[linux.git] / net / core / flow_dissector.c

#include <linux/skbuff.h>
#include <linux/export.h>
#include <linux/ip.h>
#include <linux/ipv6.h>
#include <linux/if_vlan.h>
#include <net/ip.h>
#include <net/ipv6.h>
#include <linux/igmp.h>
#include <linux/icmp.h>
#include <linux/sctp.h>
#include <linux/dccp.h>
#include <linux/if_tunnel.h>
#include <linux/if_pppox.h>
#include <linux/ppp_defs.h>
#include <net/flow_keys.h>
#include <scsi/fc/fc_fcoe.h>

/* copy saddr & daddr, possibly using 64bit load/store
 * Equivalent to :	flow->src = iph->saddr;
 *			flow->dst = iph->daddr;
 */
static void iph_to_flow_copy_addrs(struct flow_keys *flow, const struct iphdr *iph)
{
	BUILD_BUG_ON(offsetof(typeof(*flow), dst) !=
		     offsetof(typeof(*flow), src) + sizeof(flow->src));
	memcpy(&flow->src, &iph->saddr, sizeof(flow->src) + sizeof(flow->dst));
}

/**
 * __skb_flow_get_ports - extract the upper layer ports and return them
 * @skb: sk_buff to extract the ports from
 * @thoff: transport header offset
 * @ip_proto: protocol for which to get port offset
 * @data: raw buffer pointer to the packet, if NULL use skb->data
 * @hlen: packet header length, if @data is NULL use skb_headlen(skb)
 *
 * The function will try to retrieve the ports at offset thoff + poff where poff
 * is the protocol port offset returned from proto_ports_offset
 */
__be32 __skb_flow_get_ports(const struct sk_buff *skb, int thoff, u8 ip_proto,
			    void *data, int hlen)
{
	int poff = proto_ports_offset(ip_proto);

	if (!data) {
		data = skb->data;
		hlen = skb_headlen(skb);
	}

	if (poff >= 0) {
		__be32 *ports, _ports;

		ports = __skb_header_pointer(skb, thoff + poff,
					     sizeof(_ports), data, hlen, &_ports);
		if (ports)
			return *ports;
	}

	return 0;
}
EXPORT_SYMBOL(__skb_flow_get_ports);

/**
 * __skb_flow_dissect - extract the flow_keys struct and return it
 * @skb: sk_buff to extract the flow from, can be NULL if the rest are specified
 * @data: raw buffer pointer to the packet, if NULL use skb->data
 * @proto: protocol for which to get the flow, if @data is NULL use skb->protocol
 * @nhoff: network header offset, if @data is NULL use skb_network_offset(skb)
 * @hlen: packet header length, if @data is NULL use skb_headlen(skb)
 *
 * The function will try to retrieve the struct flow_keys from either the skbuff
 * or a raw buffer specified by the rest parameters
 */
bool __skb_flow_dissect(const struct sk_buff *skb, struct flow_keys *flow,
			void *data, __be16 proto, int nhoff, int hlen)
{
	u8 ip_proto;

	if (!data) {
		data = skb->data;
		proto = skb->protocol;
		nhoff = skb_network_offset(skb);
		hlen = skb_headlen(skb);
	}

	memset(flow, 0, sizeof(*flow));

again:
	switch (proto) {
	case htons(ETH_P_IP): {
		const struct iphdr *iph;
		struct iphdr _iph;
ip:
		iph = __skb_header_pointer(skb, nhoff, sizeof(_iph), data, hlen, &_iph);
		if (!iph || iph->ihl < 5)
			return false;
		nhoff += iph->ihl * 4;

		ip_proto = iph->protocol;
		if (ip_is_fragment(iph))
			ip_proto = 0;

		/* skip the address processing if skb is NULL.  The assumption
		 * here is that if there is no skb we are not looking for flow
		 * info but lengths and protocols.
		 */
		if (!skb)
			break;

		iph_to_flow_copy_addrs(flow, iph);
		break;
	}
	case htons(ETH_P_IPV6): {
		const struct ipv6hdr *iph;
		struct ipv6hdr _iph;
		__be32 flow_label;

ipv6:
		iph = __skb_header_pointer(skb, nhoff, sizeof(_iph), data, hlen, &_iph);
		if (!iph)
			return false;

		ip_proto = iph->nexthdr;
		nhoff += sizeof(struct ipv6hdr);

