[linux.git] / net / xfrm / xfrm_device.c

// SPDX-License-Identifier: GPL-2.0-or-later
/*
 * xfrm_device.c - IPsec device offloading code.
 *
 * Copyright (c) 2015 secunet Security Networks AG
 *
 * Author:
 * Steffen Klassert <[email protected]>
 */

#include <linux/errno.h>
#include <linux/module.h>
#include <linux/netdevice.h>
#include <linux/skbuff.h>
#include <linux/slab.h>
#include <linux/spinlock.h>
#include <net/dst.h>
#include <net/xfrm.h>
#include <linux/notifier.h>

#ifdef CONFIG_XFRM_OFFLOAD
static void __xfrm_transport_prep(struct xfrm_state *x, struct sk_buff *skb,
				  unsigned int hsize)
{
	struct xfrm_offload *xo = xfrm_offload(skb);

	skb_reset_mac_len(skb);
	if (xo->flags & XFRM_GSO_SEGMENT)
		skb->transport_header -= x->props.header_len;

	pskb_pull(skb, skb_transport_offset(skb) + x->props.header_len);
}

static void __xfrm_mode_tunnel_prep(struct xfrm_state *x, struct sk_buff *skb,
				    unsigned int hsize)

{
	struct xfrm_offload *xo = xfrm_offload(skb);

	if (xo->flags & XFRM_GSO_SEGMENT)
		skb->transport_header = skb->network_header + hsize;

	skb_reset_mac_len(skb);
	pskb_pull(skb, skb->mac_len + x->props.header_len);
}

static void __xfrm_mode_beet_prep(struct xfrm_state *x, struct sk_buff *skb,
				  unsigned int hsize)
{
	struct xfrm_offload *xo = xfrm_offload(skb);
	int phlen = 0;

	if (xo->flags & XFRM_GSO_SEGMENT)
		skb->transport_header = skb->network_header + hsize;

	skb_reset_mac_len(skb);
	if (x->sel.family != AF_INET6) {
		phlen = IPV4_BEET_PHMAXLEN;
		if (x->outer_mode.family == AF_INET6)
			phlen += sizeof(struct ipv6hdr) - sizeof(struct iphdr);
	}

	pskb_pull(skb, skb->mac_len + hsize + (x->props.header_len - phlen));
}

/* Adjust pointers into the packet when IPsec is done at layer2 */
static void xfrm_outer_mode_prep(struct xfrm_state *x, struct sk_buff *skb)
{
	switch (x->outer_mode.encap) {
	case XFRM_MODE_TUNNEL:
		if (x->outer_mode.family == AF_INET)
			return __xfrm_mode_tunnel_prep(x, skb,
						       sizeof(struct iphdr));
		if (x->outer_mode.family == AF_INET6)
			return __xfrm_mode_tunnel_prep(x, skb,
						       sizeof(struct ipv6hdr));
		break;
	case XFRM_MODE_TRANSPORT:
		if (x->outer_mode.family == AF_INET)
			return __xfrm_transport_prep(x, skb,
						     sizeof(struct iphdr));
		if (x->outer_mode.family == AF_INET6)
			return __xfrm_transport_prep(x, skb,
						     sizeof(struct ipv6hdr));
		break;
	case XFRM_MODE_BEET:
		if (x->outer_mode.family == AF_INET)
			return __xfrm_mode_beet_prep(x, skb,
						     sizeof(struct iphdr));
		if (x->outer_mode.family == AF_INET6)
			return __xfrm_mode_beet_prep(x, skb,
						     sizeof(struct ipv6hdr));
		break;
	case XFRM_MODE_ROUTEOPTIMIZATION:
	case XFRM_MODE_IN_TRIGGER:
		break;
	}
}

struct sk_buff *validate_xmit_xfrm(struct sk_buff *skb, netdev_features_t features, bool *again)
{
	int err;
	unsigned long flags;
	struct xfrm_state *x;
	struct softnet_data *sd;
	struct sk_buff *skb2, *nskb, *pskb = NULL;
	netdev_features_t esp_features = features;
	struct xfrm_offload *xo = xfrm_offload(skb);
	struct net_device *dev = skb->dev;
	struct sec_path *sp;

	if (!xo || (xo->flags & XFRM_XMIT))
		return skb;

	if (!(features & NETIF_F_HW_ESP))
		esp_features = features & ~(NETIF_F_SG | NETIF_F_CSUM_MASK);

	sp = skb_sec_path(skb);
	x = sp->xvec[sp->len - 1];
	if (xo->flags & XFRM_GRO || x->xso.flags & XFRM_OFFLOAD_INBOUND)
		return skb;

