[deliverable/linux.git] / net / ipv4 / tcp_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.
 *
 *	TCPv4 GSO/GRO support
 */

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

struct sk_buff *tcp_gso_segment(struct sk_buff *skb,
				netdev_features_t features)
{
	struct sk_buff *segs = ERR_PTR(-EINVAL);
	struct tcphdr *th;
	unsigned int thlen;
	unsigned int seq;
	__be32 delta;
	unsigned int oldlen;
	unsigned int mss;
	struct sk_buff *gso_skb = skb;
	__sum16 newcheck;
	bool ooo_okay, copy_destructor;

	if (!pskb_may_pull(skb, sizeof(*th)))
		goto out;

	th = tcp_hdr(skb);
	thlen = th->doff * 4;
	if (thlen < sizeof(*th))
		goto out;

	if (!pskb_may_pull(skb, thlen))
		goto out;

	oldlen = (u16)~skb->len;
	__skb_pull(skb, thlen);

	mss = tcp_skb_mss(skb);
	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_TCPV4 |
			       SKB_GSO_DODGY |
			       SKB_GSO_TCP_ECN |
			       SKB_GSO_TCPV6 |
			       SKB_GSO_GRE |
			       SKB_GSO_IPIP |
			       SKB_GSO_MPLS |
			       SKB_GSO_UDP_TUNNEL |
			       0) ||
			     !(type & (SKB_GSO_TCPV4 | SKB_GSO_TCPV6))))
			goto out;

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

		segs = NULL;
		goto out;
	}

	copy_destructor = gso_skb->destructor == tcp_wfree;
	ooo_okay = gso_skb->ooo_okay;
	/* All segments but the first should have ooo_okay cleared */
	skb->ooo_okay = 0;

	segs = skb_segment(skb, features);
	if (IS_ERR(segs))
		goto out;

	/* Only first segment might have ooo_okay set */
	segs->ooo_okay = ooo_okay;

	delta = htonl(oldlen + (thlen + mss));

	skb = segs;
	th = tcp_hdr(skb);
	seq = ntohl(th->seq);

	newcheck = ~csum_fold((__force __wsum)((__force u32)th->check +
					       (__force u32)delta));

	do {
		th->fin = th->psh = 0;
		th->check = newcheck;

		if (skb->ip_summed != CHECKSUM_PARTIAL)
			th->check =
			     csum_fold(csum_partial(skb_transport_header(skb),
						    thlen, skb->csum));

		seq += mss;
		if (copy_destructor) {
			skb->destructor = gso_skb->destructor;
			skb->sk = gso_skb->sk;
			/* {tcp|sock}_wfree() use exact truesize accounting :
			 * sum(skb->truesize) MUST be exactly be gso_skb->truesize
			 * So we account mss bytes of 'true size' for each segment.
			 * The last segment will contain the remaining.
			 */
			skb->truesize = mss;
			gso_skb->truesize -= mss;
		}
		skb = skb->next;
		th = tcp_hdr(skb);

		th->seq = htonl(seq);
		th->cwr = 0;
	} while (skb->next);

	/* Following permits TCP Small Queues to work well with GSO :
	 * The callback to TCP stack will be called at the time last frag
	 * is freed at TX completion, and not right now when gso_skb
	 * is freed by GSO engine
	 */
	if (copy_destructor) {
		swap(gso_skb->sk, skb->sk);
		swap(gso_skb->destructor, skb->destructor);
		swap(gso_skb->truesize, skb->truesize);
	}

	delta = htonl(oldlen + (skb_tail_pointer(skb) -
				skb_transport_header(skb)) +
		      skb->data_len);
	th->check = ~csum_fold((__force __wsum)((__force u32)th->check +
				(__force u32)delta));
	if (skb->ip_summed != CHECKSUM_PARTIAL)
		th->check = csum_fold(csum_partial(skb_transport_header(skb),
						   thlen, skb->csum));
out:
	return segs;
}
EXPORT_SYMBOL(tcp_gso_segment);

struct sk_buff **tcp_gro_receive(struct sk_buff **head, struct sk_buff *skb)
{
	struct sk_buff **pp = NULL;
	struct sk_buff *p;
	struct tcphdr *th;
	struct tcphdr *th2;
	unsigned int len;
	unsigned int thlen;
	__be32 flags;
	unsigned int mss = 1;
	unsigned int hlen;
	unsigned int off;
	int flush = 1;
	int i;

