[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>

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_TCP_FIXEDID |
			       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 \|
cbc53e08	92	SKB_GSO_TCP_FIXEDID \|
28850dc7 DB	93	SKB_GSO_TCPV6 \|
28850dc7 DB	94	SKB_GSO_GRE \|
4749c09c	95	SKB_GSO_GRE_CSUM \|
cb32f511	96	SKB_GSO_IPIP \|
61c1db7f	97	SKB_GSO_SIT \|
28850dc7	98	SKB_GSO_UDP_TUNNEL \|
0f4f4ffa	99	SKB_GSO_UDP_TUNNEL_CSUM \|
e585f236	100	SKB_GSO_TUNNEL_REMCSUM \|
28850dc7	101	0) \|\|
cbc53e08 AD	102	!(type & (SKB_GSO_TCPV4 \|
cbc53e08 AD	103	SKB_GSO_TCPV6))))
28850dc7 DB	104	goto out;
	105
	106	skb_shinfo(skb)->gso_segs = DIV_ROUND_UP(skb->len, mss);
	107
	108	segs = NULL;
	109	goto out;
	110	}
	111
	112	copy_destructor = gso_skb->destructor == tcp_wfree;
	113	ooo_okay = gso_skb->ooo_okay;
	114	/* All segments but the first should have ooo_okay cleared */
	115	skb->ooo_okay = 0;
	116
	117	segs = skb_segment(skb, features);
	118	if (IS_ERR(segs))
	119	goto out;
	120
	121	/* Only first segment might have ooo_okay set */
	122	segs->ooo_okay = ooo_okay;
	123
	124	delta = htonl(oldlen + (thlen + mss));
	125
	126	skb = segs;
	127	th = tcp_hdr(skb);
	128	seq = ntohl(th->seq);
	129
4ed2d765 WB	130	if (unlikely(skb_shinfo(gso_skb)->tx_flags & SKBTX_SW_TSTAMP))
	131	tcp_gso_tstamp(segs, skb_shinfo(gso_skb)->tskey, seq, mss);
	132
28850dc7 DB	133	newcheck = ~csum_fold((__force __wsum)((__force u32)th->check +
	134	(__force u32)delta));
	135
	136	do {
	137	th->fin = th->psh = 0;
	138	th->check = newcheck;
	139
08b64fcc AD	140	if (skb->ip_summed == CHECKSUM_PARTIAL)
	141	gso_reset_checksum(skb, ~th->check);
	142	else
e9c3a24b	143	th->check = gso_make_checksum(skb, ~th->check);
28850dc7 DB	144
	145	seq += mss;
	146	if (copy_destructor) {
	147	skb->destructor = gso_skb->destructor;
	148	skb->sk = gso_skb->sk;
0d08c42c	149	sum_truesize += skb->truesize;
28850dc7 DB	150	}
	151	skb = skb->next;
	152	th = tcp_hdr(skb);
	153
	154	th->seq = htonl(seq);
	155	th->cwr = 0;
	156	} while (skb->next);
	157
	158	/* Following permits TCP Small Queues to work well with GSO :
	159	* The callback to TCP stack will be called at the time last frag
	160	* is freed at TX completion, and not right now when gso_skb
	161	* is freed by GSO engine
	162	*/
	163	if (copy_destructor) {
	164	swap(gso_skb->sk, skb->sk);
	165	swap(gso_skb->destructor, skb->destructor);
0d08c42c ED	166	sum_truesize += skb->truesize;
	167	atomic_add(sum_truesize - gso_skb->truesize,
	168	&skb->sk->sk_wmem_alloc);
28850dc7 DB	169	}
	170
	171	delta = htonl(oldlen + (skb_tail_pointer(skb) -
	172	skb_transport_header(skb)) +
	173	skb->data_len);
	174	th->check = ~csum_fold((__force __wsum)((__force u32)th->check +
	175	(__force u32)delta));
08b64fcc AD	176	if (skb->ip_summed == CHECKSUM_PARTIAL)
	177	gso_reset_checksum(skb, ~th->check);
	178	else
e9c3a24b	179	th->check = gso_make_checksum(skb, ~th->check);
28850dc7 DB	180	out:
	181	return segs;
	182	}
28850dc7 DB	183
	184	struct sk_buff tcp_gro_receive(struct sk_buff head, struct sk_buff *skb)
	185	{
	186	struct sk_buff **pp = NULL;
	187	struct sk_buff *p;
	188	struct tcphdr *th;
	189	struct tcphdr *th2;
	190	unsigned int len;
	191	unsigned int thlen;
	192	__be32 flags;
	193	unsigned int mss = 1;
	194	unsigned int hlen;
	195	unsigned int off;
	196	int flush = 1;
	197	int i;
	198
	199	off = skb_gro_offset(skb);
	200	hlen = off + sizeof(*th);
	201	th = skb_gro_header_fast(skb, off);
	202	if (skb_gro_header_hard(skb, hlen)) {
	203	th = skb_gro_header_slow(skb, hlen, off);
