
+	case VMCI_TRANSPORT_PACKET_TYPE_WROTE:
+	case VMCI_TRANSPORT_PACKET_TYPE_READ:
+	case VMCI_TRANSPORT_PACKET_TYPE_RST:
+		pkt->u.size = 0;
+		break;
+
+	case VMCI_TRANSPORT_PACKET_TYPE_SHUTDOWN:
+		pkt->u.mode = mode;
+		break;
+
+	case VMCI_TRANSPORT_PACKET_TYPE_WAITING_READ:
+	case VMCI_TRANSPORT_PACKET_TYPE_WAITING_WRITE:
+		memcpy(&pkt->u.wait, wait, sizeof(pkt->u.wait));
+		break;
+
+	case VMCI_TRANSPORT_PACKET_TYPE_REQUEST2:
+	case VMCI_TRANSPORT_PACKET_TYPE_NEGOTIATE2:
+		pkt->u.size = size;
+		pkt->proto = proto;
+		break;
+	}
+}
+
+static inline void
+vmci_transport_packet_get_addresses(struct vmci_transport_packet *pkt,
+				    struct sockaddr_vm *local,
+				    struct sockaddr_vm *remote)
+{
+	vsock_addr_init(local, pkt->dg.dst.context, pkt->dst_port);
+	vsock_addr_init(remote, pkt->dg.src.context, pkt->src_port);
+}
+
+static int
+__vmci_transport_send_control_pkt(struct vmci_transport_packet *pkt,
+				  struct sockaddr_vm *src,
+				  struct sockaddr_vm *dst,
+				  enum vmci_transport_packet_type type,
+				  u64 size,
+				  u64 mode,
+				  struct vmci_transport_waiting_info *wait,
+				  u16 proto,
+				  struct vmci_handle handle,
+				  bool convert_error)
+{
+	int err;
+
+	vmci_transport_packet_init(pkt, src, dst, type, size, mode, wait,
+				   proto, handle);
+	err = vmci_datagram_send(&pkt->dg);
+	if (convert_error && (err < 0))
+		return vmci_transport_error_to_vsock_error(err);
+
+	return err;
+}
+
+static int
+vmci_transport_reply_control_pkt_fast(struct vmci_transport_packet *pkt,
+				      enum vmci_transport_packet_type type,
+				      u64 size,
+				      u64 mode,
+				      struct vmci_transport_waiting_info *wait,
+				      struct vmci_handle handle)
+{
+	struct vmci_transport_packet reply;
+	struct sockaddr_vm src, dst;
+
+	if (pkt->type == VMCI_TRANSPORT_PACKET_TYPE_RST) {
+		return 0;
+	} else {
+		vmci_transport_packet_get_addresses(pkt, &src, &dst);
+		return __vmci_transport_send_control_pkt(&reply, &src, &dst,
+							 type,
+							 size, mode, wait,
+							 VSOCK_PROTO_INVALID,
+							 handle, true);
+	}
+}
+
+static int
+vmci_transport_send_control_pkt_bh(struct sockaddr_vm *src,
+				   struct sockaddr_vm *dst,
+				   enum vmci_transport_packet_type type,
+				   u64 size,
+				   u64 mode,
+				   struct vmci_transport_waiting_info *wait,
+				   struct vmci_handle handle)
+{
+	/* Note that it is safe to use a single packet across all CPUs since
+	 * two tasklets of the same type are guaranteed to not ever run
+	 * simultaneously. If that ever changes, or VMCI stops using tasklets,
+	 * we can use per-cpu packets.
+	 */
+	static struct vmci_transport_packet pkt;
+
+	return __vmci_transport_send_control_pkt(&pkt, src, dst, type,
+						 size, mode, wait,
+						 VSOCK_PROTO_INVALID, handle,
+						 false);
+}
+
+static int
+vmci_transport_send_control_pkt(struct sock *sk,
+				enum vmci_transport_packet_type type,
+				u64 size,
+				u64 mode,
+				struct vmci_transport_waiting_info *wait,
+				u16 proto,
+				struct vmci_handle handle)
+{
+	struct vmci_transport_packet *pkt;
+	struct vsock_sock *vsk;
+	int err;
+
+	vsk = vsock_sk(sk);
+
+	if (!vsock_addr_bound(&vsk->local_addr))
+		return -EINVAL;
+
+	if (!vsock_addr_bound(&vsk->remote_addr))
+		return -EINVAL;
+
+	pkt = kmalloc(sizeof(*pkt), GFP_KERNEL);
+	if (!pkt)
+		return -ENOMEM;
+
+	err = __vmci_transport_send_control_pkt(pkt, &vsk->local_addr,
+						&vsk->remote_addr, type, size,
+						mode, wait, proto, handle,
+						true);
+	kfree(pkt);
+
+	return err;
+}
+
+static int vmci_transport_send_reset_bh(struct sockaddr_vm *dst,
+					struct sockaddr_vm *src,
+					struct vmci_transport_packet *pkt)
+{
+	if (pkt->type == VMCI_TRANSPORT_PACKET_TYPE_RST)
+		return 0;
+	return vmci_transport_send_control_pkt_bh(
+					dst, src,
+					VMCI_TRANSPORT_PACKET_TYPE_RST, 0,
+					0, NULL, VMCI_INVALID_HANDLE);
+}
+
+static int vmci_transport_send_reset(struct sock *sk,
+				     struct vmci_transport_packet *pkt)
+{
+	if (pkt->type == VMCI_TRANSPORT_PACKET_TYPE_RST)
+		return 0;
+	return vmci_transport_send_control_pkt(sk,
+					VMCI_TRANSPORT_PACKET_TYPE_RST,
+					0, 0, NULL, VSOCK_PROTO_INVALID,
+					VMCI_INVALID_HANDLE);
+}
+
+static int vmci_transport_send_negotiate(struct sock *sk, size_t size)
+{
+	return vmci_transport_send_control_pkt(
+					sk,
+					VMCI_TRANSPORT_PACKET_TYPE_NEGOTIATE,
+					size, 0, NULL,
+					VSOCK_PROTO_INVALID,
+					VMCI_INVALID_HANDLE);
+}
+
+static int vmci_transport_send_negotiate2(struct sock *sk, size_t size,
+					  u16 version)
+{
+	return vmci_transport_send_control_pkt(
+					sk,
+					VMCI_TRANSPORT_PACKET_TYPE_NEGOTIATE2,
+					size, 0, NULL, version,
+					VMCI_INVALID_HANDLE);
+}
+
+static int vmci_transport_send_qp_offer(struct sock *sk,
+					struct vmci_handle handle)
+{
+	return vmci_transport_send_control_pkt(
+					sk, VMCI_TRANSPORT_PACKET_TYPE_OFFER, 0,
+					0, NULL,
+					VSOCK_PROTO_INVALID, handle);
+}
+
+static int vmci_transport_send_attach(struct sock *sk,
+				      struct vmci_handle handle)
+{
+	return vmci_transport_send_control_pkt(
+					sk, VMCI_TRANSPORT_PACKET_TYPE_ATTACH,
+					0, 0, NULL, VSOCK_PROTO_INVALID,
+					handle);
+}
+
+static int vmci_transport_reply_reset(struct vmci_transport_packet *pkt)
+{
+	return vmci_transport_reply_control_pkt_fast(
+						pkt,
+						VMCI_TRANSPORT_PACKET_TYPE_RST,
+						0, 0, NULL,
+						VMCI_INVALID_HANDLE);
+}
+
+static int vmci_transport_send_invalid_bh(struct sockaddr_vm *dst,
+					  struct sockaddr_vm *src)
+{
+	return vmci_transport_send_control_pkt_bh(
+					dst, src,
+					VMCI_TRANSPORT_PACKET_TYPE_INVALID,
+					0, 0, NULL, VMCI_INVALID_HANDLE);
+}
+
+int vmci_transport_send_wrote_bh(struct sockaddr_vm *dst,
+				 struct sockaddr_vm *src)
+{
+	return vmci_transport_send_control_pkt_bh(
+					dst, src,
+					VMCI_TRANSPORT_PACKET_TYPE_WROTE, 0,
+					0, NULL, VMCI_INVALID_HANDLE);
+}
+
+int vmci_transport_send_read_bh(struct sockaddr_vm *dst,
+				struct sockaddr_vm *src)
+{
+	return vmci_transport_send_control_pkt_bh(
+					dst, src,
+					VMCI_TRANSPORT_PACKET_TYPE_READ, 0,
+					0, NULL, VMCI_INVALID_HANDLE);
+}
+
+int vmci_transport_send_wrote(struct sock *sk)
+{
+	return vmci_transport_send_control_pkt(
+					sk, VMCI_TRANSPORT_PACKET_TYPE_WROTE, 0,
+					0, NULL, VSOCK_PROTO_INVALID,
+					VMCI_INVALID_HANDLE);
+}
+
+int vmci_transport_send_read(struct sock *sk)
+{
+	return vmci_transport_send_control_pkt(
+					sk, VMCI_TRANSPORT_PACKET_TYPE_READ, 0,
+					0, NULL, VSOCK_PROTO_INVALID,
+					VMCI_INVALID_HANDLE);
+}
+
+int vmci_transport_send_waiting_write(struct sock *sk,
+				      struct vmci_transport_waiting_info *wait)
+{
+	return vmci_transport_send_control_pkt(
+				sk, VMCI_TRANSPORT_PACKET_TYPE_WAITING_WRITE,
+				0, 0, wait, VSOCK_PROTO_INVALID,
+				VMCI_INVALID_HANDLE);
+}
+
+int vmci_transport_send_waiting_read(struct sock *sk,
+				     struct vmci_transport_waiting_info *wait)
+{
+	return vmci_transport_send_control_pkt(
+				sk, VMCI_TRANSPORT_PACKET_TYPE_WAITING_READ,
+				0, 0, wait, VSOCK_PROTO_INVALID,
+				VMCI_INVALID_HANDLE);
+}
+
+static int vmci_transport_shutdown(struct vsock_sock *vsk, int mode)
+{
+	return vmci_transport_send_control_pkt(
+					&vsk->sk,
+					VMCI_TRANSPORT_PACKET_TYPE_SHUTDOWN,
+					0, mode, NULL,
+					VSOCK_PROTO_INVALID,
+					VMCI_INVALID_HANDLE);
+}
+
+static int vmci_transport_send_conn_request(struct sock *sk, size_t size)
+{
+	return vmci_transport_send_control_pkt(sk,
+					VMCI_TRANSPORT_PACKET_TYPE_REQUEST,
+					size, 0, NULL,
+					VSOCK_PROTO_INVALID,
+					VMCI_INVALID_HANDLE);
+}
+
+static int vmci_transport_send_conn_request2(struct sock *sk, size_t size,
+					     u16 version)
+{
+	return vmci_transport_send_control_pkt(
+					sk, VMCI_TRANSPORT_PACKET_TYPE_REQUEST2,
+					size, 0, NULL, version,
+					VMCI_INVALID_HANDLE);
+}
+
+static struct sock *vmci_transport_get_pending(
+					struct sock *listener,
+					struct vmci_transport_packet *pkt)
+{
+	struct vsock_sock *vlistener;
+	struct vsock_sock *vpending;
+	struct sock *pending;
+
+	vlistener = vsock_sk(listener);
+
+	list_for_each_entry(vpending, &vlistener->pending_links,
+			    pending_links) {
+		struct sockaddr_vm src;
+		struct sockaddr_vm dst;
+
+		vsock_addr_init(&src, pkt->dg.src.context, pkt->src_port);
+		vsock_addr_init(&dst, pkt->dg.dst.context, pkt->dst_port);
+
+		if (vsock_addr_equals_addr(&src, &vpending->remote_addr) &&
+		    vsock_addr_equals_addr(&dst, &vpending->local_addr)) {
+			pending = sk_vsock(vpending);
+			sock_hold(pending);
+			goto found;
+		}
+	}
+
+	pending = NULL;
+found:
+	return pending;
+
+}
+
+static void vmci_transport_release_pending(struct sock *pending)
+{
+	sock_put(pending);
+}
+
+/* We allow two kinds of sockets to communicate with a restricted VM: 1)
+ * trusted sockets 2) sockets from applications running as the same user as the
+ * VM (this is only true for the host side and only when using hosted products)
+ */
+
+static bool vmci_transport_is_trusted(struct vsock_sock *vsock, u32 peer_cid)
+{
+	return vsock->trusted ||
+	       vmci_is_context_owner(peer_cid, vsock->owner->uid);
+}
+
+/* We allow sending datagrams to and receiving datagrams from a restricted VM
+ * only if it is trusted as described in vmci_transport_is_trusted.
+ */
+
+static bool vmci_transport_allow_dgram(struct vsock_sock *vsock, u32 peer_cid)
+{
+	if (vsock->cached_peer != peer_cid) {
+		vsock->cached_peer = peer_cid;
+		if (!vmci_transport_is_trusted(vsock, peer_cid) &&
+		    (vmci_context_get_priv_flags(peer_cid) &
+		     VMCI_PRIVILEGE_FLAG_RESTRICTED)) {
+			vsock->cached_peer_allow_dgram = false;
+		} else {
+			vsock->cached_peer_allow_dgram = true;
+		}
+	}
+
+	return vsock->cached_peer_allow_dgram;
+}
+
+static int
+vmci_transport_queue_pair_alloc(struct vmci_qp **qpair,
+				struct vmci_handle *handle,
+				u64 produce_size,
+				u64 consume_size,
+				u32 peer, u32 flags, bool trusted)
+{
+	int err = 0;
+
+	if (trusted) {
+		/* Try to allocate our queue pair as trusted. This will only
+		 * work if vsock is running in the host.
+		 */
+
+		err = vmci_qpair_alloc(qpair, handle, produce_size,
+				       consume_size,
+				       peer, flags,
+				       VMCI_PRIVILEGE_FLAG_TRUSTED);
+		if (err != VMCI_ERROR_NO_ACCESS)
+			goto out;
+
+	}
+
+	err = vmci_qpair_alloc(qpair, handle, produce_size, consume_size,
+			       peer, flags, VMCI_NO_PRIVILEGE_FLAGS);
+out:
+	if (err < 0) {
+		pr_err("Could not attach to queue pair with %d\n",
+		       err);
+		err = vmci_transport_error_to_vsock_error(err);
+	}
+
+	return err;
+}
+
+static int
+vmci_transport_datagram_create_hnd(u32 resource_id,
+				   u32 flags,
+				   vmci_datagram_recv_cb recv_cb,
+				   void *client_data,
+				   struct vmci_handle *out_handle)
+{
+	int err = 0;
+
+	/* Try to allocate our datagram handler as trusted. This will only work
+	 * if vsock is running in the host.
+	 */
+
+	err = vmci_datagram_create_handle_priv(resource_id, flags,
+					       VMCI_PRIVILEGE_FLAG_TRUSTED,
+					       recv_cb,
+					       client_data, out_handle);
+
+	if (err == VMCI_ERROR_NO_ACCESS)
+		err = vmci_datagram_create_handle(resource_id, flags,
+						  recv_cb, client_data,
+						  out_handle);
+
+	return err;
+}
+
+/* This is invoked as part of a tasklet that's scheduled when the VMCI
+ * interrupt fires.  This is run in bottom-half context and if it ever needs to
+ * sleep it should defer that work to a work queue.