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

	thlen = th->doff * 4;
	if (thlen < sizeof(*th))
		goto out;

	hlen = off + thlen;
	if (skb_gro_header_hard(skb, hlen)) {
		th = skb_gro_header_slow(skb, hlen, off);
		if (unlikely(!th))
			goto out;
	}

	skb_gro_pull(skb, thlen);

	len = skb_gro_len(skb);
	flags = tcp_flag_word(th);

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

		th2 = tcp_hdr(p);

		if (*(u32 *)&th->source ^ *(u32 *)&th2->source) {
			NAPI_GRO_CB(p)->same_flow = 0;
			continue;
		}

		goto found;
	}

	goto out_check_final;

found:
	flush = NAPI_GRO_CB(p)->flush;
	flush |= (__force int)(flags & TCP_FLAG_CWR);
	flush |= (__force int)((flags ^ tcp_flag_word(th2)) &
		  ~(TCP_FLAG_CWR | TCP_FLAG_FIN | TCP_FLAG_PSH));
	flush |= (__force int)(th->ack_seq ^ th2->ack_seq);
	for (i = sizeof(*th); i < thlen; i += 4)
		flush |= *(u32 *)((u8 *)th + i) ^
			 *(u32 *)((u8 *)th2 + i);

	mss = tcp_skb_mss(p);

	flush |= (len - 1) >= mss;
	flush |= (ntohl(th2->seq) + skb_gro_len(p)) ^ ntohl(th->seq);

	if (flush || skb_gro_receive(head, skb)) {
		mss = 1;
		goto out_check_final;
	}

	p = *head;
	th2 = tcp_hdr(p);
	tcp_flag_word(th2) |= flags & (TCP_FLAG_FIN | TCP_FLAG_PSH);

out_check_final:
	flush = len < mss;
	flush |= (__force int)(flags & (TCP_FLAG_URG | TCP_FLAG_PSH |
					TCP_FLAG_RST | TCP_FLAG_SYN |
					TCP_FLAG_FIN));

	if (p && (!NAPI_GRO_CB(skb)->same_flow || flush))
		pp = head;

out:
	NAPI_GRO_CB(skb)->flush |= flush;

	return pp;
}
EXPORT_SYMBOL(tcp_gro_receive);

int tcp_gro_complete(struct sk_buff *skb)
{
	struct tcphdr *th = tcp_hdr(skb);

	skb->csum_start = skb_transport_header(skb) - skb->head;
	skb->csum_offset = offsetof(struct tcphdr, check);
	skb->ip_summed = CHECKSUM_PARTIAL;

	skb_shinfo(skb)->gso_segs = NAPI_GRO_CB(skb)->count;

	if (th->cwr)
		skb_shinfo(skb)->gso_type |= SKB_GSO_TCP_ECN;

	return 0;
}
EXPORT_SYMBOL(tcp_gro_complete);

static int tcp_v4_gso_send_check(struct sk_buff *skb)
{
	const struct iphdr *iph;
	struct tcphdr *th;

	if (!pskb_may_pull(skb, sizeof(*th)))
		return -EINVAL;

	iph = ip_hdr(skb);
	th = tcp_hdr(skb);

	th->check = 0;
	skb->ip_summed = CHECKSUM_PARTIAL;
	__tcp_v4_send_check(skb, iph->saddr, iph->daddr);
	return 0;
}

static struct sk_buff **tcp4_gro_receive(struct sk_buff **head, struct sk_buff *skb)
{
	const struct iphdr *iph = skb_gro_network_header(skb);
	__wsum wsum;
	__sum16 sum;

	switch (skb->ip_summed) {
	case CHECKSUM_COMPLETE:
		if (!tcp_v4_check(skb_gro_len(skb), iph->saddr, iph->daddr,
				  skb->csum)) {
			skb->ip_summed = CHECKSUM_UNNECESSARY;
			break;
		}
flush:
		NAPI_GRO_CB(skb)->flush = 1;
		return NULL;

	case CHECKSUM_NONE:
		wsum = csum_tcpudp_nofold(iph->saddr, iph->daddr,
					  skb_gro_len(skb), IPPROTO_TCP, 0);
		sum = csum_fold(skb_checksum(skb,
					     skb_gro_offset(skb),
					     skb_gro_len(skb),
					     wsum));
		if (sum)
			goto flush;

