2011-11-28 13:22:18 +08:00
|
|
|
#include <linux/skbuff.h>
|
2012-01-25 05:03:33 +08:00
|
|
|
#include <linux/export.h>
|
2011-11-28 13:22:18 +08:00
|
|
|
#include <linux/ip.h>
|
|
|
|
#include <linux/ipv6.h>
|
|
|
|
#include <linux/if_vlan.h>
|
|
|
|
#include <net/ip.h>
|
2012-07-18 16:11:12 +08:00
|
|
|
#include <net/ipv6.h>
|
2013-03-19 14:39:30 +08:00
|
|
|
#include <linux/igmp.h>
|
|
|
|
#include <linux/icmp.h>
|
|
|
|
#include <linux/sctp.h>
|
|
|
|
#include <linux/dccp.h>
|
2011-11-28 13:22:18 +08:00
|
|
|
#include <linux/if_tunnel.h>
|
|
|
|
#include <linux/if_pppox.h>
|
|
|
|
#include <linux/ppp_defs.h>
|
|
|
|
#include <net/flow_keys.h>
|
|
|
|
|
flow_dissector: use a 64bit load/store
Le lundi 28 novembre 2011 à 19:06 -0500, David Miller a écrit :
> From: Dimitris Michailidis <dm@chelsio.com>
> Date: Mon, 28 Nov 2011 08:25:39 -0800
>
> >> +bool skb_flow_dissect(const struct sk_buff *skb, struct flow_keys
> >> *flow)
> >> +{
> >> + int poff, nhoff = skb_network_offset(skb);
> >> + u8 ip_proto;
> >> + u16 proto = skb->protocol;
> >
> > __be16 instead of u16 for proto?
>
> I'll take care of this when I apply these patches.
( CC trimmed )
Thanks David !
Here is a small patch to use one 64bit load/store on x86_64 instead of
two 32bit load/stores.
[PATCH net-next] flow_dissector: use a 64bit load/store
gcc compiler is smart enough to use a single load/store if we
memcpy(dptr, sptr, 8) on x86_64, regardless of
CONFIG_CC_OPTIMIZE_FOR_SIZE
In IP header, daddr immediately follows saddr, this wont change in the
future. We only need to make sure our flow_keys (src,dst) fields wont
break the rule.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-29 04:30:35 +08:00
|
|
|
/* 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));
|
|
|
|
}
|
2011-11-28 13:22:18 +08:00
|
|
|
|
2013-10-02 19:39:24 +08:00
|
|
|
/**
|
2014-08-26 08:03:46 +08:00
|
|
|
* __skb_flow_get_ports - extract the upper layer ports and return them
|
|
|
|
* @skb: sk_buff to extract the ports from
|
2013-10-02 19:39:24 +08:00
|
|
|
* @thoff: transport header offset
|
|
|
|
* @ip_proto: protocol for which to get port offset
|
2014-08-26 08:03:46 +08:00
|
|
|
* @data: raw buffer pointer to the packet, if NULL use skb->data
|
|
|
|
* @hlen: packet header length, if @data is NULL use skb_headlen(skb)
|
2013-10-02 19:39:24 +08:00
|
|
|
*
|
|
|
|
* The function will try to retrieve the ports at offset thoff + poff where poff
|
|
|
|
* is the protocol port offset returned from proto_ports_offset
|
|
|
|
*/
|
2014-08-24 03:13:41 +08:00
|
|
|
__be32 __skb_flow_get_ports(const struct sk_buff *skb, int thoff, u8 ip_proto,
|
|
|
|
void *data, int hlen)
|
2013-10-02 19:39:24 +08:00
|
|
|
{
|
|
|
|
int poff = proto_ports_offset(ip_proto);
|
|
|
|
|
2014-08-24 03:13:41 +08:00
|
|
|
if (!data) {
|
|
|
|
data = skb->data;
|
|
|
|
hlen = skb_headlen(skb);
|
|
|
|
}
|
|
|
|
|
2013-10-02 19:39:24 +08:00
|
|
|
if (poff >= 0) {
|
|
|
|
__be32 *ports, _ports;
|
|
|
|
|
2014-08-24 03:13:41 +08:00
|
|
|
ports = __skb_header_pointer(skb, thoff + poff,
|
|
|
|