	/* This skb was already validated on the upper/virtual dev */
	if ((x->xso.dev != dev) && (x->xso.real_dev == dev))
		return skb;

	local_irq_save(flags);
	sd = this_cpu_ptr(&softnet_data);
	err = !skb_queue_empty(&sd->xfrm_backlog);
	local_irq_restore(flags);

	if (err) {
		*again = true;
		return skb;
	}

	if (skb_is_gso(skb) && unlikely(x->xso.dev != dev)) {
		struct sk_buff *segs;

		/* Packet got rerouted, fixup features and segment it. */
		esp_features = esp_features & ~(NETIF_F_HW_ESP | NETIF_F_GSO_ESP);

		segs = skb_gso_segment(skb, esp_features);
		if (IS_ERR(segs)) {
			kfree_skb(skb);
			atomic_long_inc(&dev->tx_dropped);
			return NULL;
		} else {
			consume_skb(skb);
			skb = segs;
		}
	}

	if (!skb->next) {
		esp_features |= skb->dev->gso_partial_features;
		xfrm_outer_mode_prep(x, skb);

		xo->flags |= XFRM_DEV_RESUME;

		err = x->type_offload->xmit(x, skb, esp_features);
		if (err) {
			if (err == -EINPROGRESS)
				return NULL;

			XFRM_INC_STATS(xs_net(x), LINUX_MIB_XFRMOUTSTATEPROTOERROR);
			kfree_skb(skb);
			return NULL;
		}

		skb_push(skb, skb->data - skb_mac_header(skb));

		return skb;
	}

	skb_list_walk_safe(skb, skb2, nskb) {
		esp_features |= skb->dev->gso_partial_features;
		skb_mark_not_on_list(skb2);

		xo = xfrm_offload(skb2);
		xo->flags |= XFRM_DEV_RESUME;

		xfrm_outer_mode_prep(x, skb2);

		err = x->type_offload->xmit(x, skb2, esp_features);
		if (!err) {
			skb2->next = nskb;
		} else if (err != -EINPROGRESS) {
			XFRM_INC_STATS(xs_net(x), LINUX_MIB_XFRMOUTSTATEPROTOERROR);
			skb2->next = nskb;
			kfree_skb_list(skb2);
			return NULL;
		} else {
			if (skb == skb2)
				skb = nskb;
			else
				pskb->next = nskb;

			continue;
		}

		skb_push(skb2, skb2->data - skb_mac_header(skb2));
		pskb = skb2;
	}

	return skb;
}
EXPORT_SYMBOL_GPL(validate_xmit_xfrm);

int xfrm_dev_state_add(struct net *net, struct xfrm_state *x,
		       struct xfrm_user_offload *xuo)
{
	int err;
	struct dst_entry *dst;
	struct net_device *dev;
	struct xfrm_state_offload *xso = &x->xso;
	xfrm_address_t *saddr;
	xfrm_address_t *daddr;

	if (!x->type_offload)
		return -EINVAL;

	/* We don't yet support UDP encapsulation and TFC padding. */
	if (x->encap || x->tfcpad)
		return -EINVAL;

	dev = dev_get_by_index(net, xuo->ifindex);
	if (!dev) {
		if (!(xuo->flags & XFRM_OFFLOAD_INBOUND)) {
			saddr = &x->props.saddr;
			daddr = &x->id.daddr;
		} else {
			saddr = &x->id.daddr;
			daddr = &x->props.saddr;
		}

		dst = __xfrm_dst_lookup(net, 0, 0, saddr, daddr,
					x->props.family,
					xfrm_smark_get(0, x));
		if (IS_ERR(dst))
			return 0;

		dev = dst->dev;