		skb->ip_summed = CHECKSUM_UNNECESSARY;
		break;
	}

	return tcp_gro_receive(head, skb);
}

static int tcp4_gro_complete(struct sk_buff *skb)
{
	const struct iphdr *iph = ip_hdr(skb);
	struct tcphdr *th = tcp_hdr(skb);

	th->check = ~tcp_v4_check(skb->len - skb_transport_offset(skb),
				  iph->saddr, iph->daddr, 0);
	skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;

	return tcp_gro_complete(skb);
}

static const struct net_offload tcpv4_offload = {
	.callbacks = {
		.gso_send_check	=	tcp_v4_gso_send_check,
		.gso_segment	=	tcp_gso_segment,
		.gro_receive	=	tcp4_gro_receive,
		.gro_complete	=	tcp4_gro_complete,
	},
};

int __init tcpv4_offload_init(void)
{
	return inet_add_offload(&tcpv4_offload, IPPROTO_TCP);
}
Commit	Line	Data
28850dc7 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	* TCPv4 GSO/GRO support
	11	*/
	12
	13	#include <linux/skbuff.h>
	14	#include <net/tcp.h>
	15	#include <net/protocol.h>
	16
28be6e07	17	struct sk_buff tcp_gso_segment(struct sk_buff skb,
28850dc7 DB	18	netdev_features_t features)
	19	{
	20	struct sk_buff *segs = ERR_PTR(-EINVAL);
	21	struct tcphdr *th;
	22	unsigned int thlen;
	23	unsigned int seq;
	24	__be32 delta;
	25	unsigned int oldlen;
	26	unsigned int mss;
	27	struct sk_buff *gso_skb = skb;
	28	__sum16 newcheck;
	29	bool ooo_okay, copy_destructor;
	30
	31	if (!pskb_may_pull(skb, sizeof(*th)))
	32	goto out;
	33
	34	th = tcp_hdr(skb);
	35	thlen = th->doff * 4;
	36	if (thlen < sizeof(*th))
	37	goto out;
	38
	39	if (!pskb_may_pull(skb, thlen))
	40	goto out;
	41
	42	oldlen = (u16)~skb->len;
	43	__skb_pull(skb, thlen);
	44
	45	mss = tcp_skb_mss(skb);
	46	if (unlikely(skb->len <= mss))
	47	goto out;
	48
	49	if (skb_gso_ok(skb, features \| NETIF_F_GSO_ROBUST)) {
	50	/* Packet is from an untrusted source, reset gso_segs. */
	51	int type = skb_shinfo(skb)->gso_type;
	52
	53	if (unlikely(type &
	54	~(SKB_GSO_TCPV4 \|
	55	SKB_GSO_DODGY \|
	56	SKB_GSO_TCP_ECN \|
	57	SKB_GSO_TCPV6 \|
	58	SKB_GSO_GRE \|
cb32f511	59	SKB_GSO_IPIP \|
28850dc7 DB	60	SKB_GSO_MPLS \|
	61	SKB_GSO_UDP_TUNNEL \|
	62	0) \|\|
	63	!(type & (SKB_GSO_TCPV4 \| SKB_GSO_TCPV6))))
	64	goto out;
	65
	66	skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);
	67
	68	segs = NULL;
	69	goto out;
	70	}
	71
	72	copy_destructor = gso_skb->destructor == tcp_wfree;
	73	ooo_okay = gso_skb->ooo_okay;
	74	/* All segments but the first should have ooo_okay cleared */
	75	skb->ooo_okay = 0;
	76
	77	segs = skb_segment(skb, features);
	78	if (IS_ERR(segs))
	79	goto out;
	80
	81	/* Only first segment might have ooo_okay set */
	82	segs->ooo_okay = ooo_okay;
	83
	84	delta = htonl(oldlen + (thlen + mss));
	85
	86	skb = segs;
	87	th = tcp_hdr(skb);
	88	seq = ntohl(th->seq);
	89
	90	newcheck = ~csum_fold((__force __wsum)((__force u32)th->check +
	91	(__force u32)delta));
	92
	93	do {
	94	th->fin = th->psh = 0;
	95	th->check = newcheck;
	96
	97	if (skb->ip_summed != CHECKSUM_PARTIAL)
	98	th->check =
	99	csum_fold(csum_partial(skb_transport_header(skb),
	100	thlen, skb->csum));
	101
	102	seq += mss;
