ipoib_main.c

linux-2.6.15.6

/*
 * Copyright (c) 2004 Topspin Communications.  All rights reserved.
 * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
 * Copyright (c) 2004 Voltaire, Inc. All rights reserved.
 *
 * This software is available to you under a choice of one of two
 * licenses.  You may choose to be licensed under the terms of the GNU
 * General Public License (GPL) Version 2, available from the file
 * COPYING in the main directory of this source tree, or the
 * OpenIB.org BSD license below:
 *
 *     Redistribution and use in source and binary forms, with or
 *     without modification, are permitted provided that the following
 *     conditions are met:
 *
 *      - Redistributions of source code must retain the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer.
 *
 *      - Redistributions in binary form must reproduce the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer in the documentation and/or other materials
 *        provided with the distribution.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 *
 * $Id: ipoib_main.c 1377 2004-12-23 19:57:12Z roland $
 */

#include "ipoib.h"

#include <linux/module.h>

#include <linux/init.h>
#include <linux/slab.h>
#include <linux/vmalloc.h>

#include <linux/if_arp.h>	/* For ARPHRD_xxx */

#include <linux/ip.h>
#include <linux/in.h>

MODULE_AUTHOR("Roland Dreier");
MODULE_DESCRIPTION("IP-over-InfiniBand net driver");
MODULE_LICENSE("Dual BSD/GPL");

#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG
int ipoib_debug_level;

module_param_named(debug_level, ipoib_debug_level, int, 0644);
MODULE_PARM_DESC(debug_level, "Enable debug tracing if > 0");
#endif

struct ipoib_path_iter {
	struct net_device *dev;
	struct ipoib_path  path;
};

static const u8 ipv4_bcast_addr[] = {
	0x00, 0xff, 0xff, 0xff,
	0xff, 0x12, 0x40, 0x1b,	0x00, 0x00, 0x00, 0x00,
	0x00, 0x00, 0x00, 0x00,	0xff, 0xff, 0xff, 0xff
};

struct workqueue_struct *ipoib_workqueue;

static void ipoib_add_one(struct ib_device *device);
static void ipoib_remove_one(struct ib_device *device);

static struct ib_client ipoib_client = {
	.name   = "ipoib",
	.add    = ipoib_add_one,
	.remove = ipoib_remove_one
};

int ipoib_open(struct net_device *dev)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);

	ipoib_dbg(priv, "bringing up interface\n");

	set_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags);

	if (ipoib_pkey_dev_delay_open(dev))
		return 0;

	if (ipoib_ib_dev_open(dev))
		return -EINVAL;

	if (ipoib_ib_dev_up(dev)) {
		ipoib_ib_dev_stop(dev);
		return -EINVAL;
	}

	if (!test_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags)) {
		struct ipoib_dev_priv *cpriv;

		/* Bring up any child interfaces too */
		down(&priv->vlan_mutex);
		list_for_each_entry(cpriv, &priv->child_intfs, list) {
			int flags;

			flags = cpriv->dev->flags;
			if (flags & IFF_UP)
				continue;

			dev_change_flags(cpriv->dev, flags | IFF_UP);
		}
		up(&priv->vlan_mutex);
	}

	netif_start_queue(dev);

	return 0;
}

static int ipoib_stop(struct net_device *dev)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);

	ipoib_dbg(priv, "stopping interface\n");

	clear_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags);

	netif_stop_queue(dev);

	ipoib_ib_dev_down(dev);
	ipoib_ib_dev_stop(dev);

	if (!test_bit(IPOIB_FLAG_SUBINTERFACE, &priv->flags)) {
		struct ipoib_dev_priv *cpriv;

