tcp.c

/* -*- mode: c; c-basic-offset: 8; -*-
 *
 * vim: noexpandtab sw=8 ts=8 sts=0:
 *
 * Copyright (C) 2004 Oracle.  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; either
 * version 2 of the License, or (at your option) any 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.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 *
 * ----
 *
 * Callers for this were originally written against a very simple synchronus
 * API.  This implementation reflects those simple callers.  Some day I'm sure
 * we'll need to move to a more robust posting/callback mechanism.
 *
 * Transmit calls pass in kernel virtual addresses and block copying this into
 * the socket's tx buffers via a usual blocking sendmsg.  They'll block waiting
 * for a failed socket to timeout.  TX callers can also pass in a poniter to an
 * 'int' which gets filled with an errno off the wire in response to the
 * message they send.
 *
 * Handlers for unsolicited messages are registered.  Each socket has a page
 * that incoming data is copied into.  First the header, then the data.
 * Handlers are called from only one thread with a reference to this per-socket
 * page.  This page is destroyed after the handler call, so it can't be
 * referenced beyond the call.  Handlers may block but are discouraged from
 * doing so.
 *
 * Any framing errors (bad magic, large payload lengths) close a connection.
 *
 * Our sock_container holds the state we associate with a socket.  It's current
 * framing state is held there as well as the refcounting we do around when it
 * is safe to tear down the socket.  The socket is only finally torn down from
 * the container when the container loses all of its references -- so as long
 * as you hold a ref on the container you can trust that the socket is valid
 * for use with kernel socket APIs.
 *
 * Connections are initiated between a pair of nodes when the node with the
 * higher node number gets a heartbeat callback which indicates that the lower
 * numbered node has started heartbeating.  The lower numbered node is passive
 * and only accepts the connection if the higher numbered node is heartbeating.
 */

#include <linux/kernel.h>
#include <linux/jiffies.h>
#include <linux/slab.h>
#include <linux/idr.h>
#include <linux/kref.h>
#include <net/tcp.h>

#include <asm/uaccess.h>

#include "heartbeat.h"
#include "tcp.h"
#include "nodemanager.h"
#define MLOG_MASK_PREFIX ML_TCP
#include "masklog.h"
#include "quorum.h"

#include "tcp_internal.h"

/* 
 * The linux network stack isn't sparse endian clean.. It has macros like
 * ntohs() which perform the endian checks and structs like sockaddr_in
 * which aren't annotated.  So __force is found here to get the build
 * clean.  When they emerge from the dark ages and annotate the code
 * we can remove these.
 */

#define SC_NODEF_FMT "node %s (num %u) at %u.%u.%u.%u:%u"
#define SC_NODEF_ARGS(sc) sc->sc_node->nd_name, sc->sc_node->nd_num,	\
			  NIPQUAD(sc->sc_node->nd_ipv4_address),	\
			  ntohs(sc->sc_node->nd_ipv4_port)

/*
 * In the following two log macros, the whitespace after the ',' just
 * before ##args is intentional. Otherwise, gcc 2.95 will eat the
 * previous token if args expands to nothing.
 */
#define msglog(hdr, fmt, args...) do {					\
	typeof(hdr) __hdr = (hdr);					\
	mlog(ML_MSG, "[mag %u len %u typ %u stat %d sys_stat %d "	\
	     "key %08x num %u] " fmt,					\
	     be16_to_cpu(__hdr->magic), be16_to_cpu(__hdr->data_len), 	\
	     be16_to_cpu(__hdr->msg_type), be32_to_cpu(__hdr->status),	\
	     be32_to_cpu(__hdr->sys_status), be32_to_cpu(__hdr->key),	\
	     be32_to_cpu(__hdr->msg_num) ,  ##args);			\
} while (0)

#define sclog(sc, fmt, args...) do {					\
	typeof(sc) __sc = (sc);						\
	mlog(ML_SOCKET, "[sc %p refs %d sock %p node %u page %p "	\
	     "pg_off %zu] " fmt, __sc,					\
	     atomic_read(&__sc->sc_kref.refcount), __sc->sc_sock,	\
	    __sc->sc_node->nd_num, __sc->sc_page, __sc->sc_page_off ,	\
	    ##args);							\
} while (0)

static DEFINE_RWLOCK(o2net_handler_lock);
static struct rb_root o2net_handler_tree = RB_ROOT;

static struct o2net_node o2net_nodes[O2NM_MAX_NODES];