+ */
+
+static int vmci_transport_recv_dgram_cb(void *data, struct vmci_datagram *dg)
+{
+	struct sock *sk;
+	size_t size;
+	struct sk_buff *skb;
+	struct vsock_sock *vsk;
+
+	sk = (struct sock *)data;
+
+	/* This handler is privileged when this module is running on the host.
+	 * We will get datagrams from all endpoints (even VMs that are in a
+	 * restricted context). If we get one from a restricted context then
+	 * the destination socket must be trusted.
+	 *
+	 * NOTE: We access the socket struct without holding the lock here.
+	 * This is ok because the field we are interested is never modified
+	 * outside of the create and destruct socket functions.
+	 */
+	vsk = vsock_sk(sk);
+	if (!vmci_transport_allow_dgram(vsk, dg->src.context))
+		return VMCI_ERROR_NO_ACCESS;
+
+	size = VMCI_DG_SIZE(dg);
+
+	/* Attach the packet to the socket's receive queue as an sk_buff. */
+	skb = alloc_skb(size, GFP_ATOMIC);
+	if (skb) {
+		/* sk_receive_skb() will do a sock_put(), so hold here. */
+		sock_hold(sk);
+		skb_put(skb, size);
+		memcpy(skb->data, dg, size);
+		sk_receive_skb(sk, skb, 0);
+	}
+
+	return VMCI_SUCCESS;
+}
+
+static bool vmci_transport_stream_allow(u32 cid, u32 port)
+{
+	static const u32 non_socket_contexts[] = {
+		VMADDR_CID_HYPERVISOR,
+		VMADDR_CID_RESERVED,
+	};
+	int i;
+
+	BUILD_BUG_ON(sizeof(cid) != sizeof(*non_socket_contexts));
+
+	for (i = 0; i < ARRAY_SIZE(non_socket_contexts); i++) {
+		if (cid == non_socket_contexts[i])
+			return false;
+	}
+
+	return true;
+}
+
+/* This is invoked as part of a tasklet that's scheduled when the VMCI
+ * interrupt fires.  This is run in bottom-half context but it defers most of
+ * its work to the packet handling work queue.
+ */
+
+static int vmci_transport_recv_stream_cb(void *data, struct vmci_datagram *dg)
+{
+	struct sock *sk;
+	struct sockaddr_vm dst;
+	struct sockaddr_vm src;
+	struct vmci_transport_packet *pkt;
+	struct vsock_sock *vsk;
+	bool bh_process_pkt;
+	int err;
+
+	sk = NULL;
+	err = VMCI_SUCCESS;
+	bh_process_pkt = false;
+
+	/* Ignore incoming packets from contexts without sockets, or resources
+	 * that aren't vsock implementations.
+	 */
+
+	if (!vmci_transport_stream_allow(dg->src.context, -1)
+	    || VMCI_TRANSPORT_PACKET_RID != dg->src.resource)
+		return VMCI_ERROR_NO_ACCESS;
+
+	if (VMCI_DG_SIZE(dg) < sizeof(*pkt))
+		/* Drop datagrams that do not contain full VSock packets. */
+		return VMCI_ERROR_INVALID_ARGS;
+
+	pkt = (struct vmci_transport_packet *)dg;
+
+	/* Find the socket that should handle this packet.  First we look for a
+	 * connected socket and if there is none we look for a socket bound to
+	 * the destintation address.
+	 */
+	vsock_addr_init(&src, pkt->dg.src.context, pkt->src_port);
+	vsock_addr_init(&dst, pkt->dg.dst.context, pkt->dst_port);
+
+	sk = vsock_find_connected_socket(&src, &dst);
+	if (!sk) {
+		sk = vsock_find_bound_socket(&dst);
+		if (!sk) {
+			/* We could not find a socket for this specified
+			 * address.  If this packet is a RST, we just drop it.
+			 * If it is another packet, we send a RST.  Note that
+			 * we do not send a RST reply to RSTs so that we do not
+			 * continually send RSTs between two endpoints.
+			 *
+			 * Note that since this is a reply, dst is src and src
+			 * is dst.
+			 */
+			if (vmci_transport_send_reset_bh(&dst, &src, pkt) < 0)
+				pr_err("unable to send reset\n");
+
+			err = VMCI_ERROR_NOT_FOUND;
+			goto out;
+		}
+	}
+
+	/* If the received packet type is beyond all types known to this
+	 * implementation, reply with an invalid message.  Hopefully this will
+	 * help when implementing backwards compatibility in the future.
+	 */
+	if (pkt->type >= VMCI_TRANSPORT_PACKET_TYPE_MAX) {
+		vmci_transport_send_invalid_bh(&dst, &src);
+		err = VMCI_ERROR_INVALID_ARGS;
+		goto out;
+	}
+
+	/* This handler is privileged when this module is running on the host.
+	 * We will get datagram connect requests from all endpoints (even VMs
+	 * that are in a restricted context). If we get one from a restricted
+	 * context then the destination socket must be trusted.
+	 *
+	 * NOTE: We access the socket struct without holding the lock here.
+	 * This is ok because the field we are interested is never modified
+	 * outside of the create and destruct socket functions.
+	 */
+	vsk = vsock_sk(sk);
+	if (!vmci_transport_allow_dgram(vsk, pkt->dg.src.context)) {
+		err = VMCI_ERROR_NO_ACCESS;
+		goto out;
+	}
+
+	/* We do most everything in a work queue, but let's fast path the
+	 * notification of reads and writes to help data transfer performance.
+	 * We can only do this if there is no process context code executing
+	 * for this socket since that may change the state.
+	 */
+	bh_lock_sock(sk);
+
+	if (!sock_owned_by_user(sk) && sk->sk_state == SS_CONNECTED)
+		vmci_trans(vsk)->notify_ops->handle_notify_pkt(
+				sk, pkt, true, &dst, &src,
+				&bh_process_pkt);
+
+	bh_unlock_sock(sk);
+
+	if (!bh_process_pkt) {
+		struct vmci_transport_recv_pkt_info *recv_pkt_info;
+
+		recv_pkt_info = kmalloc(sizeof(*recv_pkt_info), GFP_ATOMIC);
+		if (!recv_pkt_info) {
+			if (vmci_transport_send_reset_bh(&dst, &src, pkt) < 0)
+				pr_err("unable to send reset\n");
+
+			err = VMCI_ERROR_NO_MEM;
+			goto out;
+		}
+
+		recv_pkt_info->sk = sk;
+		memcpy(&recv_pkt_info->pkt, pkt, sizeof(recv_pkt_info->pkt));
+		INIT_WORK(&recv_pkt_info->work, vmci_transport_recv_pkt_work);
+
+		schedule_work(&recv_pkt_info->work);
+		/* Clear sk so that the reference count incremented by one of
+		 * the Find functions above is not decremented below.  We need
+		 * that reference count for the packet handler we've scheduled
+		 * to run.
+		 */
+		sk = NULL;
+	}
+
+out:
+	if (sk)
+		sock_put(sk);
+
+	return err;
+}
+
+static void vmci_transport_peer_attach_cb(u32 sub_id,
+					  const struct vmci_event_data *e_data,
+					  void *client_data)
+{
+	struct sock *sk = client_data;
+	const struct vmci_event_payload_qp *e_payload;
+	struct vsock_sock *vsk;
+
+	e_payload = vmci_event_data_const_payload(e_data);
+
+	vsk = vsock_sk(sk);
+
+	/* We don't ask for delayed CBs when we subscribe to this event (we
+	 * pass 0 as flags to vmci_event_subscribe()).  VMCI makes no
+	 * guarantees in that case about what context we might be running in,
+	 * so it could be BH or process, blockable or non-blockable.  So we
+	 * need to account for all possible contexts here.
+	 */
+	local_bh_disable();
+	bh_lock_sock(sk);
+
+	/* XXX This is lame, we should provide a way to lookup sockets by
+	 * qp_handle.
+	 */
+	if (vmci_handle_is_equal(vmci_trans(vsk)->qp_handle,
+				 e_payload->handle)) {
+		/* XXX This doesn't do anything, but in the future we may want
+		 * to set a flag here to verify the attach really did occur and
+		 * we weren't just sent a datagram claiming it was.
+		 */
+		goto out;
+	}
+
+out:
+	bh_unlock_sock(sk);
+	local_bh_enable();
+}
+
+static void vmci_transport_handle_detach(struct sock *sk)
+{
+	struct vsock_sock *vsk;
+
+	vsk = vsock_sk(sk);
+	if (!vmci_handle_is_invalid(vmci_trans(vsk)->qp_handle)) {
+		sock_set_flag(sk, SOCK_DONE);
+
+		/* On a detach the peer will not be sending or receiving
+		 * anymore.
+		 */
+		vsk->peer_shutdown = SHUTDOWN_MASK;
+
+		/* We should not be sending anymore since the peer won't be
+		 * there to receive, but we can still receive if there is data
+		 * left in our consume queue.
+		 */
+		if (vsock_stream_has_data(vsk) <= 0) {
+			if (sk->sk_state == SS_CONNECTING) {
+				/* The peer may detach from a queue pair while
+				 * we are still in the connecting state, i.e.,
+				 * if the peer VM is killed after attaching to
+				 * a queue pair, but before we complete the
+				 * handshake. In that case, we treat the detach
+				 * event like a reset.
+				 */
+
+				sk->sk_state = SS_UNCONNECTED;
+				sk->sk_err = ECONNRESET;
+				sk->sk_error_report(sk);
+				return;
+			}
+			sk->sk_state = SS_UNCONNECTED;
+		}
+		sk->sk_state_change(sk);
+	}
+}
+
+static void vmci_transport_peer_detach_cb(u32 sub_id,
+					  const struct vmci_event_data *e_data,
+					  void *client_data)
+{
+	struct sock *sk = client_data;
+	const struct vmci_event_payload_qp *e_payload;
+	struct vsock_sock *vsk;
+
+	e_payload = vmci_event_data_const_payload(e_data);
+	vsk = vsock_sk(sk);
+	if (vmci_handle_is_invalid(e_payload->handle))
+		return;
+
+	/* Same rules for locking as for peer_attach_cb(). */
+	local_bh_disable();
+	bh_lock_sock(sk);
+
+	/* XXX This is lame, we should provide a way to lookup sockets by
+	 * qp_handle.
+	 */
+	if (vmci_handle_is_equal(vmci_trans(vsk)->qp_handle,
+				 e_payload->handle))
+		vmci_transport_handle_detach(sk);
+
+	bh_unlock_sock(sk);
+	local_bh_enable();
+}
+
+static void vmci_transport_qp_resumed_cb(u32 sub_id,
+					 const struct vmci_event_data *e_data,
+					 void *client_data)
+{
+	vsock_for_each_connected_socket(vmci_transport_handle_detach);
+}
+
+static void vmci_transport_recv_pkt_work(struct work_struct *work)
+{
+	struct vmci_transport_recv_pkt_info *recv_pkt_info;
+	struct vmci_transport_packet *pkt;
+	struct sock *sk;
+
+	recv_pkt_info =
+		container_of(work, struct vmci_transport_recv_pkt_info, work);
+	sk = recv_pkt_info->sk;
+	pkt = &recv_pkt_info->pkt;
+
+	lock_sock(sk);
+
+	switch (sk->sk_state) {
+	case SS_LISTEN:
+		vmci_transport_recv_listen(sk, pkt);
+		break;
+	case SS_CONNECTING:
+		/* Processing of pending connections for servers goes through
+		 * the listening socket, so see vmci_transport_recv_listen()
+		 * for that path.
+		 */
+		vmci_transport_recv_connecting_client(sk, pkt);
+		break;
+	case SS_CONNECTED:
+		vmci_transport_recv_connected(sk, pkt);
+		break;
+	default:
+		/* Because this function does not run in the same context as
+		 * vmci_transport_recv_stream_cb it is possible that the
+		 * socket has closed. We need to let the other side know or it
+		 * could be sitting in a connect and hang forever. Send a
+		 * reset to prevent that.
+		 */
+		vmci_transport_send_reset(sk, pkt);
+		goto out;
+	}
+
+out:
+	release_sock(sk);
+	kfree(recv_pkt_info);
+	/* Release reference obtained in the stream callback when we fetched
+	 * this socket out of the bound or connected list.
+	 */
+	sock_put(sk);
+}
+
+static int vmci_transport_recv_listen(struct sock *sk,
+				      struct vmci_transport_packet *pkt)
+{
+	struct sock *pending;
+	struct vsock_sock *vpending;
+	int err;
+	u64 qp_size;
+	bool old_request = false;
+	bool old_pkt_proto = false;
+
+	err = 0;
+
+	/* Because we are in the listen state, we could be receiving a packet
+	 * for ourself or any previous connection requests that we received.
+	 * If it's the latter, we try to find a socket in our list of pending
+	 * connections and, if we do, call the appropriate handler for the
+	 * state that that socket is in.  Otherwise we try to service the
+	 * connection request.
+	 */
+	pending = vmci_transport_get_pending(sk, pkt);
+	if (pending) {
+		lock_sock(pending);
+		switch (pending->sk_state) {
+		case SS_CONNECTING:
+			err = vmci_transport_recv_connecting_server(sk,
+								    pending,
+								    pkt);
+			break;
+		default:
+			vmci_transport_send_reset(pending, pkt);
+			err = -EINVAL;
+		}
+
+		if (err < 0)
+			vsock_remove_pending(sk, pending);
+
+		release_sock(pending);
+		vmci_transport_release_pending(pending);
+
+		return err;
+	}
+
+	/* The listen state only accepts connection requests.  Reply with a
+	 * reset unless we received a reset.
+	 */
+
+	if (!(pkt->type == VMCI_TRANSPORT_PACKET_TYPE_REQUEST ||
+	      pkt->type == VMCI_TRANSPORT_PACKET_TYPE_REQUEST2)) {
+		vmci_transport_reply_reset(pkt);
+		return -EINVAL;
+	}
+
+	if (pkt->u.size == 0) {
+		vmci_transport_reply_reset(pkt);
+		return -EINVAL;
+	}
+
+	/* If this socket can't accommodate this connection request, we send a
+	 * reset.  Otherwise we create and initialize a child socket and reply
+	 * with a connection negotiation.
+	 */
+	if (sk->sk_ack_backlog >= sk->sk_max_ack_backlog) {
+		vmci_transport_reply_reset(pkt);
+		return -ECONNREFUSED;
+	}
+
+	pending = __vsock_create(sock_net(sk), NULL, sk, GFP_KERNEL,
+				 sk->sk_type);
+	if (!pending) {
+		vmci_transport_send_reset(sk, pkt);
+		return -ENOMEM;
+	}
+
+	vpending = vsock_sk(pending);
+
+	vsock_addr_init(&vpending->local_addr, pkt->dg.dst.context,
+			pkt->dst_port);
+	vsock_addr_init(&vpending->remote_addr, pkt->dg.src.context,
+			pkt->src_port);
+
+	/* If the proposed size fits within our min/max, accept it. Otherwise
+	 * propose our own size.
