[linux-2.6-block.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>

static void tcp_gso_tstamp(struct sk_buff *skb, unsigned int ts_seq,
			   unsigned int seq, unsigned int mss)
{
	while (skb) {
		if (before(ts_seq, seq + mss)) {
			skb_shinfo(skb)->tx_flags |= SKBTX_SW_TSTAMP;
			skb_shinfo(skb)->tskey = ts_seq;
			return;
		}

		skb = skb->next;
		seq += mss;
	}
}

static struct sk_buff *tcp4_gso_segment(struct sk_buff *skb,
					netdev_features_t features)
{
	if (!pskb_may_pull(skb, sizeof(struct tcphdr)))
		return ERR_PTR(-EINVAL);

	if (unlikely(skb->ip_summed != CHECKSUM_PARTIAL)) {
		const struct iphdr *iph = ip_hdr(skb);
		struct tcphdr *th = tcp_hdr(skb);

		/* Set up checksum pseudo header, usually expect stack to
		 * have done this already.
		 */

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

	return tcp_gso_segment(skb, features);
}

struct sk_buff *tcp_gso_segment(struct sk_buff *skb,
				netdev_features_t features)
{
	struct sk_buff *segs = ERR_PTR(-EINVAL);
	unsigned int sum_truesize = 0;
	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;

	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 = 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_TCPV4 |
			       SKB_GSO_DODGY |
			       SKB_GSO_TCP_ECN |
			       SKB_GSO_TCPV6 |
			       SKB_GSO_GRE |
			       SKB_GSO_GRE_CSUM |
			       SKB_GSO_IPIP |
			       SKB_GSO_SIT |
			       SKB_GSO_UDP_TUNNEL |
			       SKB_GSO_UDP_TUNNEL_CSUM |
			       SKB_GSO_TUNNEL_REMCSUM |
			       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);

	if (unlikely(skb_shinfo(gso_skb)->tx_flags & SKBTX_SW_TSTAMP))
		tcp_gso_tstamp(segs, skb_shinfo(gso_skb)->tskey, seq, mss);

	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)
			gso_reset_checksum(skb, ~th->check);
		else
			th->check = gso_make_checksum(skb, ~th->check);

		seq += mss;
		if (copy_destructor) {
			skb->destructor = gso_skb->destructor;
			skb->sk = gso_skb->sk;
			sum_truesize += skb->truesize;
		}
		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);
		sum_truesize += skb->truesize;
		atomic_add(sum_truesize - gso_skb->truesize,
			   &skb->sk->sk_wmem_alloc);
	}

	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)
		gso_reset_checksum(skb, ~th->check);
	else
		th->check = gso_make_checksum(skb, ~th->check);
out:
	return segs;
}

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:
	/* Include the IP ID check below from the inner most IP hdr */
	flush = NAPI_GRO_CB(p)->flush | NAPI_GRO_CB(p)->flush_id;
	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 = skb_shinfo(p)->gso_size;

	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 != 0);

	return pp;
}

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

	skb->csum_start = (unsigned char *)th - 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 struct sk_buff **tcp4_gro_receive(struct sk_buff **head, struct sk_buff *skb)
{
	/* Don't bother verifying checksum if we're going to flush anyway. */
	if (!NAPI_GRO_CB(skb)->flush &&
	    skb_gro_checksum_validate(skb, IPPROTO_TCP,
				      inet_gro_compute_pseudo)) {
		NAPI_GRO_CB(skb)->flush = 1;
		return NULL;
	}