/* XXX someday we'll need better accounting */
static struct socket *o2net_listen_sock = NULL;

/*
 * listen work is only queued by the listening socket callbacks on the
 * o2net_wq.  teardown detaches the callbacks before destroying the workqueue.
 * quorum work is queued as sock containers are shutdown.. stop_listening
 * tears down all the node's sock containers, preventing future shutdowns
 * and queued quroum work, before canceling delayed quorum work and
 * destroying the work queue.
 */
static struct workqueue_struct *o2net_wq;
static struct work_struct o2net_listen_work;

static struct o2hb_callback_func o2net_hb_up, o2net_hb_down;
#define O2NET_HB_PRI 0x1

static struct o2net_handshake *o2net_hand;
static struct o2net_msg *o2net_keep_req, *o2net_keep_resp;

static int o2net_sys_err_translations[O2NET_ERR_MAX] =
		{[O2NET_ERR_NONE]	= 0,
		 [O2NET_ERR_NO_HNDLR]	= -ENOPROTOOPT,
		 [O2NET_ERR_OVERFLOW]	= -EOVERFLOW,
		 [O2NET_ERR_DIED]	= -EHOSTDOWN,};

/* can't quite avoid *all* internal declarations :/ */
static void o2net_sc_connect_completed(struct work_struct *work);
static void o2net_rx_until_empty(struct work_struct *work);
static void o2net_shutdown_sc(struct work_struct *work);
static void o2net_listen_data_ready(struct sock *sk, int bytes);
static void o2net_sc_send_keep_req(struct work_struct *work);
static void o2net_idle_timer(unsigned long data);
static void o2net_sc_postpone_idle(struct o2net_sock_container *sc);
static void o2net_sc_reset_idle_timer(struct o2net_sock_container *sc);

/*
 * FIXME: These should use to_o2nm_cluster_from_node(), but we end up
 * losing our parent link to the cluster during shutdown. This can be
 * solved by adding a pre-removal callback to configfs, or passing
 * around the cluster with the node. -jeffm
 */
static inline int o2net_reconnect_delay(struct o2nm_node *node)
{
	return o2nm_single_cluster->cl_reconnect_delay_ms;
}

static inline int o2net_keepalive_delay(struct o2nm_node *node)
{
	return o2nm_single_cluster->cl_keepalive_delay_ms;
}

static inline int o2net_idle_timeout(struct o2nm_node *node)
{
	return o2nm_single_cluster->cl_idle_timeout_ms;
}

static inline int o2net_sys_err_to_errno(enum o2net_system_error err)
{
	int trans;
	BUG_ON(err >= O2NET_ERR_MAX);
	trans = o2net_sys_err_translations[err];

	/* Just in case we mess up the translation table above */
	BUG_ON(err != O2NET_ERR_NONE && trans == 0);
	return trans;
}

static struct o2net_node * o2net_nn_from_num(u8 node_num)
{
	BUG_ON(node_num >= ARRAY_SIZE(o2net_nodes));
	return &o2net_nodes[node_num];
}

static u8 o2net_num_from_nn(struct o2net_node *nn)
{
	BUG_ON(nn == NULL);
	return nn - o2net_nodes;
}

/* ------------------------------------------------------------ */

static int o2net_prep_nsw(struct o2net_node *nn, struct o2net_status_wait *nsw)
{
	int ret = 0;

	do {
		if (!idr_pre_get(&nn->nn_status_idr, GFP_ATOMIC)) {
			ret = -EAGAIN;
			break;
		}
		spin_lock(&nn->nn_lock);
		ret = idr_get_new(&nn->nn_status_idr, nsw, &nsw->ns_id);
		if (ret == 0)
			list_add_tail(&nsw->ns_node_item,
				      &nn->nn_status_list);
		spin_unlock(&nn->nn_lock);
	} while (ret == -EAGAIN);

	if (ret == 0)  {
		init_waitqueue_head(&nsw->ns_wq);
		nsw->ns_sys_status = O2NET_ERR_NONE;
		nsw->ns_status = 0;
	}

	return ret;
}

static void o2net_complete_nsw_locked(struct o2net_node *nn,
				      struct o2net_status_wait *nsw,
				      enum o2net_system_error sys_status,
				      s32 status)
{
	assert_spin_locked(&nn->nn_lock);

