[people/ms/linux.git] / net / ipv4 / udp_offload.c

/*
 *	IPV4 GSO/GRO offload support
 *	Linux INET implementation
 *
 *	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.
 *
 *	UDPv4 GSO support
 */

#include <linux/skbuff.h>
#include <net/udp.h>
#include <net/protocol.h>

static DEFINE_SPINLOCK(udp_offload_lock);
static struct udp_offload_priv __rcu *udp_offload_base __read_mostly;

#define udp_deref_protected(X) rcu_dereference_protected(X, lockdep_is_held(&udp_offload_lock))

struct udp_offload_priv {
	struct udp_offload	*offload;
	struct rcu_head		rcu;
	struct udp_offload_priv __rcu *next;
};

static int udp4_ufo_send_check(struct sk_buff *skb)
{
	if (!pskb_may_pull(skb, sizeof(struct udphdr)))
		return -EINVAL;

	if (likely(!skb->encapsulation)) {
		const struct iphdr *iph;
		struct udphdr *uh;

		iph = ip_hdr(skb);
		uh = udp_hdr(skb);

		uh->check = ~csum_tcpudp_magic(iph->saddr, iph->daddr, skb->len,
				IPPROTO_UDP, 0);
		skb->csum_start = skb_transport_header(skb) - skb->head;
		skb->csum_offset = offsetof(struct udphdr, check);
		skb->ip_summed = CHECKSUM_PARTIAL;
	}

	return 0;
}

static struct sk_buff *udp4_ufo_fragment(struct sk_buff *skb,
					 netdev_features_t features)
{
	struct sk_buff *segs = ERR_PTR(-EINVAL);
	unsigned int mss;
	int offset;
	__wsum csum;

	if (skb->encapsulation &&
	    (skb_shinfo(skb)->gso_type &
	     (SKB_GSO_UDP_TUNNEL|SKB_GSO_UDP_TUNNEL_CSUM))) {
		segs = skb_udp_tunnel_segment(skb, features);
		goto out;
	}

	mss = skb_shinfo(skb)->gso_size;
	if (unlikely(skb->len <= mss))
		goto out;

	if (skb_gso_ok(skb, features | NETIF_F_GSO_ROBUST)) {
		/* Packet is from an untrusted source, reset gso_segs. */
		int type = skb_shinfo(skb)->gso_type;

		if (unlikely(type & ~(SKB_GSO_UDP | SKB_GSO_DODGY |
				      SKB_GSO_UDP_TUNNEL |
				      SKB_GSO_UDP_TUNNEL_CSUM |
				      SKB_GSO_IPIP |
				      SKB_GSO_GRE | SKB_GSO_GRE_CSUM |
				      SKB_GSO_MPLS) ||
			     !(type & (SKB_GSO_UDP))))
			goto out;

		skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);

		segs = NULL;
		goto out;
	}

	/* Do software UFO. Complete and fill in the UDP checksum as
	 * HW cannot do checksum of UDP packets sent as multiple
	 * IP fragments.
	 */
	offset = skb_checksum_start_offset(skb);
	csum = skb_checksum(skb, offset, skb->len - offset, 0);
	offset += skb->csum_offset;
	*(__sum16 *)(skb->data + offset) = csum_fold(csum);
	skb->ip_summed = CHECKSUM_NONE;

	/* Fragment the skb. IP headers of the fragments are updated in
	 * inet_gso_segment()
	 */
	segs = skb_segment(skb, features);
out:
	return segs;
}

int udp_add_offload(struct udp_offload *uo)
{
	struct udp_offload_priv *new_offload = kzalloc(sizeof(*new_offload), GFP_ATOMIC);

	if (!new_offload)
		return -ENOMEM;

	new_offload->offload = uo;

	spin_lock(&udp_offload_lock);
	new_offload->next = udp_offload_base;
	rcu_assign_pointer(udp_offload_base, new_offload);
	spin_unlock(&udp_offload_lock);

	return 0;
}
EXPORT_SYMBOL(udp_add_offload);

static void udp_offload_free_routine(struct rcu_head *head)
{
	struct udp_offload_priv *ou_priv = container_of(head, struct udp_offload_priv, rcu);
	kfree(ou_priv);
}

void udp_del_offload(struct udp_offload *uo)
{
	struct udp_offload_priv __rcu **head = &udp_offload_base;
	struct udp_offload_priv *uo_priv;

	spin_lock(&udp_offload_lock);

	uo_priv = udp_deref_protected(*head);
	for (; uo_priv != NULL;
	     uo_priv = udp_deref_protected(*head)) {
		if (uo_priv->offload == uo) {
			rcu_assign_pointer(*head,
					   udp_deref_protected(uo_priv->next));
			goto unlock;
		}
		head = &uo_priv->next;
	}
	pr_warn("udp_del_offload: didn't find offload for port %d\n", ntohs(uo->port));
unlock:
	spin_unlock(&udp_offload_lock);
	if (uo_priv != NULL)
		call_rcu(&uo_priv->rcu, udp_offload_free_routine);
}
EXPORT_SYMBOL(udp_del_offload);