	return tcp_gro_receive(head, skb);
}

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

	th->check = ~tcp_v4_check(skb->len - thoff, 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_segment	=	tcp4_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
f066e2b0 WB	17	static void tcp_gso_tstamp(struct sk_buff *skb, unsigned int ts_seq,
f066e2b0 WB	18	unsigned int seq, unsigned int mss)
4ed2d765 WB	19	{
4ed2d765 WB	20	while (skb) {
f066e2b0 WB	21	if (before(ts_seq, seq + mss)) {
f066e2b0 WB	22	skb_shinfo(skb)->tx_flags \|= SKBTX_SW_TSTAMP;
4ed2d765 WB	23	skb_shinfo(skb)->tskey = ts_seq;
	24	return;
	25	}
	26
	27	skb = skb->next;
	28	seq += mss;
	29	}
	30	}
	31
74abc20c ED	32	static struct sk_buff tcp4_gso_segment(struct sk_buff skb,
74abc20c ED	33	netdev_features_t features)
d020f8f7 TH	34	{
	35	if (!pskb_may_pull(skb, sizeof(struct tcphdr)))
	36	return ERR_PTR(-EINVAL);
	37
	38	if (unlikely(skb->ip_summed != CHECKSUM_PARTIAL)) {
	39	const struct iphdr *iph = ip_hdr(skb);
	40	struct tcphdr *th = tcp_hdr(skb);
	41
	42	/* Set up checksum pseudo header, usually expect stack to
	43	* have done this already.
	44	*/
	45
	46	th->check = 0;
	47	skb->ip_summed = CHECKSUM_PARTIAL;
	48	__tcp_v4_send_check(skb, iph->saddr, iph->daddr);
	49	}
	50
	51	return tcp_gso_segment(skb, features);
	52	}
	53
28be6e07	54	struct sk_buff tcp_gso_segment(struct sk_buff skb,
28850dc7 DB	55	netdev_features_t features)
	56	{
	57	struct sk_buff *segs = ERR_PTR(-EINVAL);
0d08c42c	58	unsigned int sum_truesize = 0;
28850dc7 DB	59	struct tcphdr *th;
	60	unsigned int thlen;
	61	unsigned int seq;
	62	__be32 delta;
	63	unsigned int oldlen;
	64	unsigned int mss;
	65	struct sk_buff *gso_skb = skb;
	66	__sum16 newcheck;
	67	bool ooo_okay, copy_destructor;
	68
28850dc7 DB	69	th = tcp_hdr(skb);
	70	thlen = th->doff * 4;
	71	if (thlen < sizeof(*th))
	72	goto out;
	73
	74	if (!pskb_may_pull(skb, thlen))
	75	goto out;
	76
	77	oldlen = (u16)~skb->len;
	78	__skb_pull(skb, thlen);
	79
a7eea416	80	mss = skb_shinfo(skb)->gso_size;
28850dc7 DB	81	if (unlikely(skb->len <= mss))
	82	goto out;
	83
	84	if (skb_gso_ok(skb, features \| NETIF_F_GSO_ROBUST)) {
	85	/* Packet is from an untrusted source, reset gso_segs. */
	86	int type = skb_shinfo(skb)->gso_type;
	87
	88	if (unlikely(type &
	89	~(SKB_GSO_TCPV4 \|
	90	SKB_GSO_DODGY \|
	91	SKB_GSO_TCP_ECN \|
	92	SKB_GSO_TCPV6 \|
	93	SKB_GSO_GRE \|
4749c09c	94	SKB_GSO_GRE_CSUM \|
cb32f511	95	SKB_GSO_IPIP \|
61c1db7f	96	SKB_GSO_SIT \|
28850dc7	97	SKB_GSO_UDP_TUNNEL \|
0f4f4ffa	98	SKB_GSO_UDP_TUNNEL_CSUM \|
e585f236	99	SKB_GSO_TUNNEL_REMCSUM \|
28850dc7 DB	100	0) \|\|
	101	!(type & (SKB_GSO_TCPV4 \| SKB_GSO_TCPV6))))
	102	goto out;
	103
	104	skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);
	105
	106	segs = NULL;
	107	goto out;
	108	}
	109
	110	copy_destructor = gso_skb->destructor == tcp_wfree;
	111	ooo_okay = gso_skb->ooo_okay;
	112	/* All segments but the first should have ooo_okay cleared */
	113	skb->ooo_okay = 0;
	114
	115	segs = skb_segment(skb, features);
	116	if (IS_ERR(segs))
	117	goto out;
	118
	119	/* Only first segment might have ooo_okay set */