	if (!list_empty(&nsw->ns_node_item)) {
		list_del_init(&nsw->ns_node_item);
		nsw->ns_sys_status = sys_status;
		nsw->ns_status = status;
		idr_remove(&nn->nn_status_idr, nsw->ns_id);
		wake_up(&nsw->ns_wq);
	}
}

static void o2net_complete_nsw(struct o2net_node *nn,
			       struct o2net_status_wait *nsw,
			       u64 id, enum o2net_system_error sys_status,
			       s32 status)
{
	spin_lock(&nn->nn_lock);
	if (nsw == NULL) {
		if (id > INT_MAX)
			goto out;

		nsw = idr_find(&nn->nn_status_idr, id);
		if (nsw == NULL)
			goto out;
	}

	o2net_complete_nsw_locked(nn, nsw, sys_status, status);

out:
	spin_unlock(&nn->nn_lock);
	return;
}

static void o2net_complete_nodes_nsw(struct o2net_node *nn)
{
	struct list_head *iter, *tmp;
	unsigned int num_kills = 0;
	struct o2net_status_wait *nsw;

	assert_spin_locked(&nn->nn_lock);

	list_for_each_safe(iter, tmp, &nn->nn_status_list) {
		nsw = list_entry(iter, struct o2net_status_wait, ns_node_item);
		o2net_complete_nsw_locked(nn, nsw, O2NET_ERR_DIED, 0);
		num_kills++;
	}

	mlog(0, "completed %d messages for node %u\n", num_kills,
	     o2net_num_from_nn(nn));
}

static int o2net_nsw_completed(struct o2net_node *nn,
			       struct o2net_status_wait *nsw)
{
	int completed;
	spin_lock(&nn->nn_lock);
	completed = list_empty(&nsw->ns_node_item);
	spin_unlock(&nn->nn_lock);
	return completed;
}

/* ------------------------------------------------------------ */

static void sc_kref_release(struct kref *kref)
{
	struct o2net_sock_container *sc = container_of(kref,
					struct o2net_sock_container, sc_kref);
	BUG_ON(timer_pending(&sc->sc_idle_timeout));

	sclog(sc, "releasing\n");

	if (sc->sc_sock) {
		sock_release(sc->sc_sock);
		sc->sc_sock = NULL;
	}

	o2nm_node_put(sc->sc_node);
	sc->sc_node = NULL;

	kfree(sc);
}

static void sc_put(struct o2net_sock_container *sc)
{
	sclog(sc, "put\n");
	kref_put(&sc->sc_kref, sc_kref_release);
}
static void sc_get(struct o2net_sock_container *sc)
{
	sclog(sc, "get\n");
	kref_get(&sc->sc_kref);
}
static struct o2net_sock_container *sc_alloc(struct o2nm_node *node)
{
	struct o2net_sock_container *sc, *ret = NULL;
	struct page *page = NULL;

	page = alloc_page(GFP_NOFS);
	sc = kcalloc(1, sizeof(*sc), GFP_NOFS);
	if (sc == NULL || page == NULL)
		goto out;

	kref_init(&sc->sc_kref);
	o2nm_node_get(node);
	sc->sc_node = node;

	INIT_WORK(&sc->sc_connect_work, o2net_sc_connect_completed);
	INIT_WORK(&sc->sc_rx_work, o2net_rx_until_empty);
	INIT_WORK(&sc->sc_shutdown_work, o2net_shutdown_sc);
	INIT_DELAYED_WORK(&sc->sc_keepalive_work, o2net_sc_send_keep_req);

	init_timer(&sc->sc_idle_timeout);
	sc->sc_idle_timeout.function = o2net_idle_timer;
	sc->sc_idle_timeout.data = (unsigned long)sc;

	sclog(sc, "alloced\n");

	ret = sc;
	sc->sc_page = page;
	sc = NULL;
	page = NULL;

out:
	if (page)
		__free_page(page);
	kfree(sc);

	return ret;
}

/* ------------------------------------------------------------ */

static void o2net_sc_queue_work(struct o2net_sock_container *sc,
				struct work_struct *work)
{
	sc_get(sc);
	if (!queue_work(o2net_wq, work))
		sc_put(sc);
}
static void o2net_sc_queue_delayed_work(struct o2net_sock_container *sc,
					struct delayed_work *work,
					int delay)
{
	sc_get(sc);
	if (!queue_delayed_work(o2net_wq, work, delay))
		sc_put(sc);
}
static void o2net_sc_cancel_delayed_work(struct o2net_sock_container *sc,
					 struct delayed_work *work)
{
	if (cancel_delayed_work(work))
		sc_put(sc);
}