		/* Bring down any child interfaces too */
		down(&priv->vlan_mutex);
		list_for_each_entry(cpriv, &priv->child_intfs, list) {
			int flags;

			flags = cpriv->dev->flags;
			if (!(flags & IFF_UP))
				continue;

			dev_change_flags(cpriv->dev, flags & ~IFF_UP);
		}
		up(&priv->vlan_mutex);
	}

	return 0;
}

static int ipoib_change_mtu(struct net_device *dev, int new_mtu)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);

	if (new_mtu > IPOIB_PACKET_SIZE - IPOIB_ENCAP_LEN)
		return -EINVAL;

	priv->admin_mtu = new_mtu;

	dev->mtu = min(priv->mcast_mtu, priv->admin_mtu);

	return 0;
}

static struct ipoib_path *__path_find(struct net_device *dev,
				      union ib_gid *gid)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct rb_node *n = priv->path_tree.rb_node;
	struct ipoib_path *path;
	int ret;

	while (n) {
		path = rb_entry(n, struct ipoib_path, rb_node);

		ret = memcmp(gid->raw, path->pathrec.dgid.raw,
			     sizeof (union ib_gid));

		if (ret < 0)
			n = n->rb_left;
		else if (ret > 0)
			n = n->rb_right;
		else
			return path;
	}

	return NULL;
}

static int __path_add(struct net_device *dev, struct ipoib_path *path)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct rb_node **n = &priv->path_tree.rb_node;
	struct rb_node *pn = NULL;
	struct ipoib_path *tpath;
	int ret;

	while (*n) {
		pn = *n;
		tpath = rb_entry(pn, struct ipoib_path, rb_node);

		ret = memcmp(path->pathrec.dgid.raw, tpath->pathrec.dgid.raw,
			     sizeof (union ib_gid));
		if (ret < 0)
			n = &pn->rb_left;
		else if (ret > 0)
			n = &pn->rb_right;
		else
			return -EEXIST;
	}

	rb_link_node(&path->rb_node, pn, n);
	rb_insert_color(&path->rb_node, &priv->path_tree);

	list_add_tail(&path->list, &priv->path_list);

	return 0;
}

static void path_free(struct net_device *dev, struct ipoib_path *path)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_neigh *neigh, *tn;
	struct sk_buff *skb;
	unsigned long flags;

	while ((skb = __skb_dequeue(&path->queue)))
		dev_kfree_skb_irq(skb);

	spin_lock_irqsave(&priv->lock, flags);

	list_for_each_entry_safe(neigh, tn, &path->neigh_list, list) {
		/*
		 * It's safe to call ipoib_put_ah() inside priv->lock
		 * here, because we know that path->ah will always
		 * hold one more reference, so ipoib_put_ah() will
		 * never do more than decrement the ref count.
		 */
		if (neigh->ah)
			ipoib_put_ah(neigh->ah);
		*to_ipoib_neigh(neigh->neighbour) = NULL;
		neigh->neighbour->ops->destructor = NULL;
		kfree(neigh);
	}

	spin_unlock_irqrestore(&priv->lock, flags);

	if (path->ah)
		ipoib_put_ah(path->ah);

	kfree(path);
}

#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG

struct ipoib_path_iter *ipoib_path_iter_init(struct net_device *dev)
{
	struct ipoib_path_iter *iter;

	iter = kmalloc(sizeof *iter, GFP_KERNEL);
	if (!iter)
		return NULL;

	iter->dev = dev;
	memset(iter->path.pathrec.dgid.raw, 0, 16);

	if (ipoib_path_iter_next(iter)) {
		kfree(iter);
		return NULL;
	}

	return iter;
}

int ipoib_path_iter_next(struct ipoib_path_iter *iter)
{
	struct ipoib_dev_priv *priv = netdev_priv(iter->dev);
	struct rb_node *n;
	struct ipoib_path *path;
	int ret = 1;

	spin_lock_irq(&priv->lock);

	n = rb_first(&priv->path_tree);

	while (n) {
		path = rb_entry(n, struct ipoib_path, rb_node);

		if (memcmp(iter->path.pathrec.dgid.raw, path->pathrec.dgid.raw,
			   sizeof (union ib_gid)) < 0) {
			iter->path = *path;
			ret = 0;
			break;
		}

		n = rb_next(n);
	}

	spin_unlock_irq(&priv->lock);

	return ret;
}

void ipoib_path_iter_read(struct ipoib_path_iter *iter,
			  struct ipoib_path *path)
{
	*path = iter->path;
}