	120	segs->ooo_okay = ooo_okay;
	121
	122	delta = htonl(oldlen + (thlen + mss));
	123
	124	skb = segs;
	125	th = tcp_hdr(skb);
	126	seq = ntohl(th->seq);
	127
4ed2d765 WB	128	if (unlikely(skb_shinfo(gso_skb)->tx_flags & SKBTX_SW_TSTAMP))
	129	tcp_gso_tstamp(segs, skb_shinfo(gso_skb)->tskey, seq, mss);
	130
28850dc7 DB	131	newcheck = ~csum_fold((__force __wsum)((__force u32)th->check +
	132	(__force u32)delta));
	133
	134	do {
	135	th->fin = th->psh = 0;
	136	th->check = newcheck;
	137
08b64fcc AD	138	if (skb->ip_summed == CHECKSUM_PARTIAL)
	139	gso_reset_checksum(skb, ~th->check);
	140	else
e9c3a24b	141	th->check = gso_make_checksum(skb, ~th->check);
28850dc7 DB	142
	143	seq += mss;
	144	if (copy_destructor) {
	145	skb->destructor = gso_skb->destructor;
	146	skb->sk = gso_skb->sk;
0d08c42c	147	sum_truesize += skb->truesize;
28850dc7 DB	148	}
	149	skb = skb->next;
	150	th = tcp_hdr(skb);
	151
	152	th->seq = htonl(seq);
	153	th->cwr = 0;
	154	} while (skb->next);
	155
	156	/* Following permits TCP Small Queues to work well with GSO :
	157	* The callback to TCP stack will be called at the time last frag
	158	* is freed at TX completion, and not right now when gso_skb
	159	* is freed by GSO engine
	160	*/
	161	if (copy_destructor) {
	162	swap(gso_skb->sk, skb->sk);
	163	swap(gso_skb->destructor, skb->destructor);
0d08c42c ED	164	sum_truesize += skb->truesize;
	165	atomic_add(sum_truesize - gso_skb->truesize,
	166	&skb->sk->sk_wmem_alloc);
28850dc7 DB	167	}
	168
	169	delta = htonl(oldlen + (skb_tail_pointer(skb) -
	170	skb_transport_header(skb)) +
	171	skb->data_len);
	172	th->check = ~csum_fold((__force __wsum)((__force u32)th->check +
	173	(__force u32)delta));
08b64fcc AD	174	if (skb->ip_summed == CHECKSUM_PARTIAL)
	175	gso_reset_checksum(skb, ~th->check);
	176	else
e9c3a24b	177	th->check = gso_make_checksum(skb, ~th->check);
28850dc7 DB	178	out:
	179	return segs;
	180	}
28850dc7 DB	181
	182	struct sk_buff tcp_gro_receive(struct sk_buff head, struct sk_buff *skb)
	183	{
	184	struct sk_buff **pp = NULL;
	185	struct sk_buff *p;
	186	struct tcphdr *th;
	187	struct tcphdr *th2;
	188	unsigned int len;
	189	unsigned int thlen;
	190	__be32 flags;
	191	unsigned int mss = 1;
	192	unsigned int hlen;
	193	unsigned int off;
	194	int flush = 1;
	195	int i;
	196
	197	off = skb_gro_offset(skb);
	198	hlen = off + sizeof(*th);
	199	th = skb_gro_header_fast(skb, off);
	200	if (skb_gro_header_hard(skb, hlen)) {
	201	th = skb_gro_header_slow(skb, hlen, off);
	202	if (unlikely(!th))
	203	goto out;
	204	}
	205
	206	thlen = th->doff * 4;
	207	if (thlen < sizeof(*th))
	208	goto out;
	209
	210	hlen = off + thlen;
	211	if (skb_gro_header_hard(skb, hlen)) {
	212	th = skb_gro_header_slow(skb, hlen, off);
	213	if (unlikely(!th))
	214	goto out;
	215	}
	216
	217	skb_gro_pull(skb, thlen);
	218
	219	len = skb_gro_len(skb);
	220	flags = tcp_flag_word(th);
	221
	222	for (; (p = *head); head = &p->next) {
	223	if (!NAPI_GRO_CB(p)->same_flow)
	224	continue;
	225
	226	th2 = tcp_hdr(p);
	227
	228	if ((u32 )&th->source ^ (u32 )&th2->source) {