+	 */
+	if (pkt->u.size >= vmci_trans(vpending)->queue_pair_min_size &&
+	    pkt->u.size <= vmci_trans(vpending)->queue_pair_max_size) {
+		qp_size = pkt->u.size;
+	} else {
+		qp_size = vmci_trans(vpending)->queue_pair_size;
+	}
+
+	/* Figure out if we are using old or new requests based on the
+	 * overrides pkt types sent by our peer.
+	 */
+	if (vmci_transport_old_proto_override(&old_pkt_proto)) {
+		old_request = old_pkt_proto;
+	} else {
+		if (pkt->type == VMCI_TRANSPORT_PACKET_TYPE_REQUEST)
+			old_request = true;
+		else if (pkt->type == VMCI_TRANSPORT_PACKET_TYPE_REQUEST2)
+			old_request = false;
+
+	}
+
+	if (old_request) {
+		/* Handle a REQUEST (or override) */
+		u16 version = VSOCK_PROTO_INVALID;
+		if (vmci_transport_proto_to_notify_struct(
+			pending, &version, true))
+			err = vmci_transport_send_negotiate(pending, qp_size);
+		else
+			err = -EINVAL;
+
+	} else {
+		/* Handle a REQUEST2 (or override) */
+		int proto_int = pkt->proto;
+		int pos;
+		u16 active_proto_version = 0;
+
+		/* The list of possible protocols is the intersection of all
+		 * protocols the client supports ... plus all the protocols we
+		 * support.
+		 */
+		proto_int &= vmci_transport_new_proto_supported_versions();
+
+		/* We choose the highest possible protocol version and use that
+		 * one.
+		 */
+		pos = fls(proto_int);
+		if (pos) {
+			active_proto_version = (1 << (pos - 1));
+			if (vmci_transport_proto_to_notify_struct(
+				pending, &active_proto_version, false))
+				err = vmci_transport_send_negotiate2(pending,
+							qp_size,
+							active_proto_version);
+			else
+				err = -EINVAL;
+
+		} else {
+			err = -EINVAL;
+		}
+	}
+
+	if (err < 0) {
+		vmci_transport_send_reset(sk, pkt);
+		sock_put(pending);
+		err = vmci_transport_error_to_vsock_error(err);
+		goto out;
+	}
+
+	vsock_add_pending(sk, pending);
+	sk->sk_ack_backlog++;
+
+	pending->sk_state = SS_CONNECTING;
+	vmci_trans(vpending)->produce_size =
+		vmci_trans(vpending)->consume_size = qp_size;
+	vmci_trans(vpending)->queue_pair_size = qp_size;
+
+	vmci_trans(vpending)->notify_ops->process_request(pending);
+
+	/* We might never receive another message for this socket and it's not
+	 * connected to any process, so we have to ensure it gets cleaned up
+	 * ourself.  Our delayed work function will take care of that.  Note
+	 * that we do not ever cancel this function since we have few
+	 * guarantees about its state when calling cancel_delayed_work().
+	 * Instead we hold a reference on the socket for that function and make
+	 * it capable of handling cases where it needs to do nothing but
+	 * release that reference.
+	 */
+	vpending->listener = sk;
+	sock_hold(sk);
+	sock_hold(pending);
+	INIT_DELAYED_WORK(&vpending->dwork, vsock_pending_work);
+	schedule_delayed_work(&vpending->dwork, HZ);
+
+out:
+	return err;
+}
+
+static int
+vmci_transport_recv_connecting_server(struct sock *listener,
+				      struct sock *pending,
+				      struct vmci_transport_packet *pkt)
+{
+	struct vsock_sock *vpending;
+	struct vmci_handle handle;
+	struct vmci_qp *qpair;
+	bool is_local;
+	u32 flags;
+	u32 detach_sub_id;
+	int err;
+	int skerr;
+
+	vpending = vsock_sk(pending);
+	detach_sub_id = VMCI_INVALID_ID;
+
+	switch (pkt->type) {
+	case VMCI_TRANSPORT_PACKET_TYPE_OFFER:
+		if (vmci_handle_is_invalid(pkt->u.handle)) {
+			vmci_transport_send_reset(pending, pkt);
+			skerr = EPROTO;
+			err = -EINVAL;
+			goto destroy;
+		}
+		break;
+	default:
+		/* Close and cleanup the connection. */
+		vmci_transport_send_reset(pending, pkt);
+		skerr = EPROTO;
+		err = pkt->type == VMCI_TRANSPORT_PACKET_TYPE_RST ? 0 : -EINVAL;
+		goto destroy;
+	}
+
+	/* In order to complete the connection we need to attach to the offered
+	 * queue pair and send an attach notification.  We also subscribe to the
+	 * detach event so we know when our peer goes away, and we do that
+	 * before attaching so we don't miss an event.  If all this succeeds,
+	 * we update our state and wakeup anything waiting in accept() for a
+	 * connection.
+	 */
+
+	/* We don't care about attach since we ensure the other side has
+	 * attached by specifying the ATTACH_ONLY flag below.
+	 */
+	err = vmci_event_subscribe(VMCI_EVENT_QP_PEER_DETACH,
+				   vmci_transport_peer_detach_cb,
+				   pending, &detach_sub_id);
+	if (err < VMCI_SUCCESS) {
+		vmci_transport_send_reset(pending, pkt);
+		err = vmci_transport_error_to_vsock_error(err);
+		skerr = -err;
+		goto destroy;
+	}
+
+	vmci_trans(vpending)->detach_sub_id = detach_sub_id;
+
+	/* Now attach to the queue pair the client created. */
+	handle = pkt->u.handle;
+
+	/* vpending->local_addr always has a context id so we do not need to
+	 * worry about VMADDR_CID_ANY in this case.
+	 */
+	is_local =
+	    vpending->remote_addr.svm_cid == vpending->local_addr.svm_cid;
+	flags = VMCI_QPFLAG_ATTACH_ONLY;
+	flags |= is_local ? VMCI_QPFLAG_LOCAL : 0;
+
+	err = vmci_transport_queue_pair_alloc(
+					&qpair,
+					&handle,
+					vmci_trans(vpending)->produce_size,
+					vmci_trans(vpending)->consume_size,
+					pkt->dg.src.context,
+					flags,
+					vmci_transport_is_trusted(
+						vpending,
+						vpending->remote_addr.svm_cid));
+	if (err < 0) {
+		vmci_transport_send_reset(pending, pkt);
+		skerr = -err;
+		goto destroy;
+	}
+
+	vmci_trans(vpending)->qp_handle = handle;
+	vmci_trans(vpending)->qpair = qpair;
+
+	/* When we send the attach message, we must be ready to handle incoming
+	 * control messages on the newly connected socket. So we move the
+	 * pending socket to the connected state before sending the attach
+	 * message. Otherwise, an incoming packet triggered by the attach being
+	 * received by the peer may be processed concurrently with what happens
+	 * below after sending the attach message, and that incoming packet
+	 * will find the listening socket instead of the (currently) pending
+	 * socket. Note that enqueueing the socket increments the reference
+	 * count, so even if a reset comes before the connection is accepted,
+	 * the socket will be valid until it is removed from the queue.
+	 *
+	 * If we fail sending the attach below, we remove the socket from the
+	 * connected list and move the socket to SS_UNCONNECTED before
+	 * releasing the lock, so a pending slow path processing of an incoming
+	 * packet will not see the socket in the connected state in that case.
+	 */
+	pending->sk_state = SS_CONNECTED;
+
+	vsock_insert_connected(vpending);
+
+	/* Notify our peer of our attach. */
+	err = vmci_transport_send_attach(pending, handle);
+	if (err < 0) {
+		vsock_remove_connected(vpending);
+		pr_err("Could not send attach\n");
+		vmci_transport_send_reset(pending, pkt);
+		err = vmci_transport_error_to_vsock_error(err);
+		skerr = -err;
+		goto destroy;
+	}
+
+	/* We have a connection. Move the now connected socket from the
+	 * listener's pending list to the accept queue so callers of accept()
+	 * can find it.
+	 */
+	vsock_remove_pending(listener, pending);
+	vsock_enqueue_accept(listener, pending);
+
+	/* Callers of accept() will be be waiting on the listening socket, not
+	 * the pending socket.
+	 */
+	listener->sk_state_change(listener);
+
+	return 0;
+
+destroy:
+	pending->sk_err = skerr;
+	pending->sk_state = SS_UNCONNECTED;
+	/* As long as we drop our reference, all necessary cleanup will handle
+	 * when the cleanup function drops its reference and our destruct
+	 * implementation is called.  Note that since the listen handler will
+	 * remove pending from the pending list upon our failure, the cleanup
+	 * function won't drop the additional reference, which is why we do it
+	 * here.
+	 */
+	sock_put(pending);
+
+	return err;
+}
+
+static int
+vmci_transport_recv_connecting_client(struct sock *sk,
+				      struct vmci_transport_packet *pkt)
+{
+	struct vsock_sock *vsk;
+	int err;
+	int skerr;
+
+	vsk = vsock_sk(sk);
+
+	switch (pkt->type) {
+	case VMCI_TRANSPORT_PACKET_TYPE_ATTACH:
+		if (vmci_handle_is_invalid(pkt->u.handle) ||
+		    !vmci_handle_is_equal(pkt->u.handle,
+					  vmci_trans(vsk)->qp_handle)) {
+			skerr = EPROTO;
+			err = -EINVAL;
+			goto destroy;
+		}
+
+		/* Signify the socket is connected and wakeup the waiter in
+		 * connect(). Also place the socket in the connected table for
+		 * accounting (it can already be found since it's in the bound
+		 * table).
+		 */
+		sk->sk_state = SS_CONNECTED;
+		sk->sk_socket->state = SS_CONNECTED;
+		vsock_insert_connected(vsk);
+		sk->sk_state_change(sk);
+
+		break;
+	case VMCI_TRANSPORT_PACKET_TYPE_NEGOTIATE:
+	case VMCI_TRANSPORT_PACKET_TYPE_NEGOTIATE2:
+		if (pkt->u.size == 0
+		    || pkt->dg.src.context != vsk->remote_addr.svm_cid
+		    || pkt->src_port != vsk->remote_addr.svm_port
+		    || !vmci_handle_is_invalid(vmci_trans(vsk)->qp_handle)
+		    || vmci_trans(vsk)->qpair
+		    || vmci_trans(vsk)->produce_size != 0
+		    || vmci_trans(vsk)->consume_size != 0
+		    || vmci_trans(vsk)->attach_sub_id != VMCI_INVALID_ID
+		    || vmci_trans(vsk)->detach_sub_id != VMCI_INVALID_ID) {
+			skerr = EPROTO;
+			err = -EINVAL;
+
+			goto destroy;
+		}
+
+		err = vmci_transport_recv_connecting_client_negotiate(sk, pkt);
+		if (err) {
+			skerr = -err;
+			goto destroy;
+		}
+
+		break;
+	case VMCI_TRANSPORT_PACKET_TYPE_INVALID:
+		err = vmci_transport_recv_connecting_client_invalid(sk, pkt);
+		if (err) {
+			skerr = -err;
+			goto destroy;
+		}
+
+		break;
+	case VMCI_TRANSPORT_PACKET_TYPE_RST:
+		/* Older versions of the linux code (WS 6.5 / ESX 4.0) used to
+		 * continue processing here after they sent an INVALID packet.
+		 * This meant that we got a RST after the INVALID. We ignore a
+		 * RST after an INVALID. The common code doesn't send the RST
+		 * ... so we can hang if an old version of the common code
+		 * fails between getting a REQUEST and sending an OFFER back.
+		 * Not much we can do about it... except hope that it doesn't
+		 * happen.
+		 */
+		if (vsk->ignore_connecting_rst) {
+			vsk->ignore_connecting_rst = false;
+		} else {
+			skerr = ECONNRESET;
+			err = 0;
+			goto destroy;
+		}
+
+		break;
+	default:
+		/* Close and cleanup the connection. */
+		skerr = EPROTO;
+		err = -EINVAL;
+		goto destroy;
+	}
+
+	return 0;
+
+destroy:
+	vmci_transport_send_reset(sk, pkt);
+
+	sk->sk_state = SS_UNCONNECTED;
+	sk->sk_err = skerr;
+	sk->sk_error_report(sk);
+	return err;
+}
+
+static int vmci_transport_recv_connecting_client_negotiate(
+					struct sock *sk,
+					struct vmci_transport_packet *pkt)
+{
+	int err;
+	struct vsock_sock *vsk;
+	struct vmci_handle handle;
+	struct vmci_qp *qpair;
+	u32 attach_sub_id;
+	u32 detach_sub_id;
+	bool is_local;
+	u32 flags;
+	bool old_proto = true;
+	bool old_pkt_proto;
+	u16 version;
+
+	vsk = vsock_sk(sk);
+	handle = VMCI_INVALID_HANDLE;
+	attach_sub_id = VMCI_INVALID_ID;
+	detach_sub_id = VMCI_INVALID_ID;
+
+	/* If we have gotten here then we should be past the point where old
+	 * linux vsock could have sent the bogus rst.
+	 */
+	vsk->sent_request = false;
+	vsk->ignore_connecting_rst = false;
+
+	/* Verify that we're OK with the proposed queue pair size */
+	if (pkt->u.size < vmci_trans(vsk)->queue_pair_min_size ||
+	    pkt->u.size > vmci_trans(vsk)->queue_pair_max_size) {
+		err = -EINVAL;
+		goto destroy;
+	}
+
+	/* At this point we know the CID the peer is using to talk to us. */
+
+	if (vsk->local_addr.svm_cid == VMADDR_CID_ANY)
+		vsk->local_addr.svm_cid = pkt->dg.dst.context;
+
+	/* Setup the notify ops to be the highest supported version that both
+	 * the server and the client support.
+	 */
+
+	if (vmci_transport_old_proto_override(&old_pkt_proto)) {
+		old_proto = old_pkt_proto;
+	} else {
+		if (pkt->type == VMCI_TRANSPORT_PACKET_TYPE_NEGOTIATE)
+			old_proto = true;
+		else if (pkt->type == VMCI_TRANSPORT_PACKET_TYPE_NEGOTIATE2)
+			old_proto = false;
+
+	}
+
+	if (old_proto)
+		version = VSOCK_PROTO_INVALID;
+	else
+		version = pkt->proto;
+
+	if (!vmci_transport_proto_to_notify_struct(sk, &version, old_proto)) {
+		err = -EINVAL;
+		goto destroy;
+	}
+
+	/* Subscribe to attach and detach events first.
+	 *
+	 * XXX We attach once for each queue pair created for now so it is easy
+	 * to find the socket (it's provided), but later we should only
+	 * subscribe once and add a way to lookup sockets by queue pair handle.