		dev_hold(dev);
		dst_release(dst);
	}

	if (!dev->xfrmdev_ops || !dev->xfrmdev_ops->xdo_dev_state_add) {
		xso->dev = NULL;
		dev_put(dev);
		return 0;
	}

	if (x->props.flags & XFRM_STATE_ESN &&
	    !dev->xfrmdev_ops->xdo_dev_state_advance_esn) {
		xso->dev = NULL;
		dev_put(dev);
		return -EINVAL;
	}

	xso->dev = dev;
	xso->real_dev = dev;
	xso->num_exthdrs = 1;
	xso->flags = xuo->flags;

	err = dev->xfrmdev_ops->xdo_dev_state_add(x);
	if (err) {
		xso->num_exthdrs = 0;
		xso->flags = 0;
		xso->dev = NULL;
		xso->real_dev = NULL;
		dev_put(dev);

		if (err != -EOPNOTSUPP)
			return err;
	}

	return 0;
}
EXPORT_SYMBOL_GPL(xfrm_dev_state_add);

bool xfrm_dev_offload_ok(struct sk_buff *skb, struct xfrm_state *x)
{
	int mtu;
	struct dst_entry *dst = skb_dst(skb);
	struct xfrm_dst *xdst = (struct xfrm_dst *)dst;
	struct net_device *dev = x->xso.dev;

	if (!x->type_offload || x->encap)
		return false;

	if ((!dev || (dev == xfrm_dst_path(dst)->dev)) &&
	    (!xdst->child->xfrm)) {
		mtu = xfrm_state_mtu(x, xdst->child_mtu_cached);
		if (skb->len <= mtu)
			goto ok;

		if (skb_is_gso(skb) && skb_gso_validate_network_len(skb, mtu))
			goto ok;
	}

	return false;

ok:
	if (dev && dev->xfrmdev_ops && dev->xfrmdev_ops->xdo_dev_offload_ok)
		return x->xso.dev->xfrmdev_ops->xdo_dev_offload_ok(skb, x);

	return true;
}
EXPORT_SYMBOL_GPL(xfrm_dev_offload_ok);

void xfrm_dev_resume(struct sk_buff *skb)
{
	struct net_device *dev = skb->dev;
	int ret = NETDEV_TX_BUSY;
	struct netdev_queue *txq;
	struct softnet_data *sd;
	unsigned long flags;

	rcu_read_lock();
	txq = netdev_core_pick_tx(dev, skb, NULL);

	HARD_TX_LOCK(dev, txq, smp_processor_id());
	if (!netif_xmit_frozen_or_stopped(txq))
		skb = dev_hard_start_xmit(skb, dev, txq, &ret);
	HARD_TX_UNLOCK(dev, txq);

	if (!dev_xmit_complete(ret)) {
		local_irq_save(flags);
		sd = this_cpu_ptr(&softnet_data);
		skb_queue_tail(&sd->xfrm_backlog, skb);
		raise_softirq_irqoff(NET_TX_SOFTIRQ);
		local_irq_restore(flags);
	}
	rcu_read_unlock();
}
EXPORT_SYMBOL_GPL(xfrm_dev_resume);

void xfrm_dev_backlog(struct softnet_data *sd)
{
	struct sk_buff_head *xfrm_backlog = &sd->xfrm_backlog;
	struct sk_buff_head list;
	struct sk_buff *skb;

	if (skb_queue_empty(xfrm_backlog))
		return;

	__skb_queue_head_init(&list);

	spin_lock(&xfrm_backlog->lock);
	skb_queue_splice_init(xfrm_backlog, &list);
	spin_unlock(&xfrm_backlog->lock);

	while (!skb_queue_empty(&list)) {
		skb = __skb_dequeue(&list);
		xfrm_dev_resume(skb);
	}