static atomic_t o2net_connected_peers = ATOMIC_INIT(0);

int o2net_num_connected_peers(void)
{
	return atomic_read(&o2net_connected_peers);
}

static void o2net_set_nn_state(struct o2net_node *nn,
			       struct o2net_sock_container *sc,
			       unsigned valid, int err)
{
	int was_valid = nn->nn_sc_valid;
	int was_err = nn->nn_persistent_error;
	struct o2net_sock_container *old_sc = nn->nn_sc;

	assert_spin_locked(&nn->nn_lock);

	if (old_sc && !sc)
		atomic_dec(&o2net_connected_peers);
	else if (!old_sc && sc)
		atomic_inc(&o2net_connected_peers);

	/* the node num comparison and single connect/accept path should stop
	 * an non-null sc from being overwritten with another */
	BUG_ON(sc && nn->nn_sc && nn->nn_sc != sc);
	mlog_bug_on_msg(err && valid, "err %d valid %u\n", err, valid);
	mlog_bug_on_msg(valid && !sc, "valid %u sc %p\n", valid, sc);

	/* we won't reconnect after our valid conn goes away for
	 * this hb iteration.. here so it shows up in the logs */
	if (was_valid && !valid && err == 0)
		err = -ENOTCONN;

	mlog(ML_CONN, "node %u sc: %p -> %p, valid %u -> %u, err %d -> %d\n",
	     o2net_num_from_nn(nn), nn->nn_sc, sc, nn->nn_sc_valid, valid,
	     nn->nn_persistent_error, err);

	nn->nn_sc = sc;
	nn->nn_sc_valid = valid ? 1 : 0;
	nn->nn_persistent_error = err;

	/* mirrors o2net_tx_can_proceed() */
	if (nn->nn_persistent_error || nn->nn_sc_valid)
		wake_up(&nn->nn_sc_wq);

	if (!was_err && nn->nn_persistent_error) {
		o2quo_conn_err(o2net_num_from_nn(nn));
		queue_delayed_work(o2net_wq, &nn->nn_still_up,
				   msecs_to_jiffies(O2NET_QUORUM_DELAY_MS));
	}

	if (was_valid && !valid) {
		printk(KERN_INFO "o2net: no longer connected to "
		       SC_NODEF_FMT "\n", SC_NODEF_ARGS(old_sc));
		o2net_complete_nodes_nsw(nn);
	}

	if (!was_valid && valid) {
		o2quo_conn_up(o2net_num_from_nn(nn));
		/* this is a bit of a hack.  we only try reconnecting
		 * when heartbeating starts until we get a connection.
		 * if that connection then dies we don't try reconnecting.
		 * the only way to start connecting again is to down
		 * heartbeat and bring it back up. */
		cancel_delayed_work(&nn->nn_connect_expired);
		printk(KERN_INFO "o2net: %s " SC_NODEF_FMT "\n",
		       o2nm_this_node() > sc->sc_node->nd_num ?
		       		"connected to" : "accepted connection from",
		       SC_NODEF_ARGS(sc));
	}

	/* trigger the connecting worker func as long as we're not valid,
	 * it will back off if it shouldn't connect.  This can be called
	 * from node config teardown and so needs to be careful about
	 * the work queue actually being up. */
	if (!valid && o2net_wq) {
		unsigned long delay;
		/* delay if we're withing a RECONNECT_DELAY of the
		 * last attempt */
		delay = (nn->nn_last_connect_attempt +
			 msecs_to_jiffies(o2net_reconnect_delay(sc->sc_node)))
			- jiffies;
		if (delay > msecs_to_jiffies(o2net_reconnect_delay(sc->sc_node)))
			delay = 0;
		mlog(ML_CONN, "queueing conn attempt in %lu jiffies\n", delay);
		queue_delayed_work(o2net_wq, &nn->nn_connect_work, delay);
	}

	/* keep track of the nn's sc ref for the caller */
	if ((old_sc == NULL) && sc)
		sc_get(sc);
	if (old_sc && (old_sc != sc)) {
		o2net_sc_queue_work(old_sc, &old_sc->sc_shutdown_work);
		sc_put(old_sc);
	}
}