static struct sk_buff **udp_gro_receive(struct sk_buff **head, struct sk_buff *skb)
{
	struct udp_offload_priv *uo_priv;
	struct sk_buff *p, **pp = NULL;
	struct udphdr *uh, *uh2;
	unsigned int hlen, off;
	int flush = 1;

	if (NAPI_GRO_CB(skb)->udp_mark ||
	    (!skb->encapsulation && skb->ip_summed != CHECKSUM_COMPLETE))
		goto out;

	/* mark that this skb passed once through the udp gro layer */
	NAPI_GRO_CB(skb)->udp_mark = 1;

	off  = skb_gro_offset(skb);
	hlen = off + sizeof(*uh);
	uh   = skb_gro_header_fast(skb, off);
	if (skb_gro_header_hard(skb, hlen)) {
		uh = skb_gro_header_slow(skb, hlen, off);
		if (unlikely(!uh))
			goto out;
	}

	rcu_read_lock();
	uo_priv = rcu_dereference(udp_offload_base);
	for (; uo_priv != NULL; uo_priv = rcu_dereference(uo_priv->next)) {
		if (uo_priv->offload->port == uh->dest &&
		    uo_priv->offload->callbacks.gro_receive)
			goto unflush;
	}
	goto out_unlock;

unflush:
	flush = 0;

	for (p = *head; p; p = p->next) {
		if (!NAPI_GRO_CB(p)->same_flow)
			continue;

		uh2 = (struct udphdr   *)(p->data + off);
		if ((*(u32 *)&uh->source != *(u32 *)&uh2->source)) {
			NAPI_GRO_CB(p)->same_flow = 0;
			continue;
		}
	}

	skb_gro_pull(skb, sizeof(struct udphdr)); /* pull encapsulating udp header */
	skb_gro_postpull_rcsum(skb, uh, sizeof(struct udphdr));
	pp = uo_priv->offload->callbacks.gro_receive(head, skb);

out_unlock:
	rcu_read_unlock();
out:
	NAPI_GRO_CB(skb)->flush |= flush;
	return pp;
}

static int udp_gro_complete(struct sk_buff *skb, int nhoff)
{
	struct udp_offload_priv *uo_priv;
	__be16 newlen = htons(skb->len - nhoff);
	struct udphdr *uh = (struct udphdr *)(skb->data + nhoff);
	int err = -ENOSYS;

	uh->len = newlen;

	rcu_read_lock();

	uo_priv = rcu_dereference(udp_offload_base);
	for (; uo_priv != NULL; uo_priv = rcu_dereference(uo_priv->next)) {
		if (uo_priv->offload->port == uh->dest &&
		    uo_priv->offload->callbacks.gro_complete)
			break;
	}

	if (uo_priv != NULL)
		err = uo_priv->offload->callbacks.gro_complete(skb, nhoff + sizeof(struct udphdr));

	rcu_read_unlock();
	return err;
}

static const struct net_offload udpv4_offload = {
	.callbacks = {
		.gso_send_check = udp4_ufo_send_check,
		.gso_segment = udp4_ufo_fragment,
		.gro_receive  =	udp_gro_receive,
		.gro_complete =	udp_gro_complete,
	},
};