}
#endif

static int xfrm_api_check(struct net_device *dev)
{
#ifdef CONFIG_XFRM_OFFLOAD
	if ((dev->features & NETIF_F_HW_ESP_TX_CSUM) &&
	    !(dev->features & NETIF_F_HW_ESP))
		return NOTIFY_BAD;

	if ((dev->features & NETIF_F_HW_ESP) &&
	    (!(dev->xfrmdev_ops &&
	       dev->xfrmdev_ops->xdo_dev_state_add &&
	       dev->xfrmdev_ops->xdo_dev_state_delete)))
		return NOTIFY_BAD;
#else
	if (dev->features & (NETIF_F_HW_ESP | NETIF_F_HW_ESP_TX_CSUM))
		return NOTIFY_BAD;
#endif

	return NOTIFY_DONE;
}

static int xfrm_dev_register(struct net_device *dev)
{
	return xfrm_api_check(dev);
}

static int xfrm_dev_feat_change(struct net_device *dev)
{
	return xfrm_api_check(dev);
}

static int xfrm_dev_down(struct net_device *dev)
{
	if (dev->features & NETIF_F_HW_ESP)
		xfrm_dev_state_flush(dev_net(dev), dev, true);

	return NOTIFY_DONE;
}

static int xfrm_dev_event(struct notifier_block *this, unsigned long event, void *ptr)
{
	struct net_device *dev = netdev_notifier_info_to_dev(ptr);

	switch (event) {
	case NETDEV_REGISTER:
		return xfrm_dev_register(dev);

	case NETDEV_FEAT_CHANGE:
		return xfrm_dev_feat_change(dev);

	case NETDEV_DOWN:
	case NETDEV_UNREGISTER:
		return xfrm_dev_down(dev);
	}
	return NOTIFY_DONE;
}

static struct notifier_block xfrm_dev_notifier = {
	.notifier_call	= xfrm_dev_event,
};