sizeof(_ports), data, hlen, &_ports);
|
2013-10-02 19:39:24 +08:00
|
|
|
if (ports)
|
|
|
|
return *ports;
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
2014-08-24 03:13:41 +08:00
|
|
|
EXPORT_SYMBOL(__skb_flow_get_ports);
|
2013-10-02 19:39:24 +08:00
|
|
|
|
2014-08-26 08:03:47 +08:00
|
|
|
/**
|
|
|
|
* __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)
|
2011-11-28 13:22:18 +08:00
|
|
|
{
|
|
|
|
u8 ip_proto;
|
|
|
|
|
2014-08-24 03:13:41 +08:00
|
|
|
if (!data) {
|
|
|
|
data = skb->data;
|
2014-08-26 08:03:47 +08:00
|
|
|
proto = skb->protocol;
|
|
|
|
nhoff = skb_network_offset(skb);
|
2014-08-24 03:13:41 +08:00
|
|
|
hlen = skb_headlen(skb);
|
|
|
|
}
|
|
|
|
|
2011-11-28 13:22:18 +08:00
|
|
|
memset(flow, 0, sizeof(*flow));
|
|
|
|
|
|
|
|
again:
|
|
|
|
switch (proto) {
|
2014-03-13 01:04:17 +08:00
|
|
|
case htons(ETH_P_IP): {
|
2011-11-28 13:22:18 +08:00
|
|
|
const struct iphdr *iph;
|
|
|
|
struct iphdr _iph;
|
|
|
|
ip:
|
2014-08-24 03:13:41 +08:00
|
|
|
iph = __skb_header_pointer(skb, nhoff, sizeof(_iph), data, hlen, &_iph);
|
2013-11-01 15:01:10 +08:00
|
|
|
if (!iph || iph->ihl < 5)
|
2011-11-28 13:22:18 +08:00
|
|
|
return false;
|
2013-11-08 00:37:28 +08:00
|
|
|
nhoff += iph->ihl * 4;
|
2011-11-28 13:22:18 +08:00
|
|
|
|
2013-11-08 00:37:28 +08:00
|
|
|
ip_proto = iph->protocol;
|
2011-11-28 13:22:18 +08:00
|
|
|
if (ip_is_fragment(iph))
|
|
|
|
ip_proto = 0;
|
2013-11-08 00:37:28 +08:00
|
|
|
|
flow_dissector: use a 64bit load/store
Le lundi 28 novembre 2011 à 19:06 -0500, David Miller a écrit :
> From: Dimitris Michailidis <dm@chelsio.com>
> Date: Mon, 28 Nov 2011 08:25:39 -0800
>
> >> +bool skb_flow_dissect(const struct sk_buff *skb, struct flow_keys
> >> *flow)
> >> +{
> >> + int poff, nhoff = skb_network_offset(skb);
> >> + u8 ip_proto;
> >> + u16 proto = skb->protocol;
> >
> > __be16 instead of u16 for proto?
>
> I'll take care of this when I apply these patches.
( CC trimmed )
Thanks David !
Here is a small patch to use one 64bit load/store on x86_64 instead of
two 32bit load/stores.
[PATCH net-next] flow_dissector: use a 64bit load/store
gcc compiler is smart enough to use a single load/store if we
memcpy(dptr, sptr, 8) on x86_64, regardless of
CONFIG_CC_OPTIMIZE_FOR_SIZE
In IP header, daddr immediately follows saddr, this wont change in the
future. We only need to make sure our flow_keys (src,dst) fields wont
break the rule.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2011-11-29 04:30:35 +08:00
|
|
|
iph_to_flow_copy_addrs(flow, iph);
|
2011-11-28 13:22:18 +08:00
|
|
|
break;
|
|
|
|
}
|
2014-03-13 01:04:17 +08:00
|
|
|
case htons(ETH_P_IPV6): {
|
2011-11-28 13:22:18 +08:00
|
|
|
const struct ipv6hdr *iph;
|
|
|
|
struct ipv6hdr _iph;
|
2014-07-02 12:33:01 +08:00
|
|
|
__be32 flow_label;
|
|
|
|
|
2011-11-28 13:22:18 +08:00
|
|
|
ipv6:
|
2014-08-24 03:13:41 +08:00
|
|
|
iph = __skb_header_pointer(skb, nhoff, sizeof(_iph), data, hlen, &_iph);
|
2011-11-28 13:22:18 +08:00
|
|
|
if (!iph)
|
|
|
|
return false;
|
|
|
|
|
|
|
|
ip_proto = iph->nexthdr;
|
2012-07-18 16:11:12 +08:00