+	 */
+	err = vmci_event_subscribe(VMCI_EVENT_QP_PEER_ATTACH,
+				   vmci_transport_peer_attach_cb,
+				   sk, &attach_sub_id);
+	if (err < VMCI_SUCCESS) {
+		err = vmci_transport_error_to_vsock_error(err);
+		goto destroy;
+	}
+
+	err = vmci_event_subscribe(VMCI_EVENT_QP_PEER_DETACH,
+				   vmci_transport_peer_detach_cb,
+				   sk, &detach_sub_id);
+	if (err < VMCI_SUCCESS) {
+		err = vmci_transport_error_to_vsock_error(err);
+		goto destroy;
+	}
+
+	/* Make VMCI select the handle for us. */
+	handle = VMCI_INVALID_HANDLE;
+	is_local = vsk->remote_addr.svm_cid == vsk->local_addr.svm_cid;
+	flags = is_local ? VMCI_QPFLAG_LOCAL : 0;
+
+	err = vmci_transport_queue_pair_alloc(&qpair,
+					      &handle,
+					      pkt->u.size,
+					      pkt->u.size,
+					      vsk->remote_addr.svm_cid,
+					      flags,
+					      vmci_transport_is_trusted(
+						  vsk,
+						  vsk->
+						  remote_addr.svm_cid));
+	if (err < 0)
+		goto destroy;
+
+	err = vmci_transport_send_qp_offer(sk, handle);
+	if (err < 0) {
+		err = vmci_transport_error_to_vsock_error(err);
+		goto destroy;
+	}
+
+	vmci_trans(vsk)->qp_handle = handle;
+	vmci_trans(vsk)->qpair = qpair;
+
+	vmci_trans(vsk)->produce_size = vmci_trans(vsk)->consume_size =
+		pkt->u.size;
+
+	vmci_trans(vsk)->attach_sub_id = attach_sub_id;
+	vmci_trans(vsk)->detach_sub_id = detach_sub_id;
+
+	vmci_trans(vsk)->notify_ops->process_negotiate(sk);
+
+	return 0;
+
+destroy:
+	if (attach_sub_id != VMCI_INVALID_ID)
+		vmci_event_unsubscribe(attach_sub_id);
+
+	if (detach_sub_id != VMCI_INVALID_ID)
+		vmci_event_unsubscribe(detach_sub_id);
+
+	if (!vmci_handle_is_invalid(handle))
+		vmci_qpair_detach(&qpair);
+
+	return err;
+}
+
+static int
+vmci_transport_recv_connecting_client_invalid(struct sock *sk,
+					      struct vmci_transport_packet *pkt)
+{
+	int err = 0;
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	if (vsk->sent_request) {
+		vsk->sent_request = false;
+		vsk->ignore_connecting_rst = true;
+
+		err = vmci_transport_send_conn_request(
+			sk, vmci_trans(vsk)->queue_pair_size);
+		if (err < 0)
+			err = vmci_transport_error_to_vsock_error(err);
+		else
+			err = 0;
+
+	}
+
+	return err;
+}
+
+static int vmci_transport_recv_connected(struct sock *sk,
+					 struct vmci_transport_packet *pkt)
+{
+	struct vsock_sock *vsk;
+	bool pkt_processed = false;
+
+	/* In cases where we are closing the connection, it's sufficient to
+	 * mark the state change (and maybe error) and wake up any waiting
+	 * threads. Since this is a connected socket, it's owned by a user
+	 * process and will be cleaned up when the failure is passed back on
+	 * the current or next system call.  Our system call implementations
+	 * must therefore check for error and state changes on entry and when
+	 * being awoken.
+	 */
+	switch (pkt->type) {
+	case VMCI_TRANSPORT_PACKET_TYPE_SHUTDOWN:
+		if (pkt->u.mode) {
+			vsk = vsock_sk(sk);
+
+			vsk->peer_shutdown |= pkt->u.mode;
+			sk->sk_state_change(sk);
+		}
+		break;
+
+	case VMCI_TRANSPORT_PACKET_TYPE_RST:
+		vsk = vsock_sk(sk);
+		/* It is possible that we sent our peer a message (e.g a
+		 * WAITING_READ) right before we got notified that the peer had
+		 * detached. If that happens then we can get a RST pkt back
+		 * from our peer even though there is data available for us to
+		 * read. In that case, don't shutdown the socket completely but
+		 * instead allow the local client to finish reading data off
+		 * the queuepair. Always treat a RST pkt in connected mode like
+		 * a clean shutdown.
+		 */
+		sock_set_flag(sk, SOCK_DONE);
+		vsk->peer_shutdown = SHUTDOWN_MASK;
+		if (vsock_stream_has_data(vsk) <= 0)
+			sk->sk_state = SS_DISCONNECTING;
+
+		sk->sk_state_change(sk);
+		break;
+
+	default:
+		vsk = vsock_sk(sk);
+		vmci_trans(vsk)->notify_ops->handle_notify_pkt(
+				sk, pkt, false, NULL, NULL,
+				&pkt_processed);
+		if (!pkt_processed)
+			return -EINVAL;
+
+		break;
+	}
+
+	return 0;
+}
+
+static int vmci_transport_socket_init(struct vsock_sock *vsk,
+				      struct vsock_sock *psk)
+{
+	vsk->trans = kmalloc(sizeof(struct vmci_transport), GFP_KERNEL);
+	if (!vsk->trans)
+		return -ENOMEM;
+
+	vmci_trans(vsk)->dg_handle = VMCI_INVALID_HANDLE;
+	vmci_trans(vsk)->qp_handle = VMCI_INVALID_HANDLE;
+	vmci_trans(vsk)->qpair = NULL;
+	vmci_trans(vsk)->produce_size = vmci_trans(vsk)->consume_size = 0;
+	vmci_trans(vsk)->attach_sub_id = vmci_trans(vsk)->detach_sub_id =
+		VMCI_INVALID_ID;
+	vmci_trans(vsk)->notify_ops = NULL;
+	if (psk) {
+		vmci_trans(vsk)->queue_pair_size =
+			vmci_trans(psk)->queue_pair_size;
+		vmci_trans(vsk)->queue_pair_min_size =
+			vmci_trans(psk)->queue_pair_min_size;
+		vmci_trans(vsk)->queue_pair_max_size =
+			vmci_trans(psk)->queue_pair_max_size;
+	} else {
+		vmci_trans(vsk)->queue_pair_size =
+			VMCI_TRANSPORT_DEFAULT_QP_SIZE;
+		vmci_trans(vsk)->queue_pair_min_size =
+			 VMCI_TRANSPORT_DEFAULT_QP_SIZE_MIN;
+		vmci_trans(vsk)->queue_pair_max_size =
+			VMCI_TRANSPORT_DEFAULT_QP_SIZE_MAX;
+	}
+
+	return 0;
+}
+
+static void vmci_transport_destruct(struct vsock_sock *vsk)
+{
+	if (vmci_trans(vsk)->attach_sub_id != VMCI_INVALID_ID) {
+		vmci_event_unsubscribe(vmci_trans(vsk)->attach_sub_id);
+		vmci_trans(vsk)->attach_sub_id = VMCI_INVALID_ID;
+	}
+
+	if (vmci_trans(vsk)->detach_sub_id != VMCI_INVALID_ID) {
+		vmci_event_unsubscribe(vmci_trans(vsk)->detach_sub_id);
+		vmci_trans(vsk)->detach_sub_id = VMCI_INVALID_ID;
+	}
+
+	if (!vmci_handle_is_invalid(vmci_trans(vsk)->qp_handle)) {
+		vmci_qpair_detach(&vmci_trans(vsk)->qpair);
+		vmci_trans(vsk)->qp_handle = VMCI_INVALID_HANDLE;
+		vmci_trans(vsk)->produce_size = 0;
+		vmci_trans(vsk)->consume_size = 0;
+	}
+
+	if (vmci_trans(vsk)->notify_ops)
+		vmci_trans(vsk)->notify_ops->socket_destruct(vsk);
+
+	kfree(vsk->trans);
+	vsk->trans = NULL;
+}
+
+static void vmci_transport_release(struct vsock_sock *vsk)
+{
+	if (!vmci_handle_is_invalid(vmci_trans(vsk)->dg_handle)) {
+		vmci_datagram_destroy_handle(vmci_trans(vsk)->dg_handle);
+		vmci_trans(vsk)->dg_handle = VMCI_INVALID_HANDLE;
+	}
+}
+
+static int vmci_transport_dgram_bind(struct vsock_sock *vsk,
+				     struct sockaddr_vm *addr)
+{
+	u32 port;
+	u32 flags;
+	int err;
+
+	/* VMCI will select a resource ID for us if we provide
+	 * VMCI_INVALID_ID.
+	 */
+	port = addr->svm_port == VMADDR_PORT_ANY ?
+			VMCI_INVALID_ID : addr->svm_port;
+
+	if (port <= LAST_RESERVED_PORT && !capable(CAP_NET_BIND_SERVICE))
+		return -EACCES;
+
+	flags = addr->svm_cid == VMADDR_CID_ANY ?
+				VMCI_FLAG_ANYCID_DG_HND : 0;
+
+	err = vmci_transport_datagram_create_hnd(port, flags,
+						 vmci_transport_recv_dgram_cb,
+						 &vsk->sk,
+						 &vmci_trans(vsk)->dg_handle);
+	if (err < VMCI_SUCCESS)
+		return vmci_transport_error_to_vsock_error(err);
+	vsock_addr_init(&vsk->local_addr, addr->svm_cid,
+			vmci_trans(vsk)->dg_handle.resource);
+
+	return 0;
+}
+
+static int vmci_transport_dgram_enqueue(
+	struct vsock_sock *vsk,
+	struct sockaddr_vm *remote_addr,
+	struct iovec *iov,
+	size_t len)
+{
+	int err;
+	struct vmci_datagram *dg;
+
+	if (len > VMCI_MAX_DG_PAYLOAD_SIZE)
+		return -EMSGSIZE;
+
+	if (!vmci_transport_allow_dgram(vsk, remote_addr->svm_cid))
+		return -EPERM;
+
+	/* Allocate a buffer for the user's message and our packet header. */
+	dg = kmalloc(len + sizeof(*dg), GFP_KERNEL);
+	if (!dg)
+		return -ENOMEM;
+
+	memcpy_fromiovec(VMCI_DG_PAYLOAD(dg), iov, len);
+
+	dg->dst = vmci_make_handle(remote_addr->svm_cid,
+				   remote_addr->svm_port);
+	dg->src = vmci_make_handle(vsk->local_addr.svm_cid,
+				   vsk->local_addr.svm_port);
+	dg->payload_size = len;
+
+	err = vmci_datagram_send(dg);
+	kfree(dg);
+	if (err < 0)
+		return vmci_transport_error_to_vsock_error(err);
+
+	return err - sizeof(*dg);
+}
+
+static int vmci_transport_dgram_dequeue(struct kiocb *kiocb,
+					struct vsock_sock *vsk,
+					struct msghdr *msg, size_t len,
+					int flags)
+{
+	int err;
+	int noblock;
+	struct vmci_datagram *dg;
+	size_t payload_len;
+	struct sk_buff *skb;
+
+	noblock = flags & MSG_DONTWAIT;
+
+	if (flags & MSG_OOB || flags & MSG_ERRQUEUE)
+		return -EOPNOTSUPP;
+
+	/* Retrieve the head sk_buff from the socket's receive queue. */
+	err = 0;
+	skb = skb_recv_datagram(&vsk->sk, flags, noblock, &err);
+	if (err)
+		return err;
+
+	if (!skb)
+		return -EAGAIN;
+
+	dg = (struct vmci_datagram *)skb->data;
+	if (!dg)
+		/* err is 0, meaning we read zero bytes. */
+		goto out;
+
+	payload_len = dg->payload_size;
+	/* Ensure the sk_buff matches the payload size claimed in the packet. */
+	if (payload_len != skb->len - sizeof(*dg)) {
+		err = -EINVAL;
+		goto out;
+	}
+
+	if (payload_len > len) {
+		payload_len = len;
+		msg->msg_flags |= MSG_TRUNC;
+	}
+
+	/* Place the datagram payload in the user's iovec. */
+	err = skb_copy_datagram_iovec(skb, sizeof(*dg), msg->msg_iov,
+		payload_len);
+	if (err)
+		goto out;
+
+	msg->msg_namelen = 0;
+	if (msg->msg_name) {
+		struct sockaddr_vm *vm_addr;
+
+		/* Provide the address of the sender. */
+		vm_addr = (struct sockaddr_vm *)msg->msg_name;
+		vsock_addr_init(vm_addr, dg->src.context, dg->src.resource);
+		msg->msg_namelen = sizeof(*vm_addr);
+	}
+	err = payload_len;
+
+out:
+	skb_free_datagram(&vsk->sk, skb);
+	return err;
+}
+
+static bool vmci_transport_dgram_allow(u32 cid, u32 port)
+{
+	if (cid == VMADDR_CID_HYPERVISOR) {
+		/* Registrations of PBRPC Servers do not modify VMX/Hypervisor
+		 * state and are allowed.