void __init xfrm_dev_init(void)
{
	register_netdevice_notifier(&xfrm_dev_notifier);
}
Commit	Line	Data
2874c5fd	1	// SPDX-License-Identifier: GPL-2.0-or-later
21f42cc9 SK	2	/*
	3	* xfrm_device.c - IPsec device offloading code.
	4	*
	5	* Copyright (c) 2015 secunet Security Networks AG
	6	*
	7	* Author:
	8	* Steffen Klassert <[email protected]>
21f42cc9 SK	9	*/
	10
	11	#include <linux/errno.h>
	12	#include <linux/module.h>
	13	#include <linux/netdevice.h>
	14	#include <linux/skbuff.h>
	15	#include <linux/slab.h>
	16	#include <linux/spinlock.h>
	17	#include <net/dst.h>
	18	#include <net/xfrm.h>
	19	#include <linux/notifier.h>
	20
b81f884a	21	#ifdef CONFIG_XFRM_OFFLOAD
303c5fab FW	22	static void __xfrm_transport_prep(struct xfrm_state x, struct sk_buff skb,
	23	unsigned int hsize)
	24	{
	25	struct xfrm_offload *xo = xfrm_offload(skb);
	26
	27	skb_reset_mac_len(skb);
06a0afcf	28	if (xo->flags & XFRM_GSO_SEGMENT)
303c5fab	29	skb->transport_header -= x->props.header_len;
06a0afcf XL	30
06a0afcf XL	31	pskb_pull(skb, skb_transport_offset(skb) + x->props.header_len);
303c5fab FW	32	}
	33
	34	static void __xfrm_mode_tunnel_prep(struct xfrm_state x, struct sk_buff skb,
	35	unsigned int hsize)
	36
	37	{
	38	struct xfrm_offload *xo = xfrm_offload(skb);
	39
	40	if (xo->flags & XFRM_GSO_SEGMENT)
	41	skb->transport_header = skb->network_header + hsize;
	42
	43	skb_reset_mac_len(skb);
	44	pskb_pull(skb, skb->mac_len + x->props.header_len);
	45	}
	46
30849175 XL	47	static void __xfrm_mode_beet_prep(struct xfrm_state x, struct sk_buff skb,
	48	unsigned int hsize)
	49	{
	50	struct xfrm_offload *xo = xfrm_offload(skb);
	51	int phlen = 0;
	52
	53	if (xo->flags & XFRM_GSO_SEGMENT)
	54	skb->transport_header = skb->network_header + hsize;
	55
	56	skb_reset_mac_len(skb);
	57	if (x->sel.family != AF_INET6) {
	58	phlen = IPV4_BEET_PHMAXLEN;
	59	if (x->outer_mode.family == AF_INET6)
	60	phlen += sizeof(struct ipv6hdr) - sizeof(struct iphdr);
	61	}
	62
	63	pskb_pull(skb, skb->mac_len + hsize + (x->props.header_len - phlen));
	64	}
	65
303c5fab FW	66	/* Adjust pointers into the packet when IPsec is done at layer2 */
	67	static void xfrm_outer_mode_prep(struct xfrm_state x, struct sk_buff skb)
	68	{
c9500d7b	69	switch (x->outer_mode.encap) {
303c5fab	70	case XFRM_MODE_TUNNEL:
c9500d7b	71	if (x->outer_mode.family == AF_INET)
303c5fab FW	72	return __xfrm_mode_tunnel_prep(x, skb,
303c5fab FW	73	sizeof(struct iphdr));
c9500d7b	74	if (x->outer_mode.family == AF_INET6)
303c5fab FW	75	return __xfrm_mode_tunnel_prep(x, skb,
	76	sizeof(struct ipv6hdr));
	77	break;
	78	case XFRM_MODE_TRANSPORT:
c9500d7b	79	if (x->outer_mode.family == AF_INET)
303c5fab FW	80	return __xfrm_transport_prep(x, skb,
303c5fab FW	81	sizeof(struct iphdr));
c9500d7b	82	if (x->outer_mode.family == AF_INET6)
303c5fab FW	83	return __xfrm_transport_prep(x, skb,
	84	sizeof(struct ipv6hdr));
	85	break;
30849175 XL	86	case XFRM_MODE_BEET:
	87	if (x->outer_mode.family == AF_INET)
	88	return __xfrm_mode_beet_prep(x, skb,
	89	sizeof(struct iphdr));
	90	if (x->outer_mode.family == AF_INET6)
	91	return __xfrm_mode_beet_prep(x, skb,
	92	sizeof(struct ipv6hdr));
	93	break;
303c5fab FW	94	case XFRM_MODE_ROUTEOPTIMIZATION:
303c5fab FW	95	case XFRM_MODE_IN_TRIGGER:
303c5fab FW	96	break;
	97	}
	98	}
	99
f53c7239	100	struct sk_buff validate_xmit_xfrm(struct sk_buff skb, netdev_features_t features, bool *again)
f6e27114 SK	101	{
f6e27114 SK	102	int err;
f53c7239	103	unsigned long flags;
f6e27114	104	struct xfrm_state *x;
f53c7239	105	struct softnet_data *sd;
d1d17a35	106	struct sk_buff skb2, nskb, *pskb = NULL;
3dca3f38	107	netdev_features_t esp_features = features;
f6e27114	108	struct xfrm_offload *xo = xfrm_offload(skb);
272c2330	109	struct net_device *dev = skb->dev;
2294be0f	110	struct sec_path *sp;
f6e27114	111
94579ac3	112	if (!xo \|\| (xo->flags & XFRM_XMIT))
3dca3f38	113	return skb;
f6e27114	114
3dca3f38 SK	115	if (!(features & NETIF_F_HW_ESP))
3dca3f38 SK	116	esp_features = features & ~(NETIF_F_SG \| NETIF_F_CSUM_MASK);
f6e27114	117
2294be0f FW	118	sp = skb_sec_path(skb);
2294be0f FW	119	x = sp->xvec[sp->len - 1];
3dca3f38 SK	120	if (xo->flags & XFRM_GRO \|\| x->xso.flags & XFRM_OFFLOAD_INBOUND)
	121	return skb;
	122
bdfd2d1f JW	123	/* This skb was already validated on the upper/virtual dev */
bdfd2d1f JW	124	if ((x->xso.dev != dev) && (x->xso.real_dev == dev))
272c2330 JW	125	return skb;
272c2330 JW	126
f53c7239 SK	127	local_irq_save(flags);
	128	sd = this_cpu_ptr(&softnet_data);
	129	err = !skb_queue_empty(&sd->xfrm_backlog);
	130	local_irq_restore(flags);
	131
	132	if (err) {
	133	*again = true;
	134	return skb;
	135	}
	136
272c2330 JW	137	if (skb_is_gso(skb) && unlikely(x->xso.dev != dev)) {
272c2330 JW	138	struct sk_buff *segs;
3dca3f38	139
272c2330 JW	140	/* Packet got rerouted, fixup features and segment it. */
272c2330 JW	141	esp_features = esp_features & ~(NETIF_F_HW_ESP \| NETIF_F_GSO_ESP);
3dca3f38	142
272c2330 JW	143	segs = skb_gso_segment(skb, esp_features);
	144	if (IS_ERR(segs)) {
	145	kfree_skb(skb);
	146	atomic_long_inc(&dev->tx_dropped);
	147	return NULL;
	148	} else {
	149	consume_skb(skb);
	150	skb = segs;
3dca3f38 SK	151	}
	152	}
	153
	154	if (!skb->next) {
65fd2c2a	155	esp_features \|= skb->dev->gso_partial_features;
303c5fab	156	xfrm_outer_mode_prep(x, skb);
f6e27114	157
f53c7239 SK	158	xo->flags \|= XFRM_DEV_RESUME;
f53c7239 SK	159
3dca3f38	160	err = x->type_offload->xmit(x, skb, esp_features);
f6e27114	161	if (err) {
f53c7239 SK	162	if (err == -EINPROGRESS)
	163	return NULL;
	164
f6e27114	165	XFRM_INC_STATS(xs_net(x), LINUX_MIB_XFRMOUTSTATEPROTOERROR);
3dca3f38 SK	166	kfree_skb(skb);
3dca3f38 SK	167	return NULL;
f6e27114 SK	168	}
	169
	170	skb_push(skb, skb->data - skb_mac_header(skb));
3dca3f38 SK	171
3dca3f38 SK	172	return skb;
f6e27114 SK	173	}
f6e27114 SK	174
c3b18e0d	175	skb_list_walk_safe(skb, skb2, nskb) {
65fd2c2a	176	esp_features \|= skb->dev->gso_partial_features;
a8305bff	177	skb_mark_not_on_list(skb2);
3dca3f38 SK	178
3dca3f38 SK	179	xo = xfrm_offload(skb2);
f53c7239	180	xo->flags \|= XFRM_DEV_RESUME;
3dca3f38	181
303c5fab	182	xfrm_outer_mode_prep(x, skb2);
3dca3f38 SK	183
3dca3f38 SK	184	err = x->type_offload->xmit(x, skb2, esp_features);
f53c7239 SK	185	if (!err) {
	186	skb2->next = nskb;
	187	} else if (err != -EINPROGRESS) {
3dca3f38 SK	188	XFRM_INC_STATS(xs_net(x), LINUX_MIB_XFRMOUTSTATEPROTOERROR);
	189	skb2->next = nskb;
	190	kfree_skb_list(skb2);
	191	return NULL;