		/* see comment above in IPv4 section */
		if (!skb)
			break;

		flow->src = (__force __be32)ipv6_addr_hash(&iph->saddr);
		flow->dst = (__force __be32)ipv6_addr_hash(&iph->daddr);

		flow_label = ip6_flowlabel(iph);
		if (flow_label) {
			/* Awesome, IPv6 packet has a flow label so we can
			 * use that to represent the ports without any
			 * further dissection.
			 */
			flow->n_proto = proto;
			flow->ip_proto = ip_proto;
			flow->ports = flow_label;
			flow->thoff = (u16)nhoff;

			return true;
		}

		break;
	}
	case htons(ETH_P_8021AD):
	case htons(ETH_P_8021Q): {
		const struct vlan_hdr *vlan;
		struct vlan_hdr _vlan;

		vlan = __skb_header_pointer(skb, nhoff, sizeof(_vlan), data, hlen, &_vlan);
		if (!vlan)
			return false;

		proto = vlan->h_vlan_encapsulated_proto;
		nhoff += sizeof(*vlan);
		goto again;
	}
	case htons(ETH_P_PPP_SES): {
		struct {
			struct pppoe_hdr hdr;
			__be16 proto;
		} *hdr, _hdr;
		hdr = __skb_header_pointer(skb, nhoff, sizeof(_hdr), data, hlen, &_hdr);
		if (!hdr)
			return false;
		proto = hdr->proto;
		nhoff += PPPOE_SES_HLEN;
		switch (proto) {
		case htons(PPP_IP):
			goto ip;
		case htons(PPP_IPV6):
			goto ipv6;
		default:
			return false;
		}
	}
	case htons(ETH_P_TIPC): {
		struct {
			__be32 pre[3];
			__be32 srcnode;
		} *hdr, _hdr;
		hdr = __skb_header_pointer(skb, nhoff, sizeof(_hdr), data, hlen, &_hdr);
		if (!hdr)
			return false;
		flow->src = hdr->srcnode;
		flow->dst = 0;
		flow->n_proto = proto;
		flow->thoff = (u16)nhoff;
		return true;
	}
	case htons(ETH_P_FCOE):
		flow->thoff = (u16)(nhoff + FCOE_HEADER_LEN);
		/* fall through */
	default:
		return false;
	}

	switch (ip_proto) {
	case IPPROTO_GRE: {
		struct gre_hdr {
			__be16 flags;
			__be16 proto;
		} *hdr, _hdr;

		hdr = __skb_header_pointer(skb, nhoff, sizeof(_hdr), data, hlen, &_hdr);
		if (!hdr)
			return false;
		/*
		 * Only look inside GRE if version zero and no
		 * routing
		 */
		if (!(hdr->flags & (GRE_VERSION|GRE_ROUTING))) {
			proto = hdr->proto;
			nhoff += 4;
			if (hdr->flags & GRE_CSUM)
				nhoff += 4;
			if (hdr->flags & GRE_KEY)
				nhoff += 4;
			if (hdr->flags & GRE_SEQ)
				nhoff += 4;
			if (proto == htons(ETH_P_TEB)) {
				const struct ethhdr *eth;
				struct ethhdr _eth;

				eth = __skb_header_pointer(skb, nhoff,
							   sizeof(_eth),
							   data, hlen, &_eth);
				if (!eth)
					return false;
				proto = eth->h_proto;
				nhoff += sizeof(*eth);
			}
			goto again;
		}
		break;
	}
	case IPPROTO_IPIP:
		proto = htons(ETH_P_IP);
		goto ip;
	case IPPROTO_IPV6:
		proto = htons(ETH_P_IPV6);
		goto ipv6;
	default:
		break;
	}

	flow->n_proto = proto;
	flow->ip_proto = ip_proto;
	flow->thoff = (u16) nhoff;

	/* unless skb is set we don't need to record port info */
	if (skb)
		flow->ports = __skb_flow_get_ports(skb, nhoff, ip_proto,
						   data, hlen);

	return true;
}
EXPORT_SYMBOL(__skb_flow_dissect);

static u32 hashrnd __read_mostly;
static __always_inline void __flow_hash_secret_init(void)
{
	net_get_random_once(&hashrnd, sizeof(hashrnd));
}

static __always_inline u32 __flow_hash_3words(u32 a, u32 b, u32 c)
{
	__flow_hash_secret_init();
	return jhash_3words(a, b, c, hashrnd);
}

static inline u32 __flow_hash_from_keys(struct flow_keys *keys)
{
	u32 hash;