+		 */
+		return port == VMCI_UNITY_PBRPC_REGISTER;
+	}
+
+	return true;
+}
+
+static int vmci_transport_connect(struct vsock_sock *vsk)
+{
+	int err;
+	bool old_pkt_proto = false;
+	struct sock *sk = &vsk->sk;
+
+	if (vmci_transport_old_proto_override(&old_pkt_proto) &&
+		old_pkt_proto) {
+		err = vmci_transport_send_conn_request(
+			sk, vmci_trans(vsk)->queue_pair_size);
+		if (err < 0) {
+			sk->sk_state = SS_UNCONNECTED;
+			return err;
+		}
+	} else {
+		int supported_proto_versions =
+			vmci_transport_new_proto_supported_versions();
+		err = vmci_transport_send_conn_request2(
+				sk, vmci_trans(vsk)->queue_pair_size,
+				supported_proto_versions);
+		if (err < 0) {
+			sk->sk_state = SS_UNCONNECTED;
+			return err;
+		}
+
+		vsk->sent_request = true;
+	}
+
+	return err;
+}
+
+static ssize_t vmci_transport_stream_dequeue(
+	struct vsock_sock *vsk,
+	struct iovec *iov,
+	size_t len,
+	int flags)
+{
+	if (flags & MSG_PEEK)
+		return vmci_qpair_peekv(vmci_trans(vsk)->qpair, iov, len, 0);
+	else
+		return vmci_qpair_dequev(vmci_trans(vsk)->qpair, iov, len, 0);
+}
+
+static ssize_t vmci_transport_stream_enqueue(
+	struct vsock_sock *vsk,
+	struct iovec *iov,
+	size_t len)
+{
+	return vmci_qpair_enquev(vmci_trans(vsk)->qpair, iov, len, 0);
+}
+
+static s64 vmci_transport_stream_has_data(struct vsock_sock *vsk)
+{
+	return vmci_qpair_consume_buf_ready(vmci_trans(vsk)->qpair);
+}
+
+static s64 vmci_transport_stream_has_space(struct vsock_sock *vsk)
+{
+	return vmci_qpair_produce_free_space(vmci_trans(vsk)->qpair);
+}
+
+static u64 vmci_transport_stream_rcvhiwat(struct vsock_sock *vsk)
+{
+	return vmci_trans(vsk)->consume_size;
+}
+
+static bool vmci_transport_stream_is_active(struct vsock_sock *vsk)
+{
+	return !vmci_handle_is_invalid(vmci_trans(vsk)->qp_handle);
+}
+
+static u64 vmci_transport_get_buffer_size(struct vsock_sock *vsk)
+{
+	return vmci_trans(vsk)->queue_pair_size;
+}
+
+static u64 vmci_transport_get_min_buffer_size(struct vsock_sock *vsk)
+{
+	return vmci_trans(vsk)->queue_pair_min_size;
+}
+
+static u64 vmci_transport_get_max_buffer_size(struct vsock_sock *vsk)
+{
+	return vmci_trans(vsk)->queue_pair_max_size;
+}
+
+static void vmci_transport_set_buffer_size(struct vsock_sock *vsk, u64 val)
+{
+	if (val < vmci_trans(vsk)->queue_pair_min_size)
+		vmci_trans(vsk)->queue_pair_min_size = val;
+	if (val > vmci_trans(vsk)->queue_pair_max_size)
+		vmci_trans(vsk)->queue_pair_max_size = val;
+	vmci_trans(vsk)->queue_pair_size = val;
+}
+
+static void vmci_transport_set_min_buffer_size(struct vsock_sock *vsk,
+					       u64 val)
+{
+	if (val > vmci_trans(vsk)->queue_pair_size)
+		vmci_trans(vsk)->queue_pair_size = val;
+	vmci_trans(vsk)->queue_pair_min_size = val;
+}
+
+static void vmci_transport_set_max_buffer_size(struct vsock_sock *vsk,
+					       u64 val)
+{
+	if (val < vmci_trans(vsk)->queue_pair_size)
+		vmci_trans(vsk)->queue_pair_size = val;
+	vmci_trans(vsk)->queue_pair_max_size = val;
+}
+
+static int vmci_transport_notify_poll_in(
+	struct vsock_sock *vsk,
+	size_t target,
+	bool *data_ready_now)
+{
+	return vmci_trans(vsk)->notify_ops->poll_in(
+			&vsk->sk, target, data_ready_now);
+}
+
+static int vmci_transport_notify_poll_out(
+	struct vsock_sock *vsk,
+	size_t target,
+	bool *space_available_now)
+{
+	return vmci_trans(vsk)->notify_ops->poll_out(
+			&vsk->sk, target, space_available_now);
+}
+
+static int vmci_transport_notify_recv_init(
+	struct vsock_sock *vsk,
+	size_t target,
+	struct vsock_transport_recv_notify_data *data)
+{
+	return vmci_trans(vsk)->notify_ops->recv_init(
+			&vsk->sk, target,
+			(struct vmci_transport_recv_notify_data *)data);
+}
+
+static int vmci_transport_notify_recv_pre_block(
+	struct vsock_sock *vsk,
+	size_t target,
+	struct vsock_transport_recv_notify_data *data)
+{
+	return vmci_trans(vsk)->notify_ops->recv_pre_block(
+			&vsk->sk, target,
+			(struct vmci_transport_recv_notify_data *)data);
+}
+
+static int vmci_transport_notify_recv_pre_dequeue(
+	struct vsock_sock *vsk,
+	size_t target,
+	struct vsock_transport_recv_notify_data *data)
+{
+	return vmci_trans(vsk)->notify_ops->recv_pre_dequeue(
+			&vsk->sk, target,
+			(struct vmci_transport_recv_notify_data *)data);
+}
+
+static int vmci_transport_notify_recv_post_dequeue(
+	struct vsock_sock *vsk,
+	size_t target,
+	ssize_t copied,
+	bool data_read,
+	struct vsock_transport_recv_notify_data *data)
+{
+	return vmci_trans(vsk)->notify_ops->recv_post_dequeue(
+			&vsk->sk, target, copied, data_read,
+			(struct vmci_transport_recv_notify_data *)data);
+}
+
+static int vmci_transport_notify_send_init(
+	struct vsock_sock *vsk,
+	struct vsock_transport_send_notify_data *data)
+{
+	return vmci_trans(vsk)->notify_ops->send_init(
+			&vsk->sk,
+			(struct vmci_transport_send_notify_data *)data);
+}
+
+static int vmci_transport_notify_send_pre_block(
+	struct vsock_sock *vsk,
+	struct vsock_transport_send_notify_data *data)
+{
+	return vmci_trans(vsk)->notify_ops->send_pre_block(
+			&vsk->sk,
+			(struct vmci_transport_send_notify_data *)data);
+}
+
+static int vmci_transport_notify_send_pre_enqueue(
+	struct vsock_sock *vsk,
+	struct vsock_transport_send_notify_data *data)
+{
+	return vmci_trans(vsk)->notify_ops->send_pre_enqueue(
+			&vsk->sk,
+			(struct vmci_transport_send_notify_data *)data);
+}
+
+static int vmci_transport_notify_send_post_enqueue(
+	struct vsock_sock *vsk,
+	ssize_t written,
+	struct vsock_transport_send_notify_data *data)
+{
+	return vmci_trans(vsk)->notify_ops->send_post_enqueue(
+			&vsk->sk, written,
+			(struct vmci_transport_send_notify_data *)data);
+}
+
+static bool vmci_transport_old_proto_override(bool *old_pkt_proto)
+{
+	if (PROTOCOL_OVERRIDE != -1) {
+		if (PROTOCOL_OVERRIDE == 0)
+			*old_pkt_proto = true;
+		else
+			*old_pkt_proto = false;
+
+		pr_info("Proto override in use\n");
+		return true;
+	}
+
+	return false;
+}
+
+static bool vmci_transport_proto_to_notify_struct(struct sock *sk,
+						  u16 *proto,
+						  bool old_pkt_proto)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	if (old_pkt_proto) {
+		if (*proto != VSOCK_PROTO_INVALID) {
+			pr_err("Can't set both an old and new protocol\n");
+			return false;
+		}
+		vmci_trans(vsk)->notify_ops = &vmci_transport_notify_pkt_ops;
+		goto exit;
+	}
+
+	switch (*proto) {
+	case VSOCK_PROTO_PKT_ON_NOTIFY:
+		vmci_trans(vsk)->notify_ops =
+			&vmci_transport_notify_pkt_q_state_ops;
+		break;
+	default:
+		pr_err("Unknown notify protocol version\n");
+		return false;
+	}
+
+exit:
+	vmci_trans(vsk)->notify_ops->socket_init(sk);
+	return true;
+}
+
+static u16 vmci_transport_new_proto_supported_versions(void)
+{
+	if (PROTOCOL_OVERRIDE != -1)
+		return PROTOCOL_OVERRIDE;
+
+	return VSOCK_PROTO_ALL_SUPPORTED;
+}
+
+static u32 vmci_transport_get_local_cid(void)
+{
+	return vmci_get_context_id();
+}
+
+static struct vsock_transport vmci_transport = {
+	.init = vmci_transport_socket_init,
+	.destruct = vmci_transport_destruct,
+	.release = vmci_transport_release,
+	.connect = vmci_transport_connect,
+	.dgram_bind = vmci_transport_dgram_bind,
+	.dgram_dequeue = vmci_transport_dgram_dequeue,
+	.dgram_enqueue = vmci_transport_dgram_enqueue,
+	.dgram_allow = vmci_transport_dgram_allow,
+	.stream_dequeue = vmci_transport_stream_dequeue,
+	.stream_enqueue = vmci_transport_stream_enqueue,
+	.stream_has_data = vmci_transport_stream_has_data,
+	.stream_has_space = vmci_transport_stream_has_space,
+	.stream_rcvhiwat = vmci_transport_stream_rcvhiwat,
+	.stream_is_active = vmci_transport_stream_is_active,
+	.stream_allow = vmci_transport_stream_allow,
+	.notify_poll_in = vmci_transport_notify_poll_in,
+	.notify_poll_out = vmci_transport_notify_poll_out,
+	.notify_recv_init = vmci_transport_notify_recv_init,
+	.notify_recv_pre_block = vmci_transport_notify_recv_pre_block,
+	.notify_recv_pre_dequeue = vmci_transport_notify_recv_pre_dequeue,
+	.notify_recv_post_dequeue = vmci_transport_notify_recv_post_dequeue,
+	.notify_send_init = vmci_transport_notify_send_init,
+	.notify_send_pre_block = vmci_transport_notify_send_pre_block,
+	.notify_send_pre_enqueue = vmci_transport_notify_send_pre_enqueue,
+	.notify_send_post_enqueue = vmci_transport_notify_send_post_enqueue,
+	.shutdown = vmci_transport_shutdown,
+	.set_buffer_size = vmci_transport_set_buffer_size,
+	.set_min_buffer_size = vmci_transport_set_min_buffer_size,
+	.set_max_buffer_size = vmci_transport_set_max_buffer_size,
+	.get_buffer_size = vmci_transport_get_buffer_size,
+	.get_min_buffer_size = vmci_transport_get_min_buffer_size,
+	.get_max_buffer_size = vmci_transport_get_max_buffer_size,
+	.get_local_cid = vmci_transport_get_local_cid,
+};
+
+static int __init vmci_transport_init(void)
+{
+	int err;
+
+	/* Create the datagram handle that we will use to send and receive all
+	 * VSocket control messages for this context.
+	 */
+	err = vmci_transport_datagram_create_hnd(VMCI_TRANSPORT_PACKET_RID,
+						 VMCI_FLAG_ANYCID_DG_HND,
+						 vmci_transport_recv_stream_cb,
+						 NULL,
+						 &vmci_transport_stream_handle);
+	if (err < VMCI_SUCCESS) {
+		pr_err("Unable to create datagram handle. (%d)\n", err);
+		return vmci_transport_error_to_vsock_error(err);
+	}
+
+	err = vmci_event_subscribe(VMCI_EVENT_QP_RESUMED,
+				   vmci_transport_qp_resumed_cb,
+				   NULL, &vmci_transport_qp_resumed_sub_id);
+	if (err < VMCI_SUCCESS) {
+		pr_err("Unable to subscribe to resumed event. (%d)\n", err);
+		err = vmci_transport_error_to_vsock_error(err);
+		vmci_transport_qp_resumed_sub_id = VMCI_INVALID_ID;
+		goto err_destroy_stream_handle;
+	}
+
+	err = vsock_core_init(&vmci_transport);
+	if (err < 0)
+		goto err_unsubscribe;
+
+	return 0;
+
+err_unsubscribe:
+	vmci_event_unsubscribe(vmci_transport_qp_resumed_sub_id);
+err_destroy_stream_handle:
+	vmci_datagram_destroy_handle(vmci_transport_stream_handle);
+	return err;
+}
+module_init(vmci_transport_init);
+
+static void __exit vmci_transport_exit(void)
+{
+	if (!vmci_handle_is_invalid(vmci_transport_stream_handle)) {
+		if (vmci_datagram_destroy_handle(
+			vmci_transport_stream_handle) != VMCI_SUCCESS)
+			pr_err("Couldn't destroy datagram handle\n");
+		vmci_transport_stream_handle = VMCI_INVALID_HANDLE;
+	}
+
+	if (vmci_transport_qp_resumed_sub_id != VMCI_INVALID_ID) {
+		vmci_event_unsubscribe(vmci_transport_qp_resumed_sub_id);
+		vmci_transport_qp_resumed_sub_id = VMCI_INVALID_ID;
+	}
+
+	vsock_core_exit();
+}
+module_exit(vmci_transport_exit);
+
+MODULE_AUTHOR("VMware, Inc.");
+MODULE_DESCRIPTION("VMCI transport for Virtual Sockets");
+MODULE_LICENSE("GPL v2");
+MODULE_ALIAS("vmware_vsock");
+MODULE_ALIAS_NETPROTO(PF_VSOCK);

net/vmw_vsock/vmci_transport.h

@@ -0,0 +1,139 @@
+/*
+ * VMware vSockets Driver
+ *
+ * Copyright (C) 2013 VMware, Inc. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation version 2 and no later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ */
+
+#ifndef _VMCI_TRANSPORT_H_
+#define _VMCI_TRANSPORT_H_
+
+#include <linux/vmw_vmci_defs.h>
+#include <linux/vmw_vmci_api.h>
+
+#include "vsock_addr.h"
+#include "af_vsock.h"
+
+/* If the packet format changes in a release then this should change too. */
+#define VMCI_TRANSPORT_PACKET_VERSION 1
+
+/* The resource ID on which control packets are sent. */
+#define VMCI_TRANSPORT_PACKET_RID 1
+
+#define VSOCK_PROTO_INVALID        0
+#define VSOCK_PROTO_PKT_ON_NOTIFY (1 << 0)
+#define VSOCK_PROTO_ALL_SUPPORTED (VSOCK_PROTO_PKT_ON_NOTIFY)
+
+#define vmci_trans(_vsk) ((struct vmci_transport *)((_vsk)->trans))
+
+enum vmci_transport_packet_type {
+	VMCI_TRANSPORT_PACKET_TYPE_INVALID = 0,
+	VMCI_TRANSPORT_PACKET_TYPE_REQUEST,
+	VMCI_TRANSPORT_PACKET_TYPE_NEGOTIATE,
+	VMCI_TRANSPORT_PACKET_TYPE_OFFER,
+	VMCI_TRANSPORT_PACKET_TYPE_ATTACH,
+	VMCI_TRANSPORT_PACKET_TYPE_WROTE,
+	VMCI_TRANSPORT_PACKET_TYPE_READ,
+	VMCI_TRANSPORT_PACKET_TYPE_RST,
+	VMCI_TRANSPORT_PACKET_TYPE_SHUTDOWN,
+	VMCI_TRANSPORT_PACKET_TYPE_WAITING_WRITE,
+	VMCI_TRANSPORT_PACKET_TYPE_WAITING_READ,
+	VMCI_TRANSPORT_PACKET_TYPE_REQUEST2,
+	VMCI_TRANSPORT_PACKET_TYPE_NEGOTIATE2,
+	VMCI_TRANSPORT_PACKET_TYPE_MAX
+};
+
+struct vmci_transport_waiting_info {
+	u64 generation;
+	u64 offset;
+};
+
+/* Control packet type for STREAM sockets.  DGRAMs have no control packets nor
+ * special packet header for data packets, they are just raw VMCI DGRAM
+ * messages.  For STREAMs, control packets are sent over the control channel
+ * while data is written and read directly from queue pairs with no packet
+ * format.