f53c7239 SK	192	} else {
	193	if (skb == skb2)
	194	skb = nskb;
d1d17a35 XL	195	else
d1d17a35 XL	196	pskb->next = nskb;
3dca3f38	197
c3b18e0d	198	continue;
f53c7239	199	}
3dca3f38 SK	200
3dca3f38 SK	201	skb_push(skb2, skb2->data - skb_mac_header(skb2));
d1d17a35	202	pskb = skb2;
c3b18e0d	203	}
3dca3f38 SK	204
3dca3f38 SK	205	return skb;
f6e27114 SK	206	}
	207	EXPORT_SYMBOL_GPL(validate_xmit_xfrm);
	208
d77e38e6 SK	209	int xfrm_dev_state_add(struct net net, struct xfrm_state x,
	210	struct xfrm_user_offload *xuo)
	211	{
	212	int err;
	213	struct dst_entry *dst;
	214	struct net_device *dev;
	215	struct xfrm_state_offload *xso = &x->xso;
	216	xfrm_address_t *saddr;
	217	xfrm_address_t *daddr;
	218
	219	if (!x->type_offload)
ffdb5211	220	return -EINVAL;
d77e38e6	221
50bd870a YE	222	/* We don't yet support UDP encapsulation and TFC padding. */
50bd870a YE	223	if (x->encap \|\| x->tfcpad)
43024b9c	224	return -EINVAL;
d77e38e6 SK	225
	226	dev = dev_get_by_index(net, xuo->ifindex);
	227	if (!dev) {
	228	if (!(xuo->flags & XFRM_OFFLOAD_INBOUND)) {
	229	saddr = &x->props.saddr;
	230	daddr = &x->id.daddr;
	231	} else {
	232	saddr = &x->id.daddr;
	233	daddr = &x->props.saddr;
	234	}
	235
077fbac4	236	dst = __xfrm_dst_lookup(net, 0, 0, saddr, daddr,
9b42c1f1 SK	237	x->props.family,
9b42c1f1 SK	238	xfrm_smark_get(0, x));
d77e38e6 SK	239	if (IS_ERR(dst))
	240	return 0;
	241
	242	dev = dst->dev;
	243
	244	dev_hold(dev);
	245	dst_release(dst);
	246	}
	247
	248	if (!dev->xfrmdev_ops \|\| !dev->xfrmdev_ops->xdo_dev_state_add) {
67a63387	249	xso->dev = NULL;
d77e38e6 SK	250	dev_put(dev);
	251	return 0;
	252	}
	253
50bd870a YE	254	if (x->props.flags & XFRM_STATE_ESN &&
	255	!dev->xfrmdev_ops->xdo_dev_state_advance_esn) {
	256	xso->dev = NULL;
	257	dev_put(dev);
	258	return -EINVAL;
	259	}
	260
d77e38e6	261	xso->dev = dev;
bdfd2d1f	262	xso->real_dev = dev;
d77e38e6 SK	263	xso->num_exthdrs = 1;
	264	xso->flags = xuo->flags;
	265
	266	err = dev->xfrmdev_ops->xdo_dev_state_add(x);
	267	if (err) {
4a132095 SN	268	xso->num_exthdrs = 0;
4a132095 SN	269	xso->flags = 0;
aa5dd6fa	270	xso->dev = NULL;
dd72fadf	271	xso->real_dev = NULL;
d77e38e6	272	dev_put(dev);
4a132095 SN	273
	274	if (err != -EOPNOTSUPP)
	275	return err;
d77e38e6 SK	276	}
	277
	278	return 0;
	279	}
	280	EXPORT_SYMBOL_GPL(xfrm_dev_state_add);
	281
	282	bool xfrm_dev_offload_ok(struct sk_buff skb, struct xfrm_state x)
	283	{
	284	int mtu;
	285	struct dst_entry *dst = skb_dst(skb);
	286	struct xfrm_dst xdst = (struct xfrm_dst )dst;
	287	struct net_device *dev = x->xso.dev;
	288
	289	if (!x->type_offload \|\| x->encap)
	290	return false;
	291
fcb662de	292	if ((!dev \|\| (dev == xfrm_dst_path(dst)->dev)) &&
c7b37c76 FW	293	(!xdst->child->xfrm)) {
c7b37c76 FW	294	mtu = xfrm_state_mtu(x, xdst->child_mtu_cached);
d77e38e6 SK	295	if (skb->len <= mtu)
	296	goto ok;
	297
779b7931	298	if (skb_is_gso(skb) && skb_gso_validate_network_len(skb, mtu))
d77e38e6 SK	299	goto ok;
	300	}
	301
	302	return false;
	303
	304	ok:
	305	if (dev && dev->xfrmdev_ops && dev->xfrmdev_ops->xdo_dev_offload_ok)
	306	return x->xso.dev->xfrmdev_ops->xdo_dev_offload_ok(skb, x);
	307
	308	return true;
	309	}
	310	EXPORT_SYMBOL_GPL(xfrm_dev_offload_ok);
f53c7239 SK	311