	/* get a consistent hash (same value on both flow directions) */
	if (((__force u32)keys->dst < (__force u32)keys->src) ||
	    (((__force u32)keys->dst == (__force u32)keys->src) &&
	     ((__force u16)keys->port16[1] < (__force u16)keys->port16[0]))) {
		swap(keys->dst, keys->src);
		swap(keys->port16[0], keys->port16[1]);
	}

	hash = __flow_hash_3words((__force u32)keys->dst,
				  (__force u32)keys->src,
				  (__force u32)keys->ports);
	if (!hash)
		hash = 1;

	return hash;
}

u32 flow_hash_from_keys(struct flow_keys *keys)
{
	return __flow_hash_from_keys(keys);
}
EXPORT_SYMBOL(flow_hash_from_keys);

/*
 * __skb_get_hash: calculate a flow hash based on src/dst addresses
 * and src/dst port numbers.  Sets hash in skb to non-zero hash value
 * on success, zero indicates no valid hash.  Also, sets l4_hash in skb
 * if hash is a canonical 4-tuple hash over transport ports.
 */
void __skb_get_hash(struct sk_buff *skb)
{
	struct flow_keys keys;

	if (!skb_flow_dissect(skb, &keys))
		return;

	if (keys.ports)
		skb->l4_hash = 1;

	skb->sw_hash = 1;

	skb->hash = __flow_hash_from_keys(&keys);
}
EXPORT_SYMBOL(__skb_get_hash);

/*
 * Returns a Tx hash based on the given packet descriptor a Tx queues' number
 * to be used as a distribution range.
 */
u16 __skb_tx_hash(const struct net_device *dev, struct sk_buff *skb,
		  unsigned int num_tx_queues)
{
	u32 hash;
	u16 qoffset = 0;
	u16 qcount = num_tx_queues;

	if (skb_rx_queue_recorded(skb)) {
		hash = skb_get_rx_queue(skb);
		while (unlikely(hash >= num_tx_queues))
			hash -= num_tx_queues;
		return hash;
	}

	if (dev->num_tc) {
		u8 tc = netdev_get_prio_tc_map(dev, skb->priority);
		qoffset = dev->tc_to_txq[tc].offset;
		qcount = dev->tc_to_txq[tc].count;
	}

	return (u16) reciprocal_scale(skb_get_hash(skb), qcount) + qoffset;
}
EXPORT_SYMBOL(__skb_tx_hash);

u32 __skb_get_poff(const struct sk_buff *skb, void *data,
		   const struct flow_keys *keys, int hlen)
{
	u32 poff = keys->thoff;

	switch (keys->ip_proto) {
	case IPPROTO_TCP: {
		/* access doff as u8 to avoid unaligned access */
		const u8 *doff;
		u8 _doff;

		doff = __skb_header_pointer(skb, poff + 12, sizeof(_doff),
					    data, hlen, &_doff);
		if (!doff)
			return poff;

		poff += max_t(u32, sizeof(struct tcphdr), (*doff & 0xF0) >> 2);
		break;
	}
	case IPPROTO_UDP:
	case IPPROTO_UDPLITE:
		poff += sizeof(struct udphdr);
		break;
	/* For the rest, we do not really care about header
	 * extensions at this point for now.
	 */
	case IPPROTO_ICMP:
		poff += sizeof(struct icmphdr);
		break;
	case IPPROTO_ICMPV6:
		poff += sizeof(struct icmp6hdr);
		break;
	case IPPROTO_IGMP:
		poff += sizeof(struct igmphdr);
		break;
	case IPPROTO_DCCP:
		poff += sizeof(struct dccp_hdr);
		break;
	case IPPROTO_SCTP:
		poff += sizeof(struct sctphdr);
		break;
	}

	return poff;
}

/* skb_get_poff() returns the offset to the payload as far as it could
 * be dissected. The main user is currently BPF, so that we can dynamically
 * truncate packets without needing to push actual payload to the user
 * space and can analyze headers only, instead.
 */
u32 skb_get_poff(const struct sk_buff *skb)
{
	struct flow_keys keys;

	if (!skb_flow_dissect(skb, &keys))
		return 0;

	return __skb_get_poff(skb, skb->data, &keys, skb_headlen(skb));
}

static inline int get_xps_queue(struct net_device *dev, struct sk_buff *skb)
{
#ifdef CONFIG_XPS
	struct xps_dev_maps *dev_maps;
	struct xps_map *map;
	int queue_index = -1;

	rcu_read_lock();
	dev_maps = rcu_dereference(dev->xps_maps);
	if (dev_maps) {
		map = rcu_dereference(
		    dev_maps->cpu_map[skb->sender_cpu - 1]);
		if (map) {
			if (map->len == 1)
				queue_index = map->queues[0];
			else
				queue_index = map->queues[reciprocal_scale(skb_get_hash(skb),
									   map->len)];
			if (unlikely(queue_index >= dev->real_num_tx_queues))
				queue_index = -1;
		}
	}
	rcu_read_unlock();