int __init udpv4_offload_init(void)
{
	return inet_add_offload(&udpv4_offload, IPPROTO_UDP);
}
Commit	Line	Data
da5bab07 DB	1	/*
	2	* IPV4 GSO/GRO offload support
	3	* Linux INET implementation
	4	*
	5	* This program is free software; you can redistribute it and/or
	6	* modify it under the terms of the GNU General Public License
	7	* as published by the Free Software Foundation; either version
	8	* 2 of the License, or (at your option) any later version.
	9	*
	10	* UDPv4 GSO support
	11	*/
	12
	13	#include <linux/skbuff.h>
	14	#include <net/udp.h>
	15	#include <net/protocol.h>
	16
b582ef09	17	static DEFINE_SPINLOCK(udp_offload_lock);
a1d0cd8e	18	static struct udp_offload_priv __rcu *udp_offload_base __read_mostly;
b582ef09	19
a664a4f7 SP	20	#define udp_deref_protected(X) rcu_dereference_protected(X, lockdep_is_held(&udp_offload_lock))
a664a4f7 SP	21
b582ef09 OG	22	struct udp_offload_priv {
	23	struct udp_offload *offload;
	24	struct rcu_head rcu;
	25	struct udp_offload_priv __rcu *next;
	26	};
	27
da5bab07 DB	28	static int udp4_ufo_send_check(struct sk_buff *skb)
	29	{
	30	if (!pskb_may_pull(skb, sizeof(struct udphdr)))
	31	return -EINVAL;
	32
	33	if (likely(!skb->encapsulation)) {
	34	const struct iphdr *iph;
	35	struct udphdr *uh;
	36
	37	iph = ip_hdr(skb);
	38	uh = udp_hdr(skb);
	39
	40	uh->check = ~csum_tcpudp_magic(iph->saddr, iph->daddr, skb->len,
	41	IPPROTO_UDP, 0);
	42	skb->csum_start = skb_transport_header(skb) - skb->head;
	43	skb->csum_offset = offsetof(struct udphdr, check);
	44	skb->ip_summed = CHECKSUM_PARTIAL;
	45	}
	46
	47	return 0;
	48	}
	49
	50	static struct sk_buff udp4_ufo_fragment(struct sk_buff skb,
	51	netdev_features_t features)
	52	{
	53	struct sk_buff *segs = ERR_PTR(-EINVAL);
	54	unsigned int mss;
7a7ffbab WCC	55	int offset;
	56	__wsum csum;
	57
	58	if (skb->encapsulation &&
0f4f4ffa TH	59	(skb_shinfo(skb)->gso_type &
0f4f4ffa TH	60	(SKB_GSO_UDP_TUNNEL\|SKB_GSO_UDP_TUNNEL_CSUM))) {
7a7ffbab WCC	61	segs = skb_udp_tunnel_segment(skb, features);
	62	goto out;
	63	}
da5bab07 DB	64
	65	mss = skb_shinfo(skb)->gso_size;
	66	if (unlikely(skb->len <= mss))
	67	goto out;
	68
	69	if (skb_gso_ok(skb, features \| NETIF_F_GSO_ROBUST)) {
	70	/* Packet is from an untrusted source, reset gso_segs. */
	71	int type = skb_shinfo(skb)->gso_type;
	72
	73	if (unlikely(type & ~(SKB_GSO_UDP \| SKB_GSO_DODGY \|
	74	SKB_GSO_UDP_TUNNEL \|
0f4f4ffa	75	SKB_GSO_UDP_TUNNEL_CSUM \|
cb32f511	76	SKB_GSO_IPIP \|
4749c09c TH	77	SKB_GSO_GRE \| SKB_GSO_GRE_CSUM \|
4749c09c TH	78	SKB_GSO_MPLS) \|\|
da5bab07 DB	79	!(type & (SKB_GSO_UDP))))
	80	goto out;
	81
	82	skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);
	83
	84	segs = NULL;
	85	goto out;
	86	}
	87
7a7ffbab WCC	88	/* Do software UFO. Complete and fill in the UDP checksum as
	89	* HW cannot do checksum of UDP packets sent as multiple
	90	* IP fragments.
	91	*/
	92	offset = skb_checksum_start_offset(skb);
	93	csum = skb_checksum(skb, offset, skb->len - offset, 0);
	94	offset += skb->csum_offset;
	95	(__sum16 )(skb->data + offset) = csum_fold(csum);
	96	skb->ip_summed = CHECKSUM_NONE;
	97
da5bab07 DB	98	/* Fragment the skb. IP headers of the fragments are updated in
	99	* inet_gso_segment()
	100	*/
7a7ffbab	101	segs = skb_segment(skb, features);
da5bab07 DB	102	out:
	103	return segs;
	104	}
	105
b582ef09 OG	106	int udp_add_offload(struct udp_offload *uo)
b582ef09 OG	107	{
b5aaab12	108	struct udp_offload_priv new_offload = kzalloc(sizeof(new_offload), GFP_ATOMIC);
b582ef09 OG	109
	110	if (!new_offload)
	111	return -ENOMEM;
	112
	113	new_offload->offload = uo;
	114
	115	spin_lock(&udp_offload_lock);
a664a4f7 SP	116	new_offload->next = udp_offload_base;
a664a4f7 SP	117	rcu_assign_pointer(udp_offload_base, new_offload);
b582ef09 OG	118	spin_unlock(&udp_offload_lock);
	119
	120	return 0;
	121	}
	122	EXPORT_SYMBOL(udp_add_offload);