/* see o2net_register_callbacks() */
static void o2net_data_ready(struct sock *sk, int bytes)
{
	void (*ready)(struct sock *sk, int bytes);

	read_lock(&sk->sk_callback_lock);
	if (sk->sk_user_data) {
		struct o2net_sock_container *sc = sk->sk_user_data;
		sclog(sc, "data_ready hit\n");
		do_gettimeofday(&sc->sc_tv_data_ready);
		o2net_sc_queue_work(sc, &sc->sc_rx_work);
		ready = sc->sc_data_ready;
	} else {
		ready = sk->sk_data_ready;
	}
	read_unlock(&sk->sk_callback_lock);

	ready(sk, bytes);
}

/* see o2net_register_callbacks() */
static void o2net_state_change(struct sock *sk)
{
	void (*state_change)(struct sock *sk);
	struct o2net_sock_container *sc;

	read_lock(&sk->sk_callback_lock);
	sc = sk->sk_user_data;
	if (sc == NULL) {
		state_change = sk->sk_state_change;
		goto out;
	}

	sclog(sc, "state_change to %d\n", sk->sk_state);

	state_change = sc->sc_state_change;

	switch(sk->sk_state) {
		/* ignore connecting sockets as they make progress */
		case TCP_SYN_SENT:
		case TCP_SYN_RECV:
			break;
		case TCP_ESTABLISHED:
			o2net_sc_queue_work(sc, &sc->sc_connect_work);
			break;
		default:
			o2net_sc_queue_work(sc, &sc->sc_shutdown_work);
			break;
	}
out:
	read_unlock(&sk->sk_callback_lock);
	state_change(sk);
}

/*
 * we register callbacks so we can queue work on events before calling
 * the original callbacks.  our callbacks our careful to test user_data
 * to discover when they've reaced with o2net_unregister_callbacks().
 */
static void o2net_register_callbacks(struct sock *sk,
				     struct o2net_sock_container *sc)
{
	write_lock_bh(&sk->sk_callback_lock);

	/* accepted sockets inherit the old listen socket data ready */
	if (sk->sk_data_ready == o2net_listen_data_ready) {
		sk->sk_data_ready = sk->sk_user_data;
		sk->sk_user_data = NULL;
	}

	BUG_ON(sk->sk_user_data != NULL);
	sk->sk_user_data = sc;
	sc_get(sc);

	sc->sc_data_ready = sk->sk_data_ready;
	sc->sc_state_change = sk->sk_state_change;
	sk->sk_data_ready = o2net_data_ready;
	sk->sk_state_change = o2net_state_change;

	write_unlock_bh(&sk->sk_callback_lock);
}

static int o2net_unregister_callbacks(struct sock *sk,
			           struct o2net_sock_container *sc)
{
	int ret = 0;

	write_lock_bh(&sk->sk_callback_lock);
	if (sk->sk_user_data == sc) {
		ret = 1;
		sk->sk_user_data = NULL;
		sk->sk_data_ready = sc->sc_data_ready;
		sk->sk_state_change = sc->sc_state_change;
	}
	write_unlock_bh(&sk->sk_callback_lock);

	return ret;
}

/*
 * this is a little helper that is called by callers who have seen a problem
 * with an sc and want to detach it from the nn if someone already hasn't beat
 * them to it.  if an error is given then the shutdown will be persistent
 * and pending transmits will be canceled.
 */
static void o2net_ensure_shutdown(struct o2net_node *nn,
			           struct o2net_sock_container *sc,
				   int err)
{
	spin_lock(&nn->nn_lock);
	if (nn->nn_sc == sc)
		o2net_set_nn_state(nn, NULL, 0, err);
	spin_unlock(&nn->nn_lock);
}

/*
 * This work queue function performs the blocking parts of socket shutdown.  A
 * few paths lead here.  set_nn_state will trigger this callback if it sees an
 * sc detached from the nn.  state_change will also trigger this callback
 * directly when it sees errors.  In that case we need to call set_nn_state
 * ourselves as state_change couldn't get the nn_lock and call set_nn_state
 * itself.
 */
static void o2net_shutdown_sc(struct work_struct *work)
{
	struct o2net_sock_container *sc =
		container_of(work, struct o2net_sock_container,
			     sc_shutdown_work);
	struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num);

	sclog(sc, "shutting down\n");