+ */
+struct vmci_transport_packet {
+	struct vmci_datagram dg;
+	u8 version;
+	u8 type;
+	u16 proto;
+	u32 src_port;
+	u32 dst_port;
+	u32 _reserved2;
+	union {
+		u64 size;
+		u64 mode;
+		struct vmci_handle handle;
+		struct vmci_transport_waiting_info wait;
+	} u;
+};
+
+struct vmci_transport_notify_pkt {
+	u64 write_notify_window;
+	u64 write_notify_min_window;
+	bool peer_waiting_read;
+	bool peer_waiting_write;
+	bool peer_waiting_write_detected;
+	bool sent_waiting_read;
+	bool sent_waiting_write;
+	struct vmci_transport_waiting_info peer_waiting_read_info;
+	struct vmci_transport_waiting_info peer_waiting_write_info;
+	u64 produce_q_generation;
+	u64 consume_q_generation;
+};
+
+struct vmci_transport_notify_pkt_q_state {
+	u64 write_notify_window;
+	u64 write_notify_min_window;
+	bool peer_waiting_write;
+	bool peer_waiting_write_detected;
+};
+
+union vmci_transport_notify {
+	struct vmci_transport_notify_pkt pkt;
+	struct vmci_transport_notify_pkt_q_state pkt_q_state;
+};
+
+/* Our transport-specific data. */
+struct vmci_transport {
+	/* For DGRAMs. */
+	struct vmci_handle dg_handle;
+	/* For STREAMs. */
+	struct vmci_handle qp_handle;
+	struct vmci_qp *qpair;
+	u64 produce_size;
+	u64 consume_size;
+	u64 queue_pair_size;
+	u64 queue_pair_min_size;
+	u64 queue_pair_max_size;
+	u32 attach_sub_id;
+	u32 detach_sub_id;
+	union vmci_transport_notify notify;
+	struct vmci_transport_notify_ops *notify_ops;
+};
+
+int vmci_transport_register(void);
+void vmci_transport_unregister(void);
+
+int vmci_transport_send_wrote_bh(struct sockaddr_vm *dst,
+				 struct sockaddr_vm *src);
+int vmci_transport_send_read_bh(struct sockaddr_vm *dst,
+				struct sockaddr_vm *src);
+int vmci_transport_send_wrote(struct sock *sk);
+int vmci_transport_send_read(struct sock *sk);
+int vmci_transport_send_waiting_write(struct sock *sk,
+				      struct vmci_transport_waiting_info *wait);
+int vmci_transport_send_waiting_read(struct sock *sk,
+				     struct vmci_transport_waiting_info *wait);
+
+#endif

net/vmw_vsock/vmci_transport_notify.c

@@ -0,0 +1,680 @@
+/*
+ * VMware vSockets Driver
+ *
+ * Copyright (C) 2009-2013 VMware, Inc. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation version 2 and no later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ */
+
+#include <linux/types.h>
+#include <linux/socket.h>
+#include <linux/stddef.h>
+#include <net/sock.h>
+
+#include "vmci_transport_notify.h"
+
+#define PKT_FIELD(vsk, field_name) (vmci_trans(vsk)->notify.pkt.field_name)
+
+static bool vmci_transport_notify_waiting_write(struct vsock_sock *vsk)
+{
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+	bool retval;
+	u64 notify_limit;
+
+	if (!PKT_FIELD(vsk, peer_waiting_write))
+		return false;
+
+#ifdef VSOCK_OPTIMIZATION_FLOW_CONTROL
+	/* When the sender blocks, we take that as a sign that the sender is
+	 * faster than the receiver. To reduce the transmit rate of the sender,
+	 * we delay the sending of the read notification by decreasing the
+	 * write_notify_window. The notification is delayed until the number of
+	 * bytes used in the queue drops below the write_notify_window.
+	 */
+
+	if (!PKT_FIELD(vsk, peer_waiting_write_detected)) {
+		PKT_FIELD(vsk, peer_waiting_write_detected) = true;
+		if (PKT_FIELD(vsk, write_notify_window) < PAGE_SIZE) {
+			PKT_FIELD(vsk, write_notify_window) =
+			    PKT_FIELD(vsk, write_notify_min_window);
+		} else {
+			PKT_FIELD(vsk, write_notify_window) -= PAGE_SIZE;
+			if (PKT_FIELD(vsk, write_notify_window) <
+			    PKT_FIELD(vsk, write_notify_min_window))
+				PKT_FIELD(vsk, write_notify_window) =
+				    PKT_FIELD(vsk, write_notify_min_window);
+
+		}
+	}
+	notify_limit = vmci_trans(vsk)->consume_size -
+		PKT_FIELD(vsk, write_notify_window);
+#else
+	notify_limit = 0;
+#endif
+
+	/* For now we ignore the wait information and just see if the free
+	 * space exceeds the notify limit.  Note that improving this function
+	 * to be more intelligent will not require a protocol change and will
+	 * retain compatibility between endpoints with mixed versions of this
+	 * function.
+	 *
+	 * The notify_limit is used to delay notifications in the case where
+	 * flow control is enabled. Below the test is expressed in terms of
+	 * free space in the queue: if free_space > ConsumeSize -
+	 * write_notify_window then notify An alternate way of expressing this
+	 * is to rewrite the expression to use the data ready in the receive
+	 * queue: if write_notify_window > bufferReady then notify as
+	 * free_space == ConsumeSize - bufferReady.
+	 */
+	retval = vmci_qpair_consume_free_space(vmci_trans(vsk)->qpair) >
+		notify_limit;
+#ifdef VSOCK_OPTIMIZATION_FLOW_CONTROL
+	if (retval) {
+		/*
+		 * Once we notify the peer, we reset the detected flag so the
+		 * next wait will again cause a decrease in the window size.
+		 */
+
+		PKT_FIELD(vsk, peer_waiting_write_detected) = false;
+	}
+#endif
+	return retval;
+#else
+	return true;
+#endif
+}
+
+static bool vmci_transport_notify_waiting_read(struct vsock_sock *vsk)
+{
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+	if (!PKT_FIELD(vsk, peer_waiting_read))
+		return false;
+
+	/* For now we ignore the wait information and just see if there is any
+	 * data for our peer to read.  Note that improving this function to be
+	 * more intelligent will not require a protocol change and will retain
+	 * compatibility between endpoints with mixed versions of this
+	 * function.
+	 */
+	return vmci_qpair_produce_buf_ready(vmci_trans(vsk)->qpair) > 0;
+#else
+	return true;
+#endif
+}
+
+static void
+vmci_transport_handle_waiting_read(struct sock *sk,
+				   struct vmci_transport_packet *pkt,
+				   bool bottom_half,
+				   struct sockaddr_vm *dst,
+				   struct sockaddr_vm *src)
+{
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+	struct vsock_sock *vsk;
+
+	vsk = vsock_sk(sk);
+
+	PKT_FIELD(vsk, peer_waiting_read) = true;
+	memcpy(&PKT_FIELD(vsk, peer_waiting_read_info), &pkt->u.wait,
+	       sizeof(PKT_FIELD(vsk, peer_waiting_read_info)));
+
+	if (vmci_transport_notify_waiting_read(vsk)) {
+		bool sent;
+
+		if (bottom_half)
+			sent = vmci_transport_send_wrote_bh(dst, src) > 0;
+		else
+			sent = vmci_transport_send_wrote(sk) > 0;
+
+		if (sent)
+			PKT_FIELD(vsk, peer_waiting_read) = false;
+	}
+#endif
+}
+
+static void
+vmci_transport_handle_waiting_write(struct sock *sk,
+				    struct vmci_transport_packet *pkt,
+				    bool bottom_half,
+				    struct sockaddr_vm *dst,
+				    struct sockaddr_vm *src)
+{
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+	struct vsock_sock *vsk;
+
+	vsk = vsock_sk(sk);
+
+	PKT_FIELD(vsk, peer_waiting_write) = true;
+	memcpy(&PKT_FIELD(vsk, peer_waiting_write_info), &pkt->u.wait,
+	       sizeof(PKT_FIELD(vsk, peer_waiting_write_info)));
+
+	if (vmci_transport_notify_waiting_write(vsk)) {
+		bool sent;
+
+		if (bottom_half)
+			sent = vmci_transport_send_read_bh(dst, src) > 0;
+		else
+			sent = vmci_transport_send_read(sk) > 0;
+
+		if (sent)
+			PKT_FIELD(vsk, peer_waiting_write) = false;
+	}
+#endif
+}
+
+static void
+vmci_transport_handle_read(struct sock *sk,
+			   struct vmci_transport_packet *pkt,
+			   bool bottom_half,
+			   struct sockaddr_vm *dst, struct sockaddr_vm *src)
+{
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+	struct vsock_sock *vsk;
+
+	vsk = vsock_sk(sk);
+	PKT_FIELD(vsk, sent_waiting_write) = false;
+#endif
+
+	sk->sk_write_space(sk);
+}
+
+static bool send_waiting_read(struct sock *sk, u64 room_needed)
+{
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+	struct vsock_sock *vsk;
+	struct vmci_transport_waiting_info waiting_info;
+	u64 tail;
+	u64 head;
+	u64 room_left;
+	bool ret;
+
+	vsk = vsock_sk(sk);
+
+	if (PKT_FIELD(vsk, sent_waiting_read))
+		return true;
+
+	if (PKT_FIELD(vsk, write_notify_window) <
+			vmci_trans(vsk)->consume_size)
+		PKT_FIELD(vsk, write_notify_window) =
+		    min(PKT_FIELD(vsk, write_notify_window) + PAGE_SIZE,
+			vmci_trans(vsk)->consume_size);
+
+	vmci_qpair_get_consume_indexes(vmci_trans(vsk)->qpair, &tail, &head);
+	room_left = vmci_trans(vsk)->consume_size - head;
+	if (room_needed >= room_left) {
+		waiting_info.offset = room_needed - room_left;
+		waiting_info.generation =
+		    PKT_FIELD(vsk, consume_q_generation) + 1;
+	} else {
+		waiting_info.offset = head + room_needed;
+		waiting_info.generation = PKT_FIELD(vsk, consume_q_generation);
+	}
+
+	ret = vmci_transport_send_waiting_read(sk, &waiting_info) > 0;
+	if (ret)
+		PKT_FIELD(vsk, sent_waiting_read) = true;
+
+	return ret;
+#else
+	return true;
+#endif
+}
+
+static bool send_waiting_write(struct sock *sk, u64 room_needed)
+{
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+	struct vsock_sock *vsk;
+	struct vmci_transport_waiting_info waiting_info;
+	u64 tail;
+	u64 head;
+	u64 room_left;
+	bool ret;
+
+	vsk = vsock_sk(sk);
+
+	if (PKT_FIELD(vsk, sent_waiting_write))
+		return true;
+
+	vmci_qpair_get_produce_indexes(vmci_trans(vsk)->qpair, &tail, &head);
+	room_left = vmci_trans(vsk)->produce_size - tail;
+	if (room_needed + 1 >= room_left) {
+		/* Wraps around to current generation. */
+		waiting_info.offset = room_needed + 1 - room_left;
+		waiting_info.generation = PKT_FIELD(vsk, produce_q_generation);
+	} else {
+		waiting_info.offset = tail + room_needed + 1;
+		waiting_info.generation =
+		    PKT_FIELD(vsk, produce_q_generation) - 1;
+	}
+
+	ret = vmci_transport_send_waiting_write(sk, &waiting_info) > 0;
+	if (ret)
+		PKT_FIELD(vsk, sent_waiting_write) = true;
+
+	return ret;
+#else
+	return true;
+#endif
+}
+
+static int vmci_transport_send_read_notification(struct sock *sk)
+{
+	struct vsock_sock *vsk;
+	bool sent_read;
+	unsigned int retries;
+	int err;
+
+	vsk = vsock_sk(sk);
+	sent_read = false;
+	retries = 0;
+	err = 0;
+
+	if (vmci_transport_notify_waiting_write(vsk)) {
+		/* Notify the peer that we have read, retrying the send on
+		 * failure up to our maximum value.  XXX For now we just log
+		 * the failure, but later we should schedule a work item to
+		 * handle the resend until it succeeds.  That would require
+		 * keeping track of work items in the vsk and cleaning them up
+		 * upon socket close.
+		 */
+		while (!(vsk->peer_shutdown & RCV_SHUTDOWN) &&
+		       !sent_read &&
+		       retries < VMCI_TRANSPORT_MAX_DGRAM_RESENDS) {
+			err = vmci_transport_send_read(sk);
+			if (err >= 0)
+				sent_read = true;
+
+			retries++;
+		}
+
+		if (retries >= VMCI_TRANSPORT_MAX_DGRAM_RESENDS)
+			pr_err("%p unable to send read notify to peer\n", sk);
+		else
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+			PKT_FIELD(vsk, peer_waiting_write) = false;
+#endif
+
+	}
+	return err;
+}
+
+static void
+vmci_transport_handle_wrote(struct sock *sk,
+			    struct vmci_transport_packet *pkt,
+			    bool bottom_half,
+			    struct sockaddr_vm *dst, struct sockaddr_vm *src)
+{
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+	struct vsock_sock *vsk = vsock_sk(sk);
+	PKT_FIELD(vsk, sent_waiting_read) = false;
+#endif
+	sk->sk_data_ready(sk, 0);
+}
+
+static void vmci_transport_notify_pkt_socket_init(struct sock *sk)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	PKT_FIELD(vsk, write_notify_window) = PAGE_SIZE;
+	PKT_FIELD(vsk, write_notify_min_window) = PAGE_SIZE;
+	PKT_FIELD(vsk, peer_waiting_read) = false;
+	PKT_FIELD(vsk, peer_waiting_write) = false;
+	PKT_FIELD(vsk, peer_waiting_write_detected) = false;
+	PKT_FIELD(vsk, sent_waiting_read) = false;
+	PKT_FIELD(vsk, sent_waiting_write) = false;
+	PKT_FIELD(vsk, produce_q_generation) = 0;
+	PKT_FIELD(vsk, consume_q_generation) = 0;
+
+	memset(&PKT_FIELD(vsk, peer_waiting_read_info), 0,
+	       sizeof(PKT_FIELD(vsk, peer_waiting_read_info)));
+	memset(&PKT_FIELD(vsk, peer_waiting_write_info), 0,
+	       sizeof(PKT_FIELD(vsk, peer_waiting_write_info)));
+}
+
+static void vmci_transport_notify_pkt_socket_destruct(struct vsock_sock *vsk)
+{
+}
+
+static int
+vmci_transport_notify_pkt_poll_in(struct sock *sk,
+				  size_t target, bool *data_ready_now)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	if (vsock_stream_has_data(vsk)) {
+		*data_ready_now = true;
+	} else {
+		/* We can't read right now because there is nothing in the
+		 * queue. Ask for notifications when there is something to
+		 * read.
+		 */
+		if (sk->sk_state == SS_CONNECTED) {
+			if (!send_waiting_read(sk, 1))
+				return -1;
+
+		}
+		*data_ready_now = false;
+	}
+
+	return 0;
+}
+
+static int
+vmci_transport_notify_pkt_poll_out(struct sock *sk,
+				   size_t target, bool *space_avail_now)
+{
+	s64 produce_q_free_space;
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	produce_q_free_space = vsock_stream_has_space(vsk);
+	if (produce_q_free_space > 0) {
+		*space_avail_now = true;
+		return 0;
+	} else if (produce_q_free_space == 0) {
+		/* This is a connected socket but we can't currently send data.
+		 * Notify the peer that we are waiting if the queue is full. We
+		 * only send a waiting write if the queue is full because
+		 * otherwise we end up in an infinite WAITING_WRITE, READ,
+		 * WAITING_WRITE, READ, etc. loop. Treat failing to send the
+		 * notification as a socket error, passing that back through
+		 * the mask.
+		 */
+		if (!send_waiting_write(sk, 1))
+			return -1;
+
+		*space_avail_now = false;
+	}
+
+	return 0;
+}
+
+static int
+vmci_transport_notify_pkt_recv_init(
+			struct sock *sk,
+			size_t target,
+			struct vmci_transport_recv_notify_data *data)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+#ifdef VSOCK_OPTIMIZATION_WAITING_NOTIFY
+	data->consume_head = 0;
+	data->produce_tail = 0;
+#ifdef VSOCK_OPTIMIZATION_FLOW_CONTROL
+	data->notify_on_block = false;
+
+	if (PKT_FIELD(vsk, write_notify_min_window) < target + 1) {
+		PKT_FIELD(vsk, write_notify_min_window) = target + 1;
+		if (PKT_FIELD(vsk, write_notify_window) <
+		    PKT_FIELD(vsk, write_notify_min_window)) {
+			/* If the current window is smaller than the new
+			 * minimal window size, we need to reevaluate whether
+			 * we need to notify the sender. If the number of ready
+			 * bytes are smaller than the new window, we need to
+			 * send a notification to the sender before we block.