	123
	124	static void udp_offload_free_routine(struct rcu_head *head)
	125	{
	126	struct udp_offload_priv *ou_priv = container_of(head, struct udp_offload_priv, rcu);
	127	kfree(ou_priv);
	128	}
	129
	130	void udp_del_offload(struct udp_offload *uo)
	131	{
	132	struct udp_offload_priv __rcu **head = &udp_offload_base;
	133	struct udp_offload_priv *uo_priv;
	134
	135	spin_lock(&udp_offload_lock);
	136
a664a4f7	137	uo_priv = udp_deref_protected(*head);
b582ef09	138	for (; uo_priv != NULL;
a664a4f7	139	uo_priv = udp_deref_protected(*head)) {
b582ef09	140	if (uo_priv->offload == uo) {
a664a4f7 SP	141	rcu_assign_pointer(*head,
a664a4f7 SP	142	udp_deref_protected(uo_priv->next));
b582ef09 OG	143	goto unlock;
	144	}
	145	head = &uo_priv->next;
	146	}
a1d0cd8e	147	pr_warn("udp_del_offload: didn't find offload for port %d\n", ntohs(uo->port));
b582ef09 OG	148	unlock:
	149	spin_unlock(&udp_offload_lock);
	150	if (uo_priv != NULL)
	151	call_rcu(&uo_priv->rcu, udp_offload_free_routine);
	152	}
	153	EXPORT_SYMBOL(udp_del_offload);
	154
	155	static struct sk_buff udp_gro_receive(struct sk_buff head, struct sk_buff *skb)
	156	{
	157	struct udp_offload_priv *uo_priv;
	158	struct sk_buff p, *pp = NULL;
	159	struct udphdr uh, uh2;
	160	unsigned int hlen, off;
	161	int flush = 1;
	162
	163	if (NAPI_GRO_CB(skb)->udp_mark \|\|
	164	(!skb->encapsulation && skb->ip_summed != CHECKSUM_COMPLETE))
	165	goto out;
	166
	167	/* mark that this skb passed once through the udp gro layer */
	168	NAPI_GRO_CB(skb)->udp_mark = 1;
	169
	170	off = skb_gro_offset(skb);
	171	hlen = off + sizeof(*uh);
	172	uh = skb_gro_header_fast(skb, off);
	173	if (skb_gro_header_hard(skb, hlen)) {
	174	uh = skb_gro_header_slow(skb, hlen, off);
	175	if (unlikely(!uh))
	176	goto out;
	177	}
	178
	179	rcu_read_lock();
	180	uo_priv = rcu_dereference(udp_offload_base);
	181	for (; uo_priv != NULL; uo_priv = rcu_dereference(uo_priv->next)) {
	182	if (uo_priv->offload->port == uh->dest &&
	183	uo_priv->offload->callbacks.gro_receive)
	184	goto unflush;
	185	}
	186	goto out_unlock;
	187
	188	unflush:
	189	flush = 0;
	190
	191	for (p = *head; p; p = p->next) {
	192	if (!NAPI_GRO_CB(p)->same_flow)
	193	continue;
	194
	195	uh2 = (struct udphdr *)(p->data + off);
	196	if (((u32 )&uh->source != (u32 )&uh2->source)) {
	197	NAPI_GRO_CB(p)->same_flow = 0;
	198	continue;
	199	}
	200	}
	201
	202	skb_gro_pull(skb, sizeof(struct udphdr)); /* pull encapsulating udp header */
6bae1d4c	203	skb_gro_postpull_rcsum(skb, uh, sizeof(struct udphdr));
b582ef09 OG	204	pp = uo_priv->offload->callbacks.gro_receive(head, skb);
	205
	206	out_unlock:
	207	rcu_read_unlock();
	208	out:
	209	NAPI_GRO_CB(skb)->flush \|= flush;
	210	return pp;
	211	}
	212
	213	static int udp_gro_complete(struct sk_buff *skb, int nhoff)
	214	{
	215	struct udp_offload_priv *uo_priv;
	216	__be16 newlen = htons(skb->len - nhoff);
	217	struct udphdr uh = (struct udphdr )(skb->data + nhoff);
	218	int err = -ENOSYS;
	219
	220	uh->len = newlen;
	221
	222	rcu_read_lock();
	223
	224	uo_priv = rcu_dereference(udp_offload_base);
	225	for (; uo_priv != NULL; uo_priv = rcu_dereference(uo_priv->next)) {
	226	if (uo_priv->offload->port == uh->dest &&
	227	uo_priv->offload->callbacks.gro_complete)
	228	break;
	229	}
	230
	231	if (uo_priv != NULL)
	232	err = uo_priv->offload->callbacks.gro_complete(skb, nhoff + sizeof(struct udphdr));
	233
	234	rcu_read_unlock();
	235	return err;
	236	}
	237
da5bab07 DB	238	static const struct net_offload udpv4_offload = {
	239	.callbacks = {
	240	.gso_send_check = udp4_ufo_send_check,
	241	.gso_segment = udp4_ufo_fragment,
b582ef09 OG	242	.gro_receive = udp_gro_receive,
b582ef09 OG	243	.gro_complete = udp_gro_complete,
da5bab07 DB	244	},
	245	};
	246
	247	int __init udpv4_offload_init(void)
	248	{
	249	return inet_add_offload(&udpv4_offload, IPPROTO_UDP);
	250	}