	204	if (unlikely(!th))
	205	goto out;
	206	}
	207
	208	thlen = th->doff * 4;
	209	if (thlen < sizeof(*th))
	210	goto out;
	211
	212	hlen = off + thlen;
	213	if (skb_gro_header_hard(skb, hlen)) {
	214	th = skb_gro_header_slow(skb, hlen, off);
	215	if (unlikely(!th))
	216	goto out;
	217	}
	218
	219	skb_gro_pull(skb, thlen);
	220
	221	len = skb_gro_len(skb);
	222	flags = tcp_flag_word(th);
	223
	224	for (; (p = *head); head = &p->next) {
	225	if (!NAPI_GRO_CB(p)->same_flow)
	226	continue;
	227
	228	th2 = tcp_hdr(p);
	229
	230	if ((u32 )&th->source ^ (u32 )&th2->source) {
	231	NAPI_GRO_CB(p)->same_flow = 0;
	232	continue;
	233	}
	234
	235	goto found;
	236	}
	237
	238	goto out_check_final;
	239
	240	found:
bf5a755f JC	241	/* Include the IP ID check below from the inner most IP hdr */
bf5a755f JC	242	flush = NAPI_GRO_CB(p)->flush \| NAPI_GRO_CB(p)->flush_id;
28850dc7 DB	243	flush \|= (__force int)(flags & TCP_FLAG_CWR);
	244	flush \|= (__force int)((flags ^ tcp_flag_word(th2)) &
	245	~(TCP_FLAG_CWR \| TCP_FLAG_FIN \| TCP_FLAG_PSH));
	246	flush \|= (__force int)(th->ack_seq ^ th2->ack_seq);
	247	for (i = sizeof(*th); i < thlen; i += 4)
	248	flush \|= (u32 )((u8 *)th + i) ^
	249	(u32 )((u8 *)th2 + i);
	250
a7eea416	251	mss = skb_shinfo(p)->gso_size;
28850dc7 DB	252
	253	flush \|= (len - 1) >= mss;
	254	flush \|= (ntohl(th2->seq) + skb_gro_len(p)) ^ ntohl(th->seq);
	255
	256	if (flush \|\| skb_gro_receive(head, skb)) {
	257	mss = 1;
	258	goto out_check_final;
	259	}
	260
	261	p = *head;
	262	th2 = tcp_hdr(p);
	263	tcp_flag_word(th2) \|= flags & (TCP_FLAG_FIN \| TCP_FLAG_PSH);
	264
	265	out_check_final:
	266	flush = len < mss;
	267	flush \|= (__force int)(flags & (TCP_FLAG_URG \| TCP_FLAG_PSH \|
	268	TCP_FLAG_RST \| TCP_FLAG_SYN \|
	269	TCP_FLAG_FIN));
	270
	271	if (p && (!NAPI_GRO_CB(skb)->same_flow \|\| flush))
	272	pp = head;
	273
	274	out:
bf5a755f	275	NAPI_GRO_CB(skb)->flush \|= (flush != 0);
28850dc7 DB	276
	277	return pp;
	278	}
28850dc7 DB	279
	280	int tcp_gro_complete(struct sk_buff *skb)
	281	{
	282	struct tcphdr *th = tcp_hdr(skb);
	283
299603e8	284	skb->csum_start = (unsigned char *)th - skb->head;
28850dc7 DB	285	skb->csum_offset = offsetof(struct tcphdr, check);
	286	skb->ip_summed = CHECKSUM_PARTIAL;
	287
	288	skb_shinfo(skb)->gso_segs = NAPI_GRO_CB(skb)->count;
	289
	290	if (th->cwr)
	291	skb_shinfo(skb)->gso_type \|= SKB_GSO_TCP_ECN;
	292
	293	return 0;
	294	}
	295	EXPORT_SYMBOL(tcp_gro_complete);
	296
28850dc7 DB	297	static struct sk_buff tcp4_gro_receive(struct sk_buff head, struct sk_buff *skb)
28850dc7 DB	298	{
cc5c00bb	299	/* Don't bother verifying checksum if we're going to flush anyway. */
149d0774 TH	300	if (!NAPI_GRO_CB(skb)->flush &&
	301	skb_gro_checksum_validate(skb, IPPROTO_TCP,
	302	inet_gro_compute_pseudo)) {
28850dc7 DB	303	NAPI_GRO_CB(skb)->flush = 1;
28850dc7 DB	304	return NULL;
28850dc7 DB	305	}
	306
	307	return tcp_gro_receive(head, skb);
	308	}
	309
299603e8	310	static int tcp4_gro_complete(struct sk_buff *skb, int thoff)
28850dc7 DB	311	{
	312	const struct iphdr *iph = ip_hdr(skb);
	313	struct tcphdr *th = tcp_hdr(skb);
	314
299603e8 JC	315	th->check = ~tcp_v4_check(skb->len - thoff, iph->saddr,
299603e8 JC	316	iph->daddr, 0);
c3caf119	317	skb_shinfo(skb)->gso_type \|= SKB_GSO_TCPV4;
28850dc7 DB	318
	319	return tcp_gro_complete(skb);
	320	}
	321
	322	static const struct net_offload tcpv4_offload = {
	323	.callbacks = {
d020f8f7	324	.gso_segment = tcp4_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	}