	return queue_index;
#else
	return -1;
#endif
}

static u16 __netdev_pick_tx(struct net_device *dev, struct sk_buff *skb)
{
	struct sock *sk = skb->sk;
	int queue_index = sk_tx_queue_get(sk);

	if (queue_index < 0 || skb->ooo_okay ||
	    queue_index >= dev->real_num_tx_queues) {
		int new_index = get_xps_queue(dev, skb);
		if (new_index < 0)
			new_index = skb_tx_hash(dev, skb);

		if (queue_index != new_index && sk &&
		    rcu_access_pointer(sk->sk_dst_cache))
			sk_tx_queue_set(sk, new_index);

		queue_index = new_index;
	}

	return queue_index;
}

struct netdev_queue *netdev_pick_tx(struct net_device *dev,
				    struct sk_buff *skb,
				    void *accel_priv)
{
	int queue_index = 0;

#ifdef CONFIG_XPS
	if (skb->sender_cpu == 0)
		skb->sender_cpu = raw_smp_processor_id() + 1;
#endif

	if (dev->real_num_tx_queues != 1) {
		const struct net_device_ops *ops = dev->netdev_ops;
		if (ops->ndo_select_queue)
			queue_index = ops->ndo_select_queue(dev, skb, accel_priv,
							    __netdev_pick_tx);
		else
			queue_index = __netdev_pick_tx(dev, skb);