+			 */
+
+			PKT_FIELD(vsk, write_notify_window) =
+			    PKT_FIELD(vsk, write_notify_min_window);
+			data->notify_on_block = true;
+		}
+	}
+#endif
+#endif
+
+	return 0;
+}
+
+static int
+vmci_transport_notify_pkt_recv_pre_block(
+				struct sock *sk,
+				size_t target,
+				struct vmci_transport_recv_notify_data *data)
+{
+	int err = 0;
+
+	/* Notify our peer that we are waiting for data to read. */
+	if (!send_waiting_read(sk, target)) {
+		err = -EHOSTUNREACH;
+		return err;
+	}
+#ifdef VSOCK_OPTIMIZATION_FLOW_CONTROL
+	if (data->notify_on_block) {
+		err = vmci_transport_send_read_notification(sk);
+		if (err < 0)
+			return err;
+
+		data->notify_on_block = false;
+	}
+#endif
+
+	return err;
+}
+
+static int
+vmci_transport_notify_pkt_recv_pre_dequeue(
+				struct sock *sk,
+				size_t target,
+				struct vmci_transport_recv_notify_data *data)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	/* Now consume up to len bytes from the queue.  Note that since we have
+	 * the socket locked we should copy at least ready bytes.
+	 */
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+	vmci_qpair_get_consume_indexes(vmci_trans(vsk)->qpair,
+				       &data->produce_tail,
+				       &data->consume_head);
+#endif
+
+	return 0;
+}
+
+static int
+vmci_transport_notify_pkt_recv_post_dequeue(
+				struct sock *sk,
+				size_t target,
+				ssize_t copied,
+				bool data_read,
+				struct vmci_transport_recv_notify_data *data)
+{
+	struct vsock_sock *vsk;
+	int err;
+
+	vsk = vsock_sk(sk);
+	err = 0;
+
+	if (data_read) {
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+		/* Detect a wrap-around to maintain queue generation.  Note
+		 * that this is safe since we hold the socket lock across the
+		 * two queue pair operations.
+		 */
+		if (copied >=
+			vmci_trans(vsk)->consume_size - data->consume_head)
+			PKT_FIELD(vsk, consume_q_generation)++;
+#endif
+
+		err = vmci_transport_send_read_notification(sk);
+		if (err < 0)
+			return err;
+
+	}
+	return err;
+}
+
+static int
+vmci_transport_notify_pkt_send_init(
+			struct sock *sk,
+			struct vmci_transport_send_notify_data *data)
+{
+#ifdef VSOCK_OPTIMIZATION_WAITING_NOTIFY
+	data->consume_head = 0;
+	data->produce_tail = 0;
+#endif
+
+	return 0;
+}
+
+static int
+vmci_transport_notify_pkt_send_pre_block(
+				struct sock *sk,
+				struct vmci_transport_send_notify_data *data)
+{
+	/* Notify our peer that we are waiting for room to write. */
+	if (!send_waiting_write(sk, 1))
+		return -EHOSTUNREACH;
+
+	return 0;
+}
+
+static int
+vmci_transport_notify_pkt_send_pre_enqueue(
+				struct sock *sk,
+				struct vmci_transport_send_notify_data *data)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+	vmci_qpair_get_produce_indexes(vmci_trans(vsk)->qpair,
+				       &data->produce_tail,
+				       &data->consume_head);
+#endif
+
+	return 0;
+}
+
+static int
+vmci_transport_notify_pkt_send_post_enqueue(
+				struct sock *sk,
+				ssize_t written,
+				struct vmci_transport_send_notify_data *data)
+{
+	int err = 0;
+	struct vsock_sock *vsk;
+	bool sent_wrote = false;
+	int retries = 0;
+
+	vsk = vsock_sk(sk);
+
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+	/* Detect a wrap-around to maintain queue generation.  Note that this
+	 * is safe since we hold the socket lock across the two queue pair
+	 * operations.
+	 */
+	if (written >= vmci_trans(vsk)->produce_size - data->produce_tail)
+		PKT_FIELD(vsk, produce_q_generation)++;
+
+#endif
+
+	if (vmci_transport_notify_waiting_read(vsk)) {
+		/* Notify the peer that we have written, retrying the send on
+		 * failure up to our maximum value. See the XXX comment for the
+		 * corresponding piece of code in StreamRecvmsg() for potential
+		 * improvements.
+		 */
+		while (!(vsk->peer_shutdown & RCV_SHUTDOWN) &&
+		       !sent_wrote &&
+		       retries < VMCI_TRANSPORT_MAX_DGRAM_RESENDS) {
+			err = vmci_transport_send_wrote(sk);
+			if (err >= 0)
+				sent_wrote = true;
+
+			retries++;
+		}
+
+		if (retries >= VMCI_TRANSPORT_MAX_DGRAM_RESENDS) {
+			pr_err("%p unable to send wrote notify to peer\n", sk);
+			return err;
+		} else {
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+			PKT_FIELD(vsk, peer_waiting_read) = false;
+#endif
+		}
+	}
+	return err;
+}
+
+static void
+vmci_transport_notify_pkt_handle_pkt(
+			struct sock *sk,
+			struct vmci_transport_packet *pkt,
+			bool bottom_half,
+			struct sockaddr_vm *dst,
+			struct sockaddr_vm *src, bool *pkt_processed)
+{
+	bool processed = false;
+
+	switch (pkt->type) {
+	case VMCI_TRANSPORT_PACKET_TYPE_WROTE:
+		vmci_transport_handle_wrote(sk, pkt, bottom_half, dst, src);
+		processed = true;
+		break;
+	case VMCI_TRANSPORT_PACKET_TYPE_READ:
+		vmci_transport_handle_read(sk, pkt, bottom_half, dst, src);
+		processed = true;
+		break;
+	case VMCI_TRANSPORT_PACKET_TYPE_WAITING_WRITE:
+		vmci_transport_handle_waiting_write(sk, pkt, bottom_half,
+						    dst, src);
+		processed = true;
+		break;
+
+	case VMCI_TRANSPORT_PACKET_TYPE_WAITING_READ:
+		vmci_transport_handle_waiting_read(sk, pkt, bottom_half,
+						   dst, src);
+		processed = true;
+		break;
+	}
+
+	if (pkt_processed)
+		*pkt_processed = processed;
+}
+
+static void vmci_transport_notify_pkt_process_request(struct sock *sk)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	PKT_FIELD(vsk, write_notify_window) = vmci_trans(vsk)->consume_size;
+	if (vmci_trans(vsk)->consume_size <
+		PKT_FIELD(vsk, write_notify_min_window))
+		PKT_FIELD(vsk, write_notify_min_window) =
+			vmci_trans(vsk)->consume_size;
+}
+
+static void vmci_transport_notify_pkt_process_negotiate(struct sock *sk)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	PKT_FIELD(vsk, write_notify_window) = vmci_trans(vsk)->consume_size;
+	if (vmci_trans(vsk)->consume_size <
+		PKT_FIELD(vsk, write_notify_min_window))
+		PKT_FIELD(vsk, write_notify_min_window) =
+			vmci_trans(vsk)->consume_size;
+}
+
+/* Socket control packet based operations. */
+struct vmci_transport_notify_ops vmci_transport_notify_pkt_ops = {
+	vmci_transport_notify_pkt_socket_init,
+	vmci_transport_notify_pkt_socket_destruct,
+	vmci_transport_notify_pkt_poll_in,
+	vmci_transport_notify_pkt_poll_out,
+	vmci_transport_notify_pkt_handle_pkt,
+	vmci_transport_notify_pkt_recv_init,
+	vmci_transport_notify_pkt_recv_pre_block,
+	vmci_transport_notify_pkt_recv_pre_dequeue,
+	vmci_transport_notify_pkt_recv_post_dequeue,
+	vmci_transport_notify_pkt_send_init,
+	vmci_transport_notify_pkt_send_pre_block,
+	vmci_transport_notify_pkt_send_pre_enqueue,
+	vmci_transport_notify_pkt_send_post_enqueue,
+	vmci_transport_notify_pkt_process_request,
+	vmci_transport_notify_pkt_process_negotiate,
+};

net/vmw_vsock/vmci_transport_notify.h

@@ -0,0 +1,83 @@
+/*
+ * VMware vSockets Driver
+ *
+ * Copyright (C) 2009-2013 VMware, Inc. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation version 2 and no later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ */
+
+#ifndef __VMCI_TRANSPORT_NOTIFY_H__
+#define __VMCI_TRANSPORT_NOTIFY_H__
+
+#include <linux/types.h>
+#include <linux/vmw_vmci_defs.h>
+#include <linux/vmw_vmci_api.h>
+#include <linux/vm_sockets.h>
+
+#include "vmci_transport.h"
+
+/* Comment this out to compare with old protocol. */
+#define VSOCK_OPTIMIZATION_WAITING_NOTIFY 1
+#if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY)
+/* Comment this out to remove flow control for "new" protocol */
+#define VSOCK_OPTIMIZATION_FLOW_CONTROL 1
+#endif
+
+#define VMCI_TRANSPORT_MAX_DGRAM_RESENDS       10
+
+struct vmci_transport_recv_notify_data {
+	u64 consume_head;
+	u64 produce_tail;
+	bool notify_on_block;
+};
+
+struct vmci_transport_send_notify_data {
+	u64 consume_head;
+	u64 produce_tail;
+};
+
+/* Socket notification callbacks. */
+struct vmci_transport_notify_ops {
+	void (*socket_init) (struct sock *sk);
+	void (*socket_destruct) (struct vsock_sock *vsk);
+	int (*poll_in) (struct sock *sk, size_t target,
+			  bool *data_ready_now);
+	int (*poll_out) (struct sock *sk, size_t target,
+			   bool *space_avail_now);
+	void (*handle_notify_pkt) (struct sock *sk,
+				   struct vmci_transport_packet *pkt,
+				   bool bottom_half, struct sockaddr_vm *dst,
+				   struct sockaddr_vm *src,
+				   bool *pkt_processed);
+	int (*recv_init) (struct sock *sk, size_t target,
+			  struct vmci_transport_recv_notify_data *data);
+	int (*recv_pre_block) (struct sock *sk, size_t target,
+			       struct vmci_transport_recv_notify_data *data);
+	int (*recv_pre_dequeue) (struct sock *sk, size_t target,
+				 struct vmci_transport_recv_notify_data *data);
+	int (*recv_post_dequeue) (struct sock *sk, size_t target,
+				  ssize_t copied, bool data_read,
+				  struct vmci_transport_recv_notify_data *data);
+	int (*send_init) (struct sock *sk,
+			  struct vmci_transport_send_notify_data *data);
+	int (*send_pre_block) (struct sock *sk,
+			       struct vmci_transport_send_notify_data *data);
+	int (*send_pre_enqueue) (struct sock *sk,
+				 struct vmci_transport_send_notify_data *data);
+	int (*send_post_enqueue) (struct sock *sk, ssize_t written,
+				  struct vmci_transport_send_notify_data *data);
+	void (*process_request) (struct sock *sk);
+	void (*process_negotiate) (struct sock *sk);
+};
+
+extern struct vmci_transport_notify_ops vmci_transport_notify_pkt_ops;
+extern struct vmci_transport_notify_ops vmci_transport_notify_pkt_q_state_ops;
+
+#endif /* __VMCI_TRANSPORT_NOTIFY_H__ */

net/vmw_vsock/vmci_transport_notify_qstate.c

@@ -0,0 +1,438 @@
+/*
+ * VMware vSockets Driver
+ *
+ * Copyright (C) 2009-2013 VMware, Inc. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation version 2 and no later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ */
+
+#include <linux/types.h>
+#include <linux/socket.h>
+#include <linux/stddef.h>
+#include <net/sock.h>
+
+#include "vmci_transport_notify.h"
+
+#define PKT_FIELD(vsk, field_name) \
+	(vmci_trans(vsk)->notify.pkt_q_state.field_name)
+
+static bool vmci_transport_notify_waiting_write(struct vsock_sock *vsk)
+{
+	bool retval;
+	u64 notify_limit;
+
+	if (!PKT_FIELD(vsk, peer_waiting_write))
+		return false;
+
+	/* When the sender blocks, we take that as a sign that the sender is
+	 * faster than the receiver. To reduce the transmit rate of the sender,
+	 * we delay the sending of the read notification by decreasing the
+	 * write_notify_window. The notification is delayed until the number of
+	 * bytes used in the queue drops below the write_notify_window.
+	 */
+
+	if (!PKT_FIELD(vsk, peer_waiting_write_detected)) {
+		PKT_FIELD(vsk, peer_waiting_write_detected) = true;
+		if (PKT_FIELD(vsk, write_notify_window) < PAGE_SIZE) {
+			PKT_FIELD(vsk, write_notify_window) =
+			    PKT_FIELD(vsk, write_notify_min_window);
+		} else {
+			PKT_FIELD(vsk, write_notify_window) -= PAGE_SIZE;
+			if (PKT_FIELD(vsk, write_notify_window) <
+			    PKT_FIELD(vsk, write_notify_min_window))
+				PKT_FIELD(vsk, write_notify_window) =
+				    PKT_FIELD(vsk, write_notify_min_window);
+
+		}
+	}
+	notify_limit = vmci_trans(vsk)->consume_size -
+		PKT_FIELD(vsk, write_notify_window);
+
+	/* The notify_limit is used to delay notifications in the case where
+	 * flow control is enabled. Below the test is expressed in terms of
+	 * free space in the queue: if free_space > ConsumeSize -
+	 * write_notify_window then notify An alternate way of expressing this
+	 * is to rewrite the expression to use the data ready in the receive
+	 * queue: if write_notify_window > bufferReady then notify as
+	 * free_space == ConsumeSize - bufferReady.
+	 */
+
+	retval = vmci_qpair_consume_free_space(vmci_trans(vsk)->qpair) >
+		notify_limit;
+
+	if (retval) {
+		/* Once we notify the peer, we reset the detected flag so the
+		 * next wait will again cause a decrease in the window size.