#endif /* CONFIG_INFINIBAND_IPOIB_DEBUG */

void ipoib_flush_paths(struct net_device *dev)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_path *path, *tp;
	LIST_HEAD(remove_list);
	unsigned long flags;

	spin_lock_irqsave(&priv->lock, flags);

	list_splice(&priv->path_list, &remove_list);
	INIT_LIST_HEAD(&priv->path_list);

	list_for_each_entry(path, &remove_list, list)
		rb_erase(&path->rb_node, &priv->path_tree);

	spin_unlock_irqrestore(&priv->lock, flags);

	list_for_each_entry_safe(path, tp, &remove_list, list) {
		if (path->query)
			ib_sa_cancel_query(path->query_id, path->query);
		wait_for_completion(&path->done);
		path_free(dev, path);
	}
}

static void path_rec_completion(int status,
				struct ib_sa_path_rec *pathrec,
				void *path_ptr)
{
	struct ipoib_path *path = path_ptr;
	struct net_device *dev = path->dev;
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_ah *ah = NULL;
	struct ipoib_neigh *neigh;
	struct sk_buff_head skqueue;
	struct sk_buff *skb;
	unsigned long flags;

	if (pathrec)
		ipoib_dbg(priv, "PathRec LID 0x%04x for GID " IPOIB_GID_FMT "\n",
			  be16_to_cpu(pathrec->dlid), IPOIB_GID_ARG(pathrec->dgid));
	else
		ipoib_dbg(priv, "PathRec status %d for GID " IPOIB_GID_FMT "\n",
			  status, IPOIB_GID_ARG(path->pathrec.dgid));

	skb_queue_head_init(&skqueue);

	if (!status) {
		struct ib_ah_attr av = {
			.dlid 	       = be16_to_cpu(pathrec->dlid),
			.sl 	       = pathrec->sl,
			.port_num      = priv->port
		};
		int path_rate = ib_sa_rate_enum_to_int(pathrec->rate);

		if (path_rate > 0 && priv->local_rate > path_rate)
			av.static_rate = (priv->local_rate - 1) / path_rate;

		ipoib_dbg(priv, "static_rate %d for local port %dX, path %dX\n",
			  av.static_rate, priv->local_rate,
			  ib_sa_rate_enum_to_int(pathrec->rate));

		ah = ipoib_create_ah(dev, priv->pd, &av);
	}

	spin_lock_irqsave(&priv->lock, flags);

	path->ah = ah;

	if (ah) {
		path->pathrec = *pathrec;

		ipoib_dbg(priv, "created address handle %p for LID 0x%04x, SL %d\n",
			  ah, be16_to_cpu(pathrec->dlid), pathrec->sl);

		while ((skb = __skb_dequeue(&path->queue)))
			__skb_queue_tail(&skqueue, skb);

		list_for_each_entry(neigh, &path->neigh_list, list) {
			kref_get(&path->ah->ref);
			neigh->ah = path->ah;

			while ((skb = __skb_dequeue(&neigh->queue)))
				__skb_queue_tail(&skqueue, skb);
		}
	}

	path->query = NULL;
	complete(&path->done);

	spin_unlock_irqrestore(&priv->lock, flags);

	while ((skb = __skb_dequeue(&skqueue))) {
		skb->dev = dev;
		if (dev_queue_xmit(skb))
			ipoib_warn(priv, "dev_queue_xmit failed "
				   "to requeue packet\n");
	}
}

static struct ipoib_path *path_rec_create(struct net_device *dev,
					  union ib_gid *gid)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_path *path;

	path = kzalloc(sizeof *path, GFP_ATOMIC);
	if (!path)
		return NULL;

	path->dev = dev;

	skb_queue_head_init(&path->queue);

	INIT_LIST_HEAD(&path->neigh_list);

	memcpy(path->pathrec.dgid.raw, gid->raw, sizeof (union ib_gid));
	path->pathrec.sgid      = priv->local_gid;
	path->pathrec.pkey      = cpu_to_be16(priv->pkey);
	path->pathrec.numb_path = 1;

	return path;
}

static int path_rec_start(struct net_device *dev,
			  struct ipoib_path *path)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);

	ipoib_dbg(priv, "Start path record lookup for " IPOIB_GID_FMT "\n",
		  IPOIB_GID_ARG(path->pathrec.dgid));

	init_completion(&path->done);