		if (!accel_priv)
			queue_index = netdev_cap_txqueue(dev, queue_index);
	}

	skb_set_queue_mapping(skb, queue_index);
	return netdev_get_tx_queue(dev, queue_index);
}
Commit	Line	Data
0744dd00	1	#include <linux/skbuff.h>
c452ed70	2	#include <linux/export.h>
0744dd00 ED	3	#include <linux/ip.h>
	4	#include <linux/ipv6.h>
	5	#include <linux/if_vlan.h>
	6	#include <net/ip.h>
ddbe5032	7	#include <net/ipv6.h>
f77668dc DB	8	#include <linux/igmp.h>
	9	#include <linux/icmp.h>
	10	#include <linux/sctp.h>
	11	#include <linux/dccp.h>
0744dd00 ED	12	#include <linux/if_tunnel.h>
	13	#include <linux/if_pppox.h>
	14	#include <linux/ppp_defs.h>
	15	#include <net/flow_keys.h>
56193d1b	16	#include <scsi/fc/fc_fcoe.h>
0744dd00	17
4d77d2b5 ED	18	/* copy saddr & daddr, possibly using 64bit load/store
	19	* Equivalent to : flow->src = iph->saddr;
	20	* flow->dst = iph->daddr;
	21	*/
	22	static void iph_to_flow_copy_addrs(struct flow_keys flow, const struct iphdr iph)
	23	{
	24	BUILD_BUG_ON(offsetof(typeof(*flow), dst) !=
	25	offsetof(typeof(*flow), src) + sizeof(flow->src));
	26	memcpy(&flow->src, &iph->saddr, sizeof(flow->src) + sizeof(flow->dst));
	27	}
0744dd00	28
357afe9c	29	/**
6451b3f5 WC	30	* __skb_flow_get_ports - extract the upper layer ports and return them
6451b3f5 WC	31	* @skb: sk_buff to extract the ports from
357afe9c NA	32	* @thoff: transport header offset
357afe9c NA	33	* @ip_proto: protocol for which to get port offset
6451b3f5 WC	34	* @data: raw buffer pointer to the packet, if NULL use skb->data
6451b3f5 WC	35	* @hlen: packet header length, if @data is NULL use skb_headlen(skb)
357afe9c NA	36	*
	37	* The function will try to retrieve the ports at offset thoff + poff where poff
	38	* is the protocol port offset returned from proto_ports_offset
	39	*/
690e36e7 DM	40	__be32 __skb_flow_get_ports(const struct sk_buff *skb, int thoff, u8 ip_proto,
690e36e7 DM	41	void *data, int hlen)
357afe9c NA	42	{
	43	int poff = proto_ports_offset(ip_proto);
	44
690e36e7 DM	45	if (!data) {
	46	data = skb->data;
	47	hlen = skb_headlen(skb);
	48	}
	49
357afe9c NA	50	if (poff >= 0) {
	51	__be32 *ports, _ports;
	52
690e36e7 DM	53	ports = __skb_header_pointer(skb, thoff + poff,
690e36e7 DM	54	sizeof(_ports), data, hlen, &_ports);
357afe9c NA	55	if (ports)
	56	return *ports;
	57	}
	58
	59	return 0;
	60	}
690e36e7	61	EXPORT_SYMBOL(__skb_flow_get_ports);
357afe9c	62
453a940e WC	63	/**
	64	* __skb_flow_dissect - extract the flow_keys struct and return it
	65	* @skb: sk_buff to extract the flow from, can be NULL if the rest are specified
	66	* @data: raw buffer pointer to the packet, if NULL use skb->data
	67	* @proto: protocol for which to get the flow, if @data is NULL use skb->protocol
	68	* @nhoff: network header offset, if @data is NULL use skb_network_offset(skb)
	69	* @hlen: packet header length, if @data is NULL use skb_headlen(skb)
	70	*
	71	* The function will try to retrieve the struct flow_keys from either the skbuff
	72	* or a raw buffer specified by the rest parameters
	73	*/
	74	bool __skb_flow_dissect(const struct sk_buff skb, struct flow_keys flow,
	75	void *data, __be16 proto, int nhoff, int hlen)
0744dd00	76	{
0744dd00	77	u8 ip_proto;
0744dd00	78
690e36e7 DM	79	if (!data) {
690e36e7 DM	80	data = skb->data;
453a940e WC	81	proto = skb->protocol;
453a940e WC	82	nhoff = skb_network_offset(skb);
690e36e7 DM	83	hlen = skb_headlen(skb);
	84	}
	85
0744dd00 ED	86	memset(flow, 0, sizeof(*flow));
	87
	88	again:
	89	switch (proto) {
2b8837ae	90	case htons(ETH_P_IP): {
0744dd00 ED	91	const struct iphdr *iph;
	92	struct iphdr _iph;
	93	ip:
690e36e7	94	iph = __skb_header_pointer(skb, nhoff, sizeof(_iph), data, hlen, &_iph);
6f092343	95	if (!iph \|\| iph->ihl < 5)
0744dd00	96	return false;
3797d3e8	97	nhoff += iph->ihl * 4;
0744dd00	98
3797d3e8	99	ip_proto = iph->protocol;
