Initial commit; kernel source import

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.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.h

@@ -0,0 +1,142 @@
+/*
+ * 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
+
+/* The resource ID on which control packets are sent to the hypervisor. */
+#define VMCI_TRANSPORT_HYPERVISOR_PACKET_RID 15
+
+#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 +1,76 @@
+/*
+ * 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);
+
+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 +1,30 @@
+/*
+ * 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);
+int vsock_addr_cast(const struct sockaddr *addr, size_t len,
+		    struct sockaddr_vm **out_addr);
+
+#endif