	103	if (copy_destructor) {
	104	skb->destructor = gso_skb->destructor;
	105	skb->sk = gso_skb->sk;
	106	/* {tcp\|sock}_wfree() use exact truesize accounting :
	107	* sum(skb->truesize) MUST be exactly be gso_skb->truesize
	108	* So we account mss bytes of 'true size' for each segment.
	109	* The last segment will contain the remaining.
	110	*/
	111	skb->truesize = mss;
	112	gso_skb->truesize -= mss;
	113	}
	114	skb = skb->next;
	115	th = tcp_hdr(skb);
	116
	117	th->seq = htonl(seq);
	118	th->cwr = 0;
	119	} while (skb->next);
	120
	121	/* Following permits TCP Small Queues to work well with GSO :
	122	* The callback to TCP stack will be called at the time last frag
	123	* is freed at TX completion, and not right now when gso_skb
124	* is freed by GSO engine
125	*/
126	if (copy_destructor) {
127	swap(gso_skb->sk, skb->sk);
128	swap(gso_skb->destructor, skb->destructor);
129	swap(gso_skb->truesize, skb->truesize);
130	}
131
132	delta = htonl(oldlen + (skb_tail_pointer(skb) -
133	skb_transport_header(skb)) +
134	skb->data_len);
135	th->check = ~csum_fold((__force __wsum)((__force u32)th->check +
136	(__force u32)delta));
137	if (skb->ip_summed != CHECKSUM_PARTIAL)
138	th->check = csum_fold(csum_partial(skb_transport_header(skb),
139	thlen, skb->csum));
140	out:
141	return segs;
142	}
28be6e07	143	EXPORT_SYMBOL(tcp_gso_segment);
28850dc7 DB	144
	145	struct sk_buff tcp_gro_receive(struct sk_buff head, struct sk_buff *skb)
	146	{
	147	struct sk_buff **pp = NULL;
	148	struct sk_buff *p;
	149	struct tcphdr *th;
	150	struct tcphdr *th2;
	151	unsigned int len;
	152	unsigned int thlen;
	153	__be32 flags;
	154	unsigned int mss = 1;
	155	unsigned int hlen;
	156	unsigned int off;
	157	int flush = 1;
	158	int i;
	159
	160	off = skb_gro_offset(skb);
	161	hlen = off + sizeof(*th);
	162	th = skb_gro_header_fast(skb, off);
	163	if (skb_gro_header_hard(skb, hlen)) {
	164	th = skb_gro_header_slow(skb, hlen, off);
	165	if (unlikely(!th))
	166	goto out;
	167	}
	168
	169	thlen = th->doff * 4;
	170	if (thlen < sizeof(*th))
	171	goto out;
	172
	173	hlen = off + thlen;
	174	if (skb_gro_header_hard(skb, hlen)) {
	175	th = skb_gro_header_slow(skb, hlen, off);
	176	if (unlikely(!th))
	177	goto out;
	178	}
	179
	180	skb_gro_pull(skb, thlen);
	181
	182	len = skb_gro_len(skb);
	183	flags = tcp_flag_word(th);
	184
	185	for (; (p = *head); head = &p->next) {
	186	if (!NAPI_GRO_CB(p)->same_flow)
	187	continue;
	188
	189	th2 = tcp_hdr(p);
	190
	191	if ((u32 )&th->source ^ (u32 )&th2->source) {
	192	NAPI_GRO_CB(p)->same_flow = 0;
	193	continue;
	194	}
	195
	196	goto found;
	197	}
	198
	199	goto out_check_final;
	200
	201	found:
	202	flush = NAPI_GRO_CB(p)->flush;
	203	flush \|= (__force int)(flags & TCP_FLAG_CWR);
	204	flush \|= (__force int)((flags ^ tcp_flag_word(th2)) &
	205	~(TCP_FLAG_CWR \| TCP_FLAG_FIN \| TCP_FLAG_PSH));
	206	flush \|= (__force int)(th->ack_seq ^ th2->ack_seq);
	207	for (i = sizeof(*th); i < thlen; i += 4)
208	flush \|= (u32 )((u8 *)th + i) ^
209	(u32 )((u8 *)th2 + i);
210
211	mss = tcp_skb_mss(p);