0744dd00 ED	100	if (ip_is_fragment(iph))
0744dd00 ED	101	ip_proto = 0;
3797d3e8	102
5af7fb6e AD	103	/* skip the address processing if skb is NULL. The assumption
	104	* here is that if there is no skb we are not looking for flow
	105	* info but lengths and protocols.
	106	*/
	107	if (!skb)
	108	break;
	109
4d77d2b5	110	iph_to_flow_copy_addrs(flow, iph);
0744dd00 ED	111	break;
0744dd00 ED	112	}
2b8837ae	113	case htons(ETH_P_IPV6): {
0744dd00 ED	114	const struct ipv6hdr *iph;
0744dd00 ED	115	struct ipv6hdr _iph;
19469a87 TH	116	__be32 flow_label;
19469a87 TH	117
0744dd00	118	ipv6:
690e36e7	119	iph = __skb_header_pointer(skb, nhoff, sizeof(_iph), data, hlen, &_iph);
0744dd00 ED	120	if (!iph)
	121	return false;
	122
	123	ip_proto = iph->nexthdr;
0744dd00	124	nhoff += sizeof(struct ipv6hdr);
19469a87	125
5af7fb6e	126	/* see comment above in IPv4 section */
56193d1b AD	127	if (!skb)
	128	break;
	129
5af7fb6e AD	130	flow->src = (__force __be32)ipv6_addr_hash(&iph->saddr);
	131	flow->dst = (__force __be32)ipv6_addr_hash(&iph->daddr);
	132
19469a87 TH	133	flow_label = ip6_flowlabel(iph);
	134	if (flow_label) {
	135	/* Awesome, IPv6 packet has a flow label so we can
	136	* use that to represent the ports without any
	137	* further dissection.
	138	*/
	139	flow->n_proto = proto;
	140	flow->ip_proto = ip_proto;
	141	flow->ports = flow_label;
	142	flow->thoff = (u16)nhoff;
	143
	144	return true;
	145	}
	146
0744dd00 ED	147	break;
0744dd00 ED	148	}
2b8837ae JP	149	case htons(ETH_P_8021AD):
2b8837ae JP	150	case htons(ETH_P_8021Q): {
0744dd00 ED	151	const struct vlan_hdr *vlan;
	152	struct vlan_hdr _vlan;
	153
690e36e7	154	vlan = __skb_header_pointer(skb, nhoff, sizeof(_vlan), data, hlen, &_vlan);
0744dd00 ED	155	if (!vlan)
	156	return false;
	157
	158	proto = vlan->h_vlan_encapsulated_proto;
	159	nhoff += sizeof(*vlan);
	160	goto again;
	161	}
2b8837ae	162	case htons(ETH_P_PPP_SES): {
0744dd00 ED	163	struct {
	164	struct pppoe_hdr hdr;
	165	__be16 proto;
	166	} *hdr, _hdr;
690e36e7	167	hdr = __skb_header_pointer(skb, nhoff, sizeof(_hdr), data, hlen, &_hdr);
0744dd00 ED	168	if (!hdr)
	169	return false;
	170	proto = hdr->proto;
	171	nhoff += PPPOE_SES_HLEN;
	172	switch (proto) {
2b8837ae	173	case htons(PPP_IP):
0744dd00	174	goto ip;
2b8837ae	175	case htons(PPP_IPV6):
0744dd00 ED	176	goto ipv6;
	177	default:
	178	return false;
	179	}
	180	}
08bfc9cb EH	181	case htons(ETH_P_TIPC): {
	182	struct {
	183	__be32 pre[3];
	184	__be32 srcnode;
	185	} *hdr, _hdr;
	186	hdr = __skb_header_pointer(skb, nhoff, sizeof(_hdr), data, hlen, &_hdr);
	187	if (!hdr)
	188	return false;
	189	flow->src = hdr->srcnode;
	190	flow->dst = 0;
	191	flow->n_proto = proto;
	192	flow->thoff = (u16)nhoff;
	193	return true;
	194	}
56193d1b AD	195	case htons(ETH_P_FCOE):
	196	flow->thoff = (u16)(nhoff + FCOE_HEADER_LEN);
	197	/* fall through */
0744dd00 ED	198	default:
	199	return false;
	200	}
	201
	202	switch (ip_proto) {
	203	case IPPROTO_GRE: {
	204	struct gre_hdr {
	205	__be16 flags;
	206	__be16 proto;
	207	} *hdr, _hdr;
	208
690e36e7	209	hdr = __skb_header_pointer(skb, nhoff, sizeof(_hdr), data, hlen, &_hdr);
0744dd00 ED	210	if (!hdr)
	211	return false;
	212	/*
	213	* Only look inside GRE if version zero and no
	214	* routing
	215	*/
	216	if (!(hdr->flags & (GRE_VERSION\|GRE_ROUTING))) {
	217	proto = hdr->proto;
	218	nhoff += 4;
	219	if (hdr->flags & GRE_CSUM)
	220	nhoff += 4;
	221	if (hdr->flags & GRE_KEY)
	222	nhoff += 4;
	223	if (hdr->flags & GRE_SEQ)
	224	nhoff += 4;
e1733de2 MD	225	if (proto == htons(ETH_P_TEB)) {
	226	const struct ethhdr *eth;
	227	struct ethhdr _eth;
	228
690e36e7 DM	229	eth = __skb_header_pointer(skb, nhoff,
	230	sizeof(_eth),
	231	data, hlen, &_eth);
e1733de2 MD	232	if (!eth)
	233	return false;
	234	proto = eth->h_proto;
	235	nhoff += sizeof(*eth);
	236	}