	312	void xfrm_dev_resume(struct sk_buff *skb)
	313	{
	314	struct net_device *dev = skb->dev;
	315	int ret = NETDEV_TX_BUSY;
	316	struct netdev_queue *txq;
	317	struct softnet_data *sd;
	318	unsigned long flags;
	319
	320	rcu_read_lock();
4bd97d51	321	txq = netdev_core_pick_tx(dev, skb, NULL);
f53c7239 SK	322
	323	HARD_TX_LOCK(dev, txq, smp_processor_id());
	324	if (!netif_xmit_frozen_or_stopped(txq))
	325	skb = dev_hard_start_xmit(skb, dev, txq, &ret);
	326	HARD_TX_UNLOCK(dev, txq);
	327
	328	if (!dev_xmit_complete(ret)) {
	329	local_irq_save(flags);
	330	sd = this_cpu_ptr(&softnet_data);
	331	skb_queue_tail(&sd->xfrm_backlog, skb);
	332	raise_softirq_irqoff(NET_TX_SOFTIRQ);
	333	local_irq_restore(flags);
	334	}
	335	rcu_read_unlock();
	336	}
	337	EXPORT_SYMBOL_GPL(xfrm_dev_resume);
	338
	339	void xfrm_dev_backlog(struct softnet_data *sd)
	340	{
	341	struct sk_buff_head *xfrm_backlog = &sd->xfrm_backlog;
	342	struct sk_buff_head list;
	343	struct sk_buff *skb;
	344
	345	if (skb_queue_empty(xfrm_backlog))
	346	return;
	347
	348	__skb_queue_head_init(&list);
	349
	350	spin_lock(&xfrm_backlog->lock);
	351	skb_queue_splice_init(xfrm_backlog, &list);
	352	spin_unlock(&xfrm_backlog->lock);
	353
	354	while (!skb_queue_empty(&list)) {
	355	skb = __skb_dequeue(&list);
	356	xfrm_dev_resume(skb);
	357	}
	358
	359	}
b81f884a	360	#endif
d77e38e6	361
92a23206	362	static int xfrm_api_check(struct net_device *dev)
d77e38e6	363	{
92a23206	364	#ifdef CONFIG_XFRM_OFFLOAD
d77e38e6 SK	365	if ((dev->features & NETIF_F_HW_ESP_TX_CSUM) &&
	366	!(dev->features & NETIF_F_HW_ESP))
	367	return NOTIFY_BAD;
	368
92a23206 SN	369	if ((dev->features & NETIF_F_HW_ESP) &&
	370	(!(dev->xfrmdev_ops &&
	371	dev->xfrmdev_ops->xdo_dev_state_add &&
	372	dev->xfrmdev_ops->xdo_dev_state_delete)))
	373	return NOTIFY_BAD;
	374	#else
	375	if (dev->features & (NETIF_F_HW_ESP \| NETIF_F_HW_ESP_TX_CSUM))
	376	return NOTIFY_BAD;
	377	#endif
	378
d77e38e6 SK	379	return NOTIFY_DONE;
	380	}
	381
92a23206 SN	382	static int xfrm_dev_register(struct net_device *dev)
	383	{
	384	return xfrm_api_check(dev);
	385	}
	386
d77e38e6 SK	387	static int xfrm_dev_feat_change(struct net_device *dev)
d77e38e6 SK	388	{
92a23206	389	return xfrm_api_check(dev);
d77e38e6 SK	390	}
	391
	392	static int xfrm_dev_down(struct net_device *dev)
	393	{
2c1497bb	394	if (dev->features & NETIF_F_HW_ESP)
d77e38e6 SK	395	xfrm_dev_state_flush(dev_net(dev), dev, true);
d77e38e6 SK	396
d77e38e6 SK	397	return NOTIFY_DONE;
	398	}
	399
21f42cc9 SK	400	static int xfrm_dev_event(struct notifier_block this, unsigned long event, void ptr)
	401	{
	402	struct net_device *dev = netdev_notifier_info_to_dev(ptr);
	403
	404	switch (event) {
d77e38e6 SK	405	case NETDEV_REGISTER:
	406	return xfrm_dev_register(dev);
	407
d77e38e6 SK	408	case NETDEV_FEAT_CHANGE:
	409	return xfrm_dev_feat_change(dev);
	410
21f42cc9	411	case NETDEV_DOWN:
03891f82	412	case NETDEV_UNREGISTER:
d77e38e6	413	return xfrm_dev_down(dev);
21f42cc9 SK	414	}
	415	return NOTIFY_DONE;
	416	}
	417
	418	static struct notifier_block xfrm_dev_notifier = {
	419	.notifier_call = xfrm_dev_event,
	420	};
	421
e9a441b6	422	void __init xfrm_dev_init(void)
21f42cc9 SK	423	{
	424	register_netdevice_notifier(&xfrm_dev_notifier);
	425	}