	/* drop the callbacks ref and call shutdown only once */
	if (o2net_unregister_callbacks(sc->sc_sock->sk, sc)) {
		/* we shouldn't flush as we're in the thread, the
		 * races with pending sc work structs are harmless */
		del_timer_sync(&sc->sc_idle_timeout);
		o2net_sc_cancel_delayed_work(sc, &sc->sc_keepalive_work);
		sc_put(sc);
		sc->sc_sock->ops->shutdown(sc->sc_sock,
					   RCV_SHUTDOWN|SEND_SHUTDOWN);
	}

	/* not fatal so failed connects before the other guy has our
	 * heartbeat can be retried */
	o2net_ensure_shutdown(nn, sc, 0);
	sc_put(sc);
}

/* ------------------------------------------------------------ */

static int o2net_handler_cmp(struct o2net_msg_handler *nmh, u32 msg_type,
			     u32 key)
{
	int ret = memcmp(&nmh->nh_key, &key, sizeof(key));

	if (ret == 0)
		ret = memcmp(&nmh->nh_msg_type, &msg_type, sizeof(msg_type));

	return ret;
}

static struct o2net_msg_handler *
o2net_handler_tree_lookup(u32 msg_type, u32 key, struct rb_node ***ret_p,
			  struct rb_node **ret_parent)
{
        struct rb_node **p = &o2net_handler_tree.rb_node;
        struct rb_node *parent = NULL;
	struct o2net_msg_handler *nmh, *ret = NULL;
	int cmp;

        while (*p) {
                parent = *p;
                nmh = rb_entry(parent, struct o2net_msg_handler, nh_node);
		cmp = o2net_handler_cmp(nmh, msg_type, key);

                if (cmp < 0)
                        p = &(*p)->rb_left;
                else if (cmp > 0)
                        p = &(*p)->rb_right;
                else {
			ret = nmh;
                        break;
		}
        }

        if (ret_p != NULL)
                *ret_p = p;
        if (ret_parent != NULL)
                *ret_parent = parent;

        return ret;
}

static void o2net_handler_kref_release(struct kref *kref)
{
	struct o2net_msg_handler *nmh;
	nmh = container_of(kref, struct o2net_msg_handler, nh_kref);

	kfree(nmh);
}

static void o2net_handler_put(struct o2net_msg_handler *nmh)
{
	kref_put(&nmh->nh_kref, o2net_handler_kref_release);
}

/* max_len is protection for the handler func.  incoming messages won't
 * be given to the handler if their payload is longer than the max. */
int o2net_register_handler(u32 msg_type, u32 key, u32 max_len,
			   o2net_msg_handler_func *func, void *data,
			   struct list_head *unreg_list)
{
	struct o2net_msg_handler *nmh = NULL;
	struct rb_node **p, *parent;
	int ret = 0;

	if (max_len > O2NET_MAX_PAYLOAD_BYTES) {
		mlog(0, "max_len for message handler out of range: %u\n",
			max_len);
		ret = -EINVAL;
		goto out;
	}

	if (!msg_type) {
		mlog(0, "no message type provided: %u, %p\n", msg_type, func);
		ret = -EINVAL;
		goto out;

	}
	if (!func) {
		mlog(0, "no message handler provided: %u, %p\n",
		       msg_type, func);
		ret = -EINVAL;
		goto out;
	}

       	nmh = kcalloc(1, sizeof(struct o2net_msg_handler), GFP_NOFS);
	if (nmh == NULL) {
		ret = -ENOMEM;
		goto out;
	}

	nmh->nh_func = func;
	nmh->nh_func_data = data;
	nmh->nh_msg_type = msg_type;
	nmh->nh_max_len = max_len;
	nmh->nh_key = key;
	/* the tree and list get this ref.. they're both removed in
	 * unregister when this ref is dropped */
	kref_init(&nmh->nh_kref);
	INIT_LIST_HEAD(&nmh->nh_unregister_item);

	write_lock(&o2net_handler_lock);
	if (o2net_handler_tree_lookup(msg_type, key, &p, &parent))
		ret = -EEXIST;
	else {
	        rb_link_node(&nmh->nh_node, parent, p);
		rb_insert_color(&nmh->nh_node, &o2net_handler_tree);
		list_add_tail(&nmh->nh_unregister_item, unreg_list);