0744dd00 ED	237	goto again;
	238	}
	239	break;
	240	}
	241	case IPPROTO_IPIP:
fca41895 TH	242	proto = htons(ETH_P_IP);
fca41895 TH	243	goto ip;
b438f940 TH	244	case IPPROTO_IPV6:
	245	proto = htons(ETH_P_IPV6);
	246	goto ipv6;
0744dd00 ED	247	default:
	248	break;
	249	}
	250
e0f31d84	251	flow->n_proto = proto;
0744dd00	252	flow->ip_proto = ip_proto;
8ed78166 DB	253	flow->thoff = (u16) nhoff;
8ed78166 DB	254
5af7fb6e AD	255	/* unless skb is set we don't need to record port info */
	256	if (skb)
	257	flow->ports = __skb_flow_get_ports(skb, nhoff, ip_proto,
	258	data, hlen);
	259
0744dd00 ED	260	return true;
0744dd00 ED	261	}
690e36e7	262	EXPORT_SYMBOL(__skb_flow_dissect);
441d9d32 CW	263
441d9d32 CW	264	static u32 hashrnd __read_mostly;
66415cf8 HFS	265	static __always_inline void __flow_hash_secret_init(void)
	266	{
	267	net_get_random_once(&hashrnd, sizeof(hashrnd));
	268	}
	269
	270	static __always_inline u32 __flow_hash_3words(u32 a, u32 b, u32 c)
	271	{
	272	__flow_hash_secret_init();
	273	return jhash_3words(a, b, c, hashrnd);
	274	}
	275
5ed20a68 TH	276	static inline u32 __flow_hash_from_keys(struct flow_keys *keys)
	277	{
	278	u32 hash;
	279
	280	/* get a consistent hash (same value on both flow directions) */
	281	if (((__force u32)keys->dst < (__force u32)keys->src) \|\|
	282	(((__force u32)keys->dst == (__force u32)keys->src) &&
	283	((__force u16)keys->port16[1] < (__force u16)keys->port16[0]))) {
	284	swap(keys->dst, keys->src);
	285	swap(keys->port16[0], keys->port16[1]);
	286	}
	287
	288	hash = __flow_hash_3words((__force u32)keys->dst,
	289	(__force u32)keys->src,
	290	(__force u32)keys->ports);
	291	if (!hash)
	292	hash = 1;
	293
	294	return hash;
	295	}
	296
	297	u32 flow_hash_from_keys(struct flow_keys *keys)
	298	{
	299	return __flow_hash_from_keys(keys);
	300	}
	301	EXPORT_SYMBOL(flow_hash_from_keys);
	302
441d9d32	303	/*
3958afa1	304	* __skb_get_hash: calculate a flow hash based on src/dst addresses
61b905da TH	305	* and src/dst port numbers. Sets hash in skb to non-zero hash value
61b905da TH	306	* on success, zero indicates no valid hash. Also, sets l4_hash in skb
441d9d32 CW	307	* if hash is a canonical 4-tuple hash over transport ports.
441d9d32 CW	308	*/
3958afa1	309	void __skb_get_hash(struct sk_buff *skb)
441d9d32 CW	310	{
441d9d32 CW	311	struct flow_keys keys;
441d9d32 CW	312
	313	if (!skb_flow_dissect(skb, &keys))
	314	return;
	315
	316	if (keys.ports)
61b905da	317	skb->l4_hash = 1;
441d9d32	318
a3b18ddb TH	319	skb->sw_hash = 1;
a3b18ddb TH	320
5ed20a68	321	skb->hash = __flow_hash_from_keys(&keys);
441d9d32	322	}
3958afa1	323	EXPORT_SYMBOL(__skb_get_hash);
441d9d32 CW	324
	325	/*
	326	* Returns a Tx hash based on the given packet descriptor a Tx queues' number
	327	* to be used as a distribution range.
	328	*/
0e001614	329	u16 __skb_tx_hash(const struct net_device dev, struct sk_buff skb,
441d9d32 CW	330	unsigned int num_tx_queues)
	331	{
	332	u32 hash;
	333	u16 qoffset = 0;
	334	u16 qcount = num_tx_queues;
	335
	336	if (skb_rx_queue_recorded(skb)) {
	337	hash = skb_get_rx_queue(skb);
	338	while (unlikely(hash >= num_tx_queues))
	339	hash -= num_tx_queues;
	340	return hash;
	341	}
	342
	343	if (dev->num_tc) {
	344	u8 tc = netdev_get_prio_tc_map(dev, skb->priority);
	345	qoffset = dev->tc_to_txq[tc].offset;
	346	qcount = dev->tc_to_txq[tc].count;
	347	}
	348
8fc54f68	349	return (u16) reciprocal_scale(skb_get_hash(skb), qcount) + qoffset;
441d9d32 CW	350	}
	351	EXPORT_SYMBOL(__skb_tx_hash);
	352
56193d1b AD	353	u32 __skb_get_poff(const struct sk_buff skb, void data,
56193d1b AD	354	const struct flow_keys *keys, int hlen)
f77668dc	355	{
56193d1b	356	u32 poff = keys->thoff;
f77668dc	357
56193d1b	358	switch (keys->ip_proto) {
f77668dc	359	case IPPROTO_TCP: {
5af7fb6e AD	360	/* access doff as u8 to avoid unaligned access */