	path->query_id =
		ib_sa_path_rec_get(priv->ca, priv->port,
				   &path->pathrec,
				   IB_SA_PATH_REC_DGID		|
				   IB_SA_PATH_REC_SGID		|
				   IB_SA_PATH_REC_NUMB_PATH	|
				   IB_SA_PATH_REC_PKEY,
				   1000, GFP_ATOMIC,
				   path_rec_completion,
				   path, &path->query);
	if (path->query_id < 0) {
		ipoib_warn(priv, "ib_sa_path_rec_get failed\n");
		path->query = NULL;
		return path->query_id;
	}

	return 0;
}

static void neigh_add_path(struct sk_buff *skb, struct net_device *dev)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_path *path;
	struct ipoib_neigh *neigh;

	neigh = kmalloc(sizeof *neigh, GFP_ATOMIC);
	if (!neigh) {
		++priv->stats.tx_dropped;
		dev_kfree_skb_any(skb);
		return;
	}

	skb_queue_head_init(&neigh->queue);
	neigh->neighbour = skb->dst->neighbour;
	*to_ipoib_neigh(skb->dst->neighbour) = neigh;

	/*
	 * We can only be called from ipoib_start_xmit, so we're
	 * inside tx_lock -- no need to save/restore flags.
	 */
	spin_lock(&priv->lock);

	path = __path_find(dev, (union ib_gid *) (skb->dst->neighbour->ha + 4));
	if (!path) {
		path = path_rec_create(dev,
				       (union ib_gid *) (skb->dst->neighbour->ha + 4));
		if (!path)
			goto err;

		__path_add(dev, path);
	}

	list_add_tail(&neigh->list, &path->neigh_list);

	if (path->pathrec.dlid) {
		kref_get(&path->ah->ref);
		neigh->ah = path->ah;

		ipoib_send(dev, skb, path->ah,
			   be32_to_cpup((__be32 *) skb->dst->neighbour->ha));
	} else {
		neigh->ah  = NULL;
		if (skb_queue_len(&neigh->queue) < IPOIB_MAX_PATH_REC_QUEUE) {
			__skb_queue_tail(&neigh->queue, skb);
		} else {
			++priv->stats.tx_dropped;
			dev_kfree_skb_any(skb);
		}

		if (!path->query && path_rec_start(dev, path))
			goto err;
	}

	spin_unlock(&priv->lock);
	return;

err:
	*to_ipoib_neigh(skb->dst->neighbour) = NULL;
	list_del(&neigh->list);
	neigh->neighbour->ops->destructor = NULL;
	kfree(neigh);

	++priv->stats.tx_dropped;
	dev_kfree_skb_any(skb);

	spin_unlock(&priv->lock);
}

static void ipoib_path_lookup(struct sk_buff *skb, struct net_device *dev)
{
	struct ipoib_dev_priv *priv = netdev_priv(skb->dev);

	/* Look up path record for unicasts */
	if (skb->dst->neighbour->ha[4] != 0xff) {
		neigh_add_path(skb, dev);
		return;
	}

	/* Add in the P_Key for multicasts */
	skb->dst->neighbour->ha[8] = (priv->pkey >> 8) & 0xff;
	skb->dst->neighbour->ha[9] = priv->pkey & 0xff;
	ipoib_mcast_send(dev, (union ib_gid *) (skb->dst->neighbour->ha + 4), skb);
}

static void unicast_arp_send(struct sk_buff *skb, struct net_device *dev,
			     struct ipoib_pseudoheader *phdr)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_path *path;

	/*
	 * We can only be called from ipoib_start_xmit, so we're
	 * inside tx_lock -- no need to save/restore flags.
	 */
	spin_lock(&priv->lock);

	path = __path_find(dev, (union ib_gid *) (phdr->hwaddr + 4));
	if (!path) {
		path = path_rec_create(dev,
				       (union ib_gid *) (phdr->hwaddr + 4));
		if (path) {
			/* put pseudoheader back on for next time */
			skb_push(skb, sizeof *phdr);
			__skb_queue_tail(&path->queue, skb);

			if (path_rec_start(dev, path)) {
				spin_unlock(&priv->lock);
				path_free(dev, path);
				return;
			} else
				__path_add(dev, path);
		} else {
			++priv->stats.tx_dropped;
			dev_kfree_skb_any(skb);