+		 */
+
+		PKT_FIELD(vsk, peer_waiting_write_detected) = false;
+	}
+	return retval;
+}
+
+static void
+vmci_transport_handle_read(struct sock *sk,
+			   struct vmci_transport_packet *pkt,
+			   bool bottom_half,
+			   struct sockaddr_vm *dst, struct sockaddr_vm *src)
+{
+	sk->sk_write_space(sk);
+}
+
+static void
+vmci_transport_handle_wrote(struct sock *sk,
+			    struct vmci_transport_packet *pkt,
+			    bool bottom_half,
+			    struct sockaddr_vm *dst, struct sockaddr_vm *src)
+{
+	sk->sk_data_ready(sk, 0);
+}
+
+static void vsock_block_update_write_window(struct sock *sk)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	if (PKT_FIELD(vsk, write_notify_window) < vmci_trans(vsk)->consume_size)
+		PKT_FIELD(vsk, write_notify_window) =
+		    min(PKT_FIELD(vsk, write_notify_window) + PAGE_SIZE,
+			vmci_trans(vsk)->consume_size);
+}
+
+static int vmci_transport_send_read_notification(struct sock *sk)
+{
+	struct vsock_sock *vsk;
+	bool sent_read;
+	unsigned int retries;
+	int err;
+
+	vsk = vsock_sk(sk);
+	sent_read = false;
+	retries = 0;
+	err = 0;
+
+	if (vmci_transport_notify_waiting_write(vsk)) {
+		/* Notify the peer that we have read, retrying the send on
+		 * failure up to our maximum value.  XXX For now we just log
+		 * the failure, but later we should schedule a work item to
+		 * handle the resend until it succeeds.  That would require
+		 * keeping track of work items in the vsk and cleaning them up
+		 * upon socket close.
+		 */
+		while (!(vsk->peer_shutdown & RCV_SHUTDOWN) &&
+		       !sent_read &&
+		       retries < VMCI_TRANSPORT_MAX_DGRAM_RESENDS) {
+			err = vmci_transport_send_read(sk);
+			if (err >= 0)
+				sent_read = true;
+
+			retries++;
+		}
+
+		if (retries >= VMCI_TRANSPORT_MAX_DGRAM_RESENDS && !sent_read)
+			pr_err("%p unable to send read notification to peer\n",
+			       sk);
+		else
+			PKT_FIELD(vsk, peer_waiting_write) = false;
+
+	}
+	return err;
+}
+
+static void vmci_transport_notify_pkt_socket_init(struct sock *sk)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	PKT_FIELD(vsk, write_notify_window) = PAGE_SIZE;
+	PKT_FIELD(vsk, write_notify_min_window) = PAGE_SIZE;
+	PKT_FIELD(vsk, peer_waiting_write) = false;
+	PKT_FIELD(vsk, peer_waiting_write_detected) = false;
+}
+
+static void vmci_transport_notify_pkt_socket_destruct(struct vsock_sock *vsk)
+{
+	PKT_FIELD(vsk, write_notify_window) = PAGE_SIZE;
+	PKT_FIELD(vsk, write_notify_min_window) = PAGE_SIZE;
+	PKT_FIELD(vsk, peer_waiting_write) = false;
+	PKT_FIELD(vsk, peer_waiting_write_detected) = false;
+}
+
+static int
+vmci_transport_notify_pkt_poll_in(struct sock *sk,
+				  size_t target, bool *data_ready_now)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	if (vsock_stream_has_data(vsk)) {
+		*data_ready_now = true;
+	} else {
+		/* We can't read right now because there is nothing in the
+		 * queue. Ask for notifications when there is something to
+		 * read.
+		 */
+		if (sk->sk_state == SS_CONNECTED)
+			vsock_block_update_write_window(sk);
+		*data_ready_now = false;
+	}
+
+	return 0;
+}
+
+static int
+vmci_transport_notify_pkt_poll_out(struct sock *sk,
+				   size_t target, bool *space_avail_now)
+{
+	s64 produce_q_free_space;
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	produce_q_free_space = vsock_stream_has_space(vsk);
+	if (produce_q_free_space > 0) {
+		*space_avail_now = true;
+		return 0;
+	} else if (produce_q_free_space == 0) {
+		/* This is a connected socket but we can't currently send data.
+		 * Nothing else to do.
+		 */
+		*space_avail_now = false;
+	}
+
+	return 0;
+}
+
+static int
+vmci_transport_notify_pkt_recv_init(
+				struct sock *sk,
+				size_t target,
+				struct vmci_transport_recv_notify_data *data)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	data->consume_head = 0;
+	data->produce_tail = 0;
+	data->notify_on_block = false;
+
+	if (PKT_FIELD(vsk, write_notify_min_window) < target + 1) {
+		PKT_FIELD(vsk, write_notify_min_window) = target + 1;
+		if (PKT_FIELD(vsk, write_notify_window) <
+		    PKT_FIELD(vsk, write_notify_min_window)) {
+			/* If the current window is smaller than the new
+			 * minimal window size, we need to reevaluate whether
+			 * we need to notify the sender. If the number of ready
+			 * bytes are smaller than the new window, we need to
+			 * send a notification to the sender before we block.
+			 */
+
+			PKT_FIELD(vsk, write_notify_window) =
+			    PKT_FIELD(vsk, write_notify_min_window);
+			data->notify_on_block = true;
+		}
+	}
+
+	return 0;
+}
+
+static int
+vmci_transport_notify_pkt_recv_pre_block(
+				struct sock *sk,
+				size_t target,
+				struct vmci_transport_recv_notify_data *data)
+{
+	int err = 0;
+
+	vsock_block_update_write_window(sk);
+
+	if (data->notify_on_block) {
+		err = vmci_transport_send_read_notification(sk);
+		if (err < 0)
+			return err;
+		data->notify_on_block = false;
+	}
+
+	return err;
+}
+
+static int
+vmci_transport_notify_pkt_recv_post_dequeue(
+				struct sock *sk,
+				size_t target,
+				ssize_t copied,
+				bool data_read,
+				struct vmci_transport_recv_notify_data *data)
+{
+	struct vsock_sock *vsk;
+	int err;
+	bool was_full = false;
+	u64 free_space;
+
+	vsk = vsock_sk(sk);
+	err = 0;
+
+	if (data_read) {
+		smp_mb();
+
+		free_space =
+			vmci_qpair_consume_free_space(vmci_trans(vsk)->qpair);
+		was_full = free_space == copied;
+
+		if (was_full)
+			PKT_FIELD(vsk, peer_waiting_write) = true;
+
+		err = vmci_transport_send_read_notification(sk);
+		if (err < 0)
+			return err;
+
+		/* See the comment in
+		 * vmci_transport_notify_pkt_send_post_enqueue().
+		 */
+		sk->sk_data_ready(sk, 0);
+	}
+
+	return err;
+}
+
+static int
+vmci_transport_notify_pkt_send_init(
+				struct sock *sk,
+				struct vmci_transport_send_notify_data *data)
+{
+	data->consume_head = 0;
+	data->produce_tail = 0;
+
+	return 0;
+}
+
+static int
+vmci_transport_notify_pkt_send_post_enqueue(
+				struct sock *sk,
+				ssize_t written,
+				struct vmci_transport_send_notify_data *data)
+{
+	int err = 0;
+	struct vsock_sock *vsk;
+	bool sent_wrote = false;
+	bool was_empty;
+	int retries = 0;
+
+	vsk = vsock_sk(sk);
+
+	smp_mb();
+
+	was_empty =
+		vmci_qpair_produce_buf_ready(vmci_trans(vsk)->qpair) == written;
+	if (was_empty) {
+		while (!(vsk->peer_shutdown & RCV_SHUTDOWN) &&
+		       !sent_wrote &&
+		       retries < VMCI_TRANSPORT_MAX_DGRAM_RESENDS) {
+			err = vmci_transport_send_wrote(sk);
+			if (err >= 0)
+				sent_wrote = true;
+
+			retries++;
+		}
+	}
+
+	if (retries >= VMCI_TRANSPORT_MAX_DGRAM_RESENDS && !sent_wrote) {
+		pr_err("%p unable to send wrote notification to peer\n",
+		       sk);
+		return err;
+	}
+
+	return err;
+}
+
+static void
+vmci_transport_notify_pkt_handle_pkt(
+				struct sock *sk,
+				struct vmci_transport_packet *pkt,
+				bool bottom_half,
+				struct sockaddr_vm *dst,
+				struct sockaddr_vm *src, bool *pkt_processed)
+{
+	bool processed = false;
+
+	switch (pkt->type) {
+	case VMCI_TRANSPORT_PACKET_TYPE_WROTE:
+		vmci_transport_handle_wrote(sk, pkt, bottom_half, dst, src);
+		processed = true;
+		break;
+	case VMCI_TRANSPORT_PACKET_TYPE_READ:
+		vmci_transport_handle_read(sk, pkt, bottom_half, dst, src);
+		processed = true;
+		break;
+	}
+
+	if (pkt_processed)
+		*pkt_processed = processed;
+}
+
+static void vmci_transport_notify_pkt_process_request(struct sock *sk)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	PKT_FIELD(vsk, write_notify_window) = vmci_trans(vsk)->consume_size;
+	if (vmci_trans(vsk)->consume_size <
+		PKT_FIELD(vsk, write_notify_min_window))
+		PKT_FIELD(vsk, write_notify_min_window) =
+			vmci_trans(vsk)->consume_size;
+}
+
+static void vmci_transport_notify_pkt_process_negotiate(struct sock *sk)
+{
+	struct vsock_sock *vsk = vsock_sk(sk);
+
+	PKT_FIELD(vsk, write_notify_window) = vmci_trans(vsk)->consume_size;
+	if (vmci_trans(vsk)->consume_size <
+		PKT_FIELD(vsk, write_notify_min_window))
+		PKT_FIELD(vsk, write_notify_min_window) =
+			vmci_trans(vsk)->consume_size;
+}
+
+static int
+vmci_transport_notify_pkt_recv_pre_dequeue(
+				struct sock *sk,
+				size_t target,
+				struct vmci_transport_recv_notify_data *data)
+{
+	return 0; /* NOP for QState. */
+}
+
+static int
+vmci_transport_notify_pkt_send_pre_block(
+				struct sock *sk,
+				struct vmci_transport_send_notify_data *data)
+{
+	return 0; /* NOP for QState. */
+}
+
+static int
+vmci_transport_notify_pkt_send_pre_enqueue(
+				struct sock *sk,
+				struct vmci_transport_send_notify_data *data)
+{
+	return 0; /* NOP for QState. */
+}
+
+/* Socket always on control packet based operations. */
+struct vmci_transport_notify_ops vmci_transport_notify_pkt_q_state_ops = {
+	vmci_transport_notify_pkt_socket_init,
+	vmci_transport_notify_pkt_socket_destruct,
+	vmci_transport_notify_pkt_poll_in,
+	vmci_transport_notify_pkt_poll_out,
+	vmci_transport_notify_pkt_handle_pkt,
+	vmci_transport_notify_pkt_recv_init,
+	vmci_transport_notify_pkt_recv_pre_block,
+	vmci_transport_notify_pkt_recv_pre_dequeue,
+	vmci_transport_notify_pkt_recv_post_dequeue,
+	vmci_transport_notify_pkt_send_init,
+	vmci_transport_notify_pkt_send_pre_block,
+	vmci_transport_notify_pkt_send_pre_enqueue,
+	vmci_transport_notify_pkt_send_post_enqueue,
+	vmci_transport_notify_pkt_process_request,
+	vmci_transport_notify_pkt_process_negotiate,
+};

net/vmw_vsock/vsock_addr.c

@@ -0,0 +1,86 @@
+/*
+ * VMware vSockets Driver
+ *
+ * Copyright (C) 2007-2012 VMware, Inc. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation version 2 and no later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ */
+
+#include <linux/types.h>
+#include <linux/socket.h>
+#include <linux/stddef.h>
+#include <net/sock.h>
+
+#include "vsock_addr.h"
+
+void vsock_addr_init(struct sockaddr_vm *addr, u32 cid, u32 port)
+{
+	memset(addr, 0, sizeof(*addr));
+	addr->svm_family = AF_VSOCK;
+	addr->svm_cid = cid;
+	addr->svm_port = port;
+}
+EXPORT_SYMBOL_GPL(vsock_addr_init);
+
+int vsock_addr_validate(const struct sockaddr_vm *addr)
+{
+	if (!addr)
+		return -EFAULT;
+
+	if (addr->svm_family != AF_VSOCK)
+		return -EAFNOSUPPORT;
+
+	if (addr->svm_zero[0] != 0)
+		return -EINVAL;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(vsock_addr_validate);
+
+bool vsock_addr_bound(const struct sockaddr_vm *addr)
+{
+	return addr->svm_port != VMADDR_PORT_ANY;
+}
+EXPORT_SYMBOL_GPL(vsock_addr_bound);
+
+void vsock_addr_unbind(struct sockaddr_vm *addr)
+{
+	vsock_addr_init(addr, VMADDR_CID_ANY, VMADDR_PORT_ANY);
+}
+EXPORT_SYMBOL_GPL(vsock_addr_unbind);
+
+bool vsock_addr_equals_addr(const struct sockaddr_vm *addr,
+			    const struct sockaddr_vm *other)
+{
+	return addr->svm_cid == other->svm_cid &&
+		addr->svm_port == other->svm_port;
+}
+EXPORT_SYMBOL_GPL(vsock_addr_equals_addr);
+
+bool vsock_addr_equals_addr_any(const struct sockaddr_vm *addr,
+				const struct sockaddr_vm *other)
+{
+	return (addr->svm_cid == VMADDR_CID_ANY ||
+		other->svm_cid == VMADDR_CID_ANY ||
+		addr->svm_cid == other->svm_cid) &&
+	       addr->svm_port == other->svm_port;
+}
+EXPORT_SYMBOL_GPL(vsock_addr_equals_addr_any);
+
+int vsock_addr_cast(const struct sockaddr *addr,
+		    size_t len, struct sockaddr_vm **out_addr)
+{
+	if (len < sizeof(**out_addr))
+		return -EFAULT;
+
+	*out_addr = (struct sockaddr_vm *)addr;
+	return vsock_addr_validate(*out_addr);
+}
+EXPORT_SYMBOL_GPL(vsock_addr_cast);

net/vmw_vsock/vsock_addr.h

@@ -0,0 +1,32 @@
+/*
+ * VMware vSockets Driver
+ *
+ * Copyright (C) 2007-2013 VMware, Inc. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation version 2 and no later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ */
+
+#ifndef _VSOCK_ADDR_H_
+#define _VSOCK_ADDR_H_
+
+#include <linux/vm_sockets.h>
+
+void vsock_addr_init(struct sockaddr_vm *addr, u32 cid, u32 port);
+int vsock_addr_validate(const struct sockaddr_vm *addr);
+bool vsock_addr_bound(const struct sockaddr_vm *addr);
+void vsock_addr_unbind(struct sockaddr_vm *addr);
+bool vsock_addr_equals_addr(const struct sockaddr_vm *addr,
+			    const struct sockaddr_vm *other);
+bool vsock_addr_equals_addr_any(const struct sockaddr_vm *addr,
+				const struct sockaddr_vm *other);
+int vsock_addr_cast(const struct sockaddr *addr, size_t len,
+		    struct sockaddr_vm **out_addr);
+
+#endif

net/vmw_vsock/vsock_version.h

@@ -0,0 +1,22 @@
+/*
+ * VMware vSockets Driver
+ *
+ * Copyright (C) 2011-2012 VMware, Inc. All rights reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the Free
+ * Software Foundation version 2 and no later version.
+ *
+ * This program is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ */
+
+#ifndef _VSOCK_VERSION_H_
+#define _VSOCK_VERSION_H_
+
+#define VSOCK_DRIVER_VERSION_PARTS	{ 1, 0, 0, 0 }
+#define VSOCK_DRIVER_VERSION_STRING	"1.0.0.0-k"
+
+#endif /* _VSOCK_VERSION_H_ */