	361	const u8 *doff;
	362	u8 _doff;
f77668dc	363
5af7fb6e AD	364	doff = __skb_header_pointer(skb, poff + 12, sizeof(_doff),
	365	data, hlen, &_doff);
	366	if (!doff)
f77668dc DB	367	return poff;
f77668dc DB	368
5af7fb6e	369	poff += max_t(u32, sizeof(struct tcphdr), (*doff & 0xF0) >> 2);
f77668dc DB	370	break;
	371	}
	372	case IPPROTO_UDP:
	373	case IPPROTO_UDPLITE:
	374	poff += sizeof(struct udphdr);
	375	break;
	376	/* For the rest, we do not really care about header
	377	* extensions at this point for now.
	378	*/
	379	case IPPROTO_ICMP:
	380	poff += sizeof(struct icmphdr);
	381	break;
	382	case IPPROTO_ICMPV6:
	383	poff += sizeof(struct icmp6hdr);
	384	break;
	385	case IPPROTO_IGMP:
	386	poff += sizeof(struct igmphdr);
	387	break;
	388	case IPPROTO_DCCP:
	389	poff += sizeof(struct dccp_hdr);
	390	break;
	391	case IPPROTO_SCTP:
	392	poff += sizeof(struct sctphdr);
	393	break;
	394	}
	395
	396	return poff;
	397	}
	398
56193d1b AD	399	/* skb_get_poff() returns the offset to the payload as far as it could
	400	* be dissected. The main user is currently BPF, so that we can dynamically
	401	* truncate packets without needing to push actual payload to the user
	402	* space and can analyze headers only, instead.
	403	*/
	404	u32 skb_get_poff(const struct sk_buff *skb)
	405	{
	406	struct flow_keys keys;
	407
	408	if (!skb_flow_dissect(skb, &keys))
	409	return 0;
	410
	411	return __skb_get_poff(skb, skb->data, &keys, skb_headlen(skb));
	412	}
	413
441d9d32 CW	414	static inline int get_xps_queue(struct net_device dev, struct sk_buff skb)
	415	{
	416	#ifdef CONFIG_XPS
	417	struct xps_dev_maps *dev_maps;
	418	struct xps_map *map;
	419	int queue_index = -1;
	420
	421	rcu_read_lock();
	422	dev_maps = rcu_dereference(dev->xps_maps);
	423	if (dev_maps) {
	424	map = rcu_dereference(
2bd82484	425	dev_maps->cpu_map[skb->sender_cpu - 1]);
441d9d32 CW	426	if (map) {
	427	if (map->len == 1)
	428	queue_index = map->queues[0];
0e001614	429	else
8fc54f68 DB	430	queue_index = map->queues[reciprocal_scale(skb_get_hash(skb),
8fc54f68 DB	431	map->len)];
441d9d32 CW	432	if (unlikely(queue_index >= dev->real_num_tx_queues))
	433	queue_index = -1;
	434	}
	435	}
	436	rcu_read_unlock();
	437
	438	return queue_index;
	439	#else
	440	return -1;
	441	#endif
	442	}
	443
99932d4f	444	static u16 __netdev_pick_tx(struct net_device dev, struct sk_buff skb)
441d9d32 CW	445	{
	446	struct sock *sk = skb->sk;
	447	int queue_index = sk_tx_queue_get(sk);
	448
	449	if (queue_index < 0 \|\| skb->ooo_okay \|\|
	450	queue_index >= dev->real_num_tx_queues) {
	451	int new_index = get_xps_queue(dev, skb);
	452	if (new_index < 0)
	453	new_index = skb_tx_hash(dev, skb);
	454
702821f4 ED	455	if (queue_index != new_index && sk &&
702821f4 ED	456	rcu_access_pointer(sk->sk_dst_cache))
50d1784e	457	sk_tx_queue_set(sk, new_index);
441d9d32 CW	458
	459	queue_index = new_index;
	460	}
	461
	462	return queue_index;
	463	}
441d9d32 CW	464
441d9d32 CW	465	struct netdev_queue netdev_pick_tx(struct net_device dev,
f663dd9a JW	466	struct sk_buff *skb,
f663dd9a JW	467	void *accel_priv)
441d9d32 CW	468	{
	469	int queue_index = 0;
	470
2bd82484 ED	471	#ifdef CONFIG_XPS
	472	if (skb->sender_cpu == 0)
	473	skb->sender_cpu = raw_smp_processor_id() + 1;
	474	#endif
	475
441d9d32 CW	476	if (dev->real_num_tx_queues != 1) {
	477	const struct net_device_ops *ops = dev->netdev_ops;
	478	if (ops->ndo_select_queue)
99932d4f DB	479	queue_index = ops->ndo_select_queue(dev, skb, accel_priv,
99932d4f DB	480	__netdev_pick_tx);
441d9d32 CW	481	else
441d9d32 CW	482	queue_index = __netdev_pick_tx(dev, skb);
f663dd9a JW	483
f663dd9a JW	484	if (!accel_priv)
b9507bda	485	queue_index = netdev_cap_txqueue(dev, queue_index);
441d9d32 CW	486	}
	487
	488	skb_set_queue_mapping(skb, queue_index);
	489	return netdev_get_tx_queue(dev, queue_index);
	490	}