		mlog(ML_TCP, "registered handler func %p type %u key %08x\n",
		     func, msg_type, key);
		/* we've had some trouble with handlers seemingly vanishing. */
		mlog_bug_on_msg(o2net_handler_tree_lookup(msg_type, key, &p,
							  &parent) == NULL,
			        "couldn't find handler we *just* registerd "
				"for type %u key %08x\n", msg_type, key);
	}
	write_unlock(&o2net_handler_lock);
	if (ret)
		goto out;

out:
	if (ret)
		kfree(nmh);

	return ret;
}
EXPORT_SYMBOL_GPL(o2net_register_handler);

void o2net_unregister_handler_list(struct list_head *list)
{
	struct list_head *pos, *n;
	struct o2net_msg_handler *nmh;

	write_lock(&o2net_handler_lock);
	list_for_each_safe(pos, n, list) {
		nmh = list_entry(pos, struct o2net_msg_handler,
				 nh_unregister_item);
		mlog(ML_TCP, "unregistering handler func %p type %u key %08x\n",
		     nmh->nh_func, nmh->nh_msg_type, nmh->nh_key);
		rb_erase(&nmh->nh_node, &o2net_handler_tree);
		list_del_init(&nmh->nh_unregister_item);
		kref_put(&nmh->nh_kref, o2net_handler_kref_release);
	}
	write_unlock(&o2net_handler_lock);
}
EXPORT_SYMBOL_GPL(o2net_unregister_handler_list);

static struct o2net_msg_handler *o2net_handler_get(u32 msg_type, u32 key)
{
	struct o2net_msg_handler *nmh;

	read_lock(&o2net_handler_lock);
	nmh = o2net_handler_tree_lookup(msg_type, key, NULL, NULL);
	if (nmh)
		kref_get(&nmh->nh_kref);
	read_unlock(&o2net_handler_lock);

	return nmh;
}

/* ------------------------------------------------------------ */

static int o2net_recv_tcp_msg(struct socket *sock, void *data, size_t len)
{
	int ret;
	mm_segment_t oldfs;
	struct kvec vec = {
		.iov_len = len,
		.iov_base = data,
	};
	struct msghdr msg = {
		.msg_iovlen = 1,
		.msg_iov = (struct iovec *)&vec,
       		.msg_flags = MSG_DONTWAIT,
	};

	oldfs = get_fs();
	set_fs(get_ds());
	ret = sock_recvmsg(sock, &msg, len, msg.msg_flags);
	set_fs(oldfs);

	return ret;
}

static int o2net_send_tcp_msg(struct socket *sock, struct kvec *vec,
			      size_t veclen, size_t total)
{
	int ret;
	mm_segment_t oldfs;
	struct msghdr msg = {
		.msg_iov = (struct iovec *)vec,
		.msg_iovlen = veclen,
	};

	if (sock == NULL) {
		ret = -EINVAL;
		goto out;
	}

	oldfs = get_fs();
	set_fs(get_ds());
	ret = sock_sendmsg(sock, &msg, total);
	set_fs(oldfs);
	if (ret != total) {
		mlog(ML_ERROR, "sendmsg returned %d instead of %zu\n", ret,
		     total);
		if (ret >= 0)
			ret = -EPIPE; /* should be smarter, I bet */
		goto out;
	}

	ret = 0;
out:
	if (ret < 0)
		mlog(0, "returning error: %d\n", ret);
	return ret;
}

static void o2net_sendpage(struct o2net_sock_container *sc,
			   void *kmalloced_virt,
			   size_t size)
{
	struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num);
	ssize_t ret;


	ret = sc->sc_sock->ops->sendpage(sc->sc_sock,
					 virt_to_page(kmalloced_virt),
					 (long)kmalloced_virt & ~PAGE_MASK,
					 size, MSG_DONTWAIT);
	if (ret != size) {
		mlog(ML_ERROR, "sendpage of size %zu to " SC_NODEF_FMT 
		     " failed with %zd\n", size, SC_NODEF_ARGS(sc), ret);
		o2net_ensure_shutdown(nn, sc, 0);
	}
}