212
213	flush \|= (len - 1) >= mss;
214	flush \|= (ntohl(th2->seq) + skb_gro_len(p)) ^ ntohl(th->seq);
215
216	if (flush \|\| skb_gro_receive(head, skb)) {
217	mss = 1;
218	goto out_check_final;
219	}
220
221	p = *head;
222	th2 = tcp_hdr(p);
223	tcp_flag_word(th2) \|= flags & (TCP_FLAG_FIN \| TCP_FLAG_PSH);
224
225	out_check_final:
226	flush = len < mss;
227	flush \|= (__force int)(flags & (TCP_FLAG_URG \| TCP_FLAG_PSH \|
228	TCP_FLAG_RST \| TCP_FLAG_SYN \|
229	TCP_FLAG_FIN));
230
231	if (p && (!NAPI_GRO_CB(skb)->same_flow \|\| flush))
232	pp = head;
233
234	out:
235	NAPI_GRO_CB(skb)->flush \|= flush;
236
237	return pp;
238	}
239	EXPORT_SYMBOL(tcp_gro_receive);
240
241	int tcp_gro_complete(struct sk_buff *skb)
242	{
243	struct tcphdr *th = tcp_hdr(skb);
244
245	skb->csum_start = skb_transport_header(skb) - skb->head;
246	skb->csum_offset = offsetof(struct tcphdr, check);
247	skb->ip_summed = CHECKSUM_PARTIAL;
248
249	skb_shinfo(skb)->gso_segs = NAPI_GRO_CB(skb)->count;
250
251	if (th->cwr)
252	skb_shinfo(skb)->gso_type \|= SKB_GSO_TCP_ECN;
253
254	return 0;
255	}
256	EXPORT_SYMBOL(tcp_gro_complete);
257
258	static int tcp_v4_gso_send_check(struct sk_buff *skb)
259	{
260	const struct iphdr *iph;
261	struct tcphdr *th;
262
263	if (!pskb_may_pull(skb, sizeof(*th)))
264	return -EINVAL;
265
266	iph = ip_hdr(skb);
267	th = tcp_hdr(skb);
268
269	th->check = 0;
270	skb->ip_summed = CHECKSUM_PARTIAL;
271	__tcp_v4_send_check(skb, iph->saddr, iph->daddr);
272	return 0;
273	}
274
275	static struct sk_buff tcp4_gro_receive(struct sk_buff head, struct sk_buff *skb)
276	{
277	const struct iphdr *iph = skb_gro_network_header(skb);
278	__wsum wsum;
279	__sum16 sum;
280
281	switch (skb->ip_summed) {
282	case CHECKSUM_COMPLETE:
283	if (!tcp_v4_check(skb_gro_len(skb), iph->saddr, iph->daddr,
284	skb->csum)) {
285	skb->ip_summed = CHECKSUM_UNNECESSARY;
286	break;
287	}
288	flush:
289	NAPI_GRO_CB(skb)->flush = 1;
290	return NULL;
291
292	case CHECKSUM_NONE:
293	wsum = csum_tcpudp_nofold(iph->saddr, iph->daddr,
294	skb_gro_len(skb), IPPROTO_TCP, 0);
295	sum = csum_fold(skb_checksum(skb,
296	skb_gro_offset(skb),
297	skb_gro_len(skb),
298	wsum));
299	if (sum)
300	goto flush;
301
302	skb->ip_summed = CHECKSUM_UNNECESSARY;
303	break;
304	}
305
306	return tcp_gro_receive(head, skb);
307	}
308
309	static int tcp4_gro_complete(struct sk_buff *skb)
310	{
311	const struct iphdr *iph = ip_hdr(skb);
312	struct tcphdr *th = tcp_hdr(skb);
313
314	th->check = ~tcp_v4_check(skb->len - skb_transport_offset(skb),
315	iph->saddr, iph->daddr, 0);
316	skb_shinfo(skb)->gso_type = SKB_GSO_TCPV4;
317
318	return tcp_gro_complete(skb);
319	}
320
321	static const struct net_offload tcpv4_offload = {
322	.callbacks = {
323	.gso_send_check = tcp_v4_gso_send_check,
28be6e07	324	.gso_segment = tcp_gso_segment,
28850dc7 DB	325	.gro_receive = tcp4_gro_receive,
	326	.gro_complete = tcp4_gro_complete,
	327	},
	328	};
	329
	330	int __init tcpv4_offload_init(void)
	331	{
	332	return inet_add_offload(&tcpv4_offload, IPPROTO_TCP);
	333	}