	229	NAPI_GRO_CB(p)->same_flow = 0;
	230	continue;
	231	}
	232
	233	goto found;
	234	}
	235
	236	goto out_check_final;
	237
	238	found:
bf5a755f JC	239	/* Include the IP ID check below from the inner most IP hdr */
bf5a755f JC	240	flush = NAPI_GRO_CB(p)->flush \| NAPI_GRO_CB(p)->flush_id;
28850dc7 DB	241	flush \|= (__force int)(flags & TCP_FLAG_CWR);
	242	flush \|= (__force int)((flags ^ tcp_flag_word(th2)) &
	243	~(TCP_FLAG_CWR \| TCP_FLAG_FIN \| TCP_FLAG_PSH));
	244	flush \|= (__force int)(th->ack_seq ^ th2->ack_seq);
	245	for (i = sizeof(*th); i < thlen; i += 4)
	246	flush \|= (u32 )((u8 *)th + i) ^
	247	(u32 )((u8 *)th2 + i);
	248
a7eea416	249	mss = skb_shinfo(p)->gso_size;
28850dc7 DB	250
	251	flush \|= (len - 1) >= mss;
	252	flush \|= (ntohl(th2->seq) + skb_gro_len(p)) ^ ntohl(th->seq);
	253
	254	if (flush \|\| skb_gro_receive(head, skb)) {
	255	mss = 1;
	256	goto out_check_final;
	257	}
	258
	259	p = *head;
	260	th2 = tcp_hdr(p);
	261	tcp_flag_word(th2) \|= flags & (TCP_FLAG_FIN \| TCP_FLAG_PSH);
	262
	263	out_check_final:
	264	flush = len < mss;
	265	flush \|= (__force int)(flags & (TCP_FLAG_URG \| TCP_FLAG_PSH \|
	266	TCP_FLAG_RST \| TCP_FLAG_SYN \|
	267	TCP_FLAG_FIN));
	268
	269	if (p && (!NAPI_GRO_CB(skb)->same_flow \|\| flush))
	270	pp = head;
	271
	272	out:
bf5a755f	273	NAPI_GRO_CB(skb)->flush \|= (flush != 0);
28850dc7 DB	274
	275	return pp;
	276	}
28850dc7 DB	277
	278	int tcp_gro_complete(struct sk_buff *skb)
	279	{
	280	struct tcphdr *th = tcp_hdr(skb);
	281
299603e8	282	skb->csum_start = (unsigned char *)th - skb->head;
28850dc7 DB	283	skb->csum_offset = offsetof(struct tcphdr, check);
	284	skb->ip_summed = CHECKSUM_PARTIAL;
	285
	286	skb_shinfo(skb)->gso_segs = NAPI_GRO_CB(skb)->count;
	287
	288	if (th->cwr)
	289	skb_shinfo(skb)->gso_type \|= SKB_GSO_TCP_ECN;
	290
	291	return 0;
	292	}
	293	EXPORT_SYMBOL(tcp_gro_complete);
	294
28850dc7 DB	295	static struct sk_buff tcp4_gro_receive(struct sk_buff head, struct sk_buff *skb)
28850dc7 DB	296	{
cc5c00bb	297	/* Don't bother verifying checksum if we're going to flush anyway. */
149d0774 TH	298	if (!NAPI_GRO_CB(skb)->flush &&
	299	skb_gro_checksum_validate(skb, IPPROTO_TCP,
	300	inet_gro_compute_pseudo)) {
28850dc7 DB	301	NAPI_GRO_CB(skb)->flush = 1;
28850dc7 DB	302	return NULL;
28850dc7 DB	303	}
	304
	305	return tcp_gro_receive(head, skb);
	306	}
	307
299603e8	308	static int tcp4_gro_complete(struct sk_buff *skb, int thoff)
28850dc7 DB	309	{
	310	const struct iphdr *iph = ip_hdr(skb);
	311	struct tcphdr *th = tcp_hdr(skb);
	312
299603e8 JC	313	th->check = ~tcp_v4_check(skb->len - thoff, iph->saddr,
299603e8 JC	314	iph->daddr, 0);
c3caf119	315	skb_shinfo(skb)->gso_type \|= SKB_GSO_TCPV4;
28850dc7 DB	316
	317	return tcp_gro_complete(skb);
	318	}
	319
	320	static const struct net_offload tcpv4_offload = {
	321	.callbacks = {
d020f8f7	322	.gso_segment = tcp4_gso_segment,
28850dc7 DB	323	.gro_receive = tcp4_gro_receive,
	324	.gro_complete = tcp4_gro_complete,
	325	},
	326	};
	327
	328	int __init tcpv4_offload_init(void)
	329	{
	330	return inet_add_offload(&tcpv4_offload, IPPROTO_TCP);
	331	}