static void o2net_init_msg(struct o2net_msg *msg, u16 data_len, u16 msg_type, u32 key)
{
	memset(msg, 0, sizeof(struct o2net_msg));
	msg->magic = cpu_to_be16(O2NET_MSG_MAGIC);
	msg->data_len = cpu_to_be16(data_len);
	msg->msg_type = cpu_to_be16(msg_type);
	msg->sys_status = cpu_to_be32(O2NET_ERR_NONE);
	msg->status = 0;
	msg->key = cpu_to_be32(key);
}

static int o2net_tx_can_proceed(struct o2net_node *nn,
			        struct o2net_sock_container **sc_ret,
				int *error)
{
	int ret = 0;

	spin_lock(&nn->nn_lock);
	if (nn->nn_persistent_error) {
		ret = 1;
		*sc_ret = NULL;
		*error = nn->nn_persistent_error;
	} else if (nn->nn_sc_valid) {
		kref_get(&nn->nn_sc->sc_kref);

		ret = 1;
		*sc_ret = nn->nn_sc;
		*error = 0;
	}
	spin_unlock(&nn->nn_lock);

	return ret;
}

int o2net_send_message_vec(u32 msg_type, u32 key, struct kvec *caller_vec,
			   size_t caller_veclen, u8 target_node, int *status)
{
	int ret, error = 0;
	struct o2net_msg *msg = NULL;
	size_t veclen, caller_bytes = 0;
	struct kvec *vec = NULL;
	struct o2net_sock_container *sc = NULL;
	struct o2net_node *nn = o2net_nn_from_num(target_node);
	struct o2net_status_wait nsw = {
		.ns_node_item = LIST_HEAD_INIT(nsw.ns_node_item),
	};

	if (o2net_wq == NULL) {
		mlog(0, "attempt to tx without o2netd running\n");
		ret = -ESRCH;
		goto out;
	}

	if (caller_veclen == 0) {
		mlog(0, "bad kvec array length\n");
		ret = -EINVAL;
		goto out;
	}

	caller_bytes = iov_length((struct iovec *)caller_vec, caller_veclen);
	if (caller_bytes > O2NET_MAX_PAYLOAD_BYTES) {
		mlog(0, "total payload len %zu too large\n", caller_bytes);
		ret = -EINVAL;
		goto out;
	}

	if (target_node == o2nm_this_node()) {
		ret = -ELOOP;
		goto out;
	}

	ret = wait_event_interruptible(nn->nn_sc_wq,
				       o2net_tx_can_proceed(nn, &sc, &error));
	if (!ret && error)
		ret = error;
	if (ret)
		goto out;

	veclen = caller_veclen + 1;
	vec = kmalloc(sizeof(struct kvec) * veclen, GFP_ATOMIC);
	if (vec == NULL) {
		mlog(0, "failed to %zu element kvec!\n", veclen);
		ret = -ENOMEM;
		goto out;
	}

	msg = kmalloc(sizeof(struct o2net_msg), GFP_ATOMIC);
	if (!msg) {
		mlog(0, "failed to allocate a o2net_msg!\n");
		ret = -ENOMEM;
		goto out;
	}

	o2net_init_msg(msg, caller_bytes, msg_type, key);

	vec[0].iov_len = sizeof(struct o2net_msg);
	vec[0].iov_base = msg;
	memcpy(&vec[1], caller_vec, caller_veclen * sizeof(struct kvec));

	ret = o2net_prep_nsw(nn, &nsw);
	if (ret)
		goto out;

	msg->msg_num = cpu_to_be32(nsw.ns_id);

	/* finally, convert the message header to network byte-order
	 * and send */
	ret = o2net_send_tcp_msg(sc->sc_sock, vec, veclen,
				 sizeof(struct o2net_msg) + caller_bytes);
	msglog(msg, "sending returned %d\n", ret);
	if (ret < 0) {
		mlog(0, "error returned from o2net_send_tcp_msg=%d\n", ret);
		goto out;
	}

	/* wait on other node's handler */
	wait_event(nsw.ns_wq, o2net_nsw_completed(nn, &nsw));

	/* Note that we avoid overwriting the callers status return
	 * variable if a system error was reported on the other
	 * side. Callers beware. */
	ret = o2net_sys_err_to_errno(nsw.ns_sys_status);
	if (status && !ret)
		*status = nsw.ns_status;

	mlog(0, "woken, returning system status %d, user status %d\n",
	     ret, nsw.ns_status);
out:
	if (sc)
		sc_put(sc);
	if (vec)