|
|
|
flow->src = (__force __be32)ipv6_addr_hash(&iph->saddr);
|
|
|
|
flow->dst = (__force __be32)ipv6_addr_hash(&iph->daddr);
|
2011-11-28 13:22:18 +08:00
|
|
|
nhoff += sizeof(struct ipv6hdr);
|
2014-07-02 12:33:01 +08:00
|
|
|
|
|
|
|
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;
|
|
|
|
}
|
|
|
|
|
2011-11-28 13:22:18 +08:00
|
|
|
break;
|
|
|
|
}
|
2014-03-13 01:04:17 +08:00
|
|
|
case htons(ETH_P_8021AD):
|
|
|
|
case htons(ETH_P_8021Q): {
|
2011-11-28 13:22:18 +08:00
|
|
|
const struct vlan_hdr *vlan;
|
|
|
|
struct vlan_hdr _vlan;
|
|
|
|
|
2014-08-24 03:13:41 +08:00
|
|
|
vlan = __skb_header_pointer(skb, nhoff, sizeof(_vlan), data, hlen, &_vlan);
|
2011-11-28 13:22:18 +08:00
|
|
|
if (!vlan)
|
|
|
|
return false;
|
|
|
|
|
|
|
|
proto = vlan->h_vlan_encapsulated_proto;
|
|
|
|
nhoff += sizeof(*vlan);
|
|
|
|
goto again;
|
|
|
|
}
|
2014-03-13 01:04:17 +08:00
|
|
|
case htons(ETH_P_PPP_SES): {
|
2011-11-28 13:22:18 +08:00
|
|
|
struct {
|
|
|
|
struct pppoe_hdr hdr;
|
|
|
|
__be16 proto;
|
|
|
|
} *hdr, _hdr;
|
2014-08-24 03:13:41 +08:00
|
|
|
hdr = __skb_header_pointer(skb, nhoff, sizeof(_hdr), data, hlen, &_hdr);
|
2011-11-28 13:22:18 +08:00
|
|
|
if (!hdr)
|
|
|
|
return false;
|
|
|
|
proto = hdr->proto;
|
|
|
|
nhoff += PPPOE_SES_HLEN;
|
|
|
|
switch (proto) {
|
2014-03-13 01:04:17 +08:00
|
|
|
case htons(PPP_IP):
|
2011-11-28 13:22:18 +08:00
|
|
|
goto ip;
|
2014-03-13 01:04:17 +08:00
|
|
|
case htons(PPP_IPV6):
|
2011-11-28 13:22:18 +08:00
|
|
|
goto ipv6;
|
|
|
|
default:
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
default:
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
switch (ip_proto) {
|
|
|
|
case IPPROTO_GRE: {
|
|
|
|
struct gre_hdr {
|
|
|
|
__be16 flags;
|
|
|
|
__be16 proto;
|
|
|
|
} *hdr, _hdr;
|
|
|
|
|
2014-08-24 03:13:41 +08:00
|
|
|
hdr = __skb_header_pointer(skb, nhoff, sizeof(_hdr), data, hlen, &_hdr);
|
2011-11-28 13:22:18 +08:00
|
|
|
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;
|
2013-03-11 14:52:28 +08:00
|
|
|
if (proto == htons(ETH_P_TEB)) {
|
|
|
|
const struct ethhdr *eth;
|
|
|
|
struct ethhdr _eth;
|
|
|
|
|
2014-08-24 03:13:41 +08:00
|
|
|
eth = __skb_header_pointer(skb, nhoff,
|
|
|
|
sizeof(_eth),
|
|
|
|
data, hlen, &_eth);
|
2013-03-11 14:52:28 +08:00
|
|
|
if (!eth)
|
|
|
|
return false;
|
|
|
|
proto = eth->h_proto;
|
|
|
|
nhoff += sizeof(*eth);
|
|
|
|
}
|
2011-11-28 13:22:18 +08:00
|
|
|
goto again;
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
case IPPROTO_IPIP:
|
2013-07-30 02:07:36 +08:00
|
|
|
proto = htons(ETH_P_IP);
|
|
|
|
goto ip;
|
2013-07-30 02:07:42 +08:00
|
|
|
case IPPROTO_IPV6:
|
|
|
|
proto = htons(ETH_P_IPV6);
|
|
|
|
goto ipv6;
|
2011-11-28 13:22:18 +08:00
|
|
|
default:
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2014-06-23 18:37:58 +08:00
|
|
|
flow->n_proto = proto;
|
2011-11-28 13:22:18 +08:00
|
|
|
flow->ip_proto = ip_proto;
|
2014-08-24 03:13:41 +08:00
|
|
|
flow->ports = __skb_flow_get_ports(skb, nhoff, ip_proto, data, hlen);
|
2013-03-19 14:39:29 +08:00
|
|
|
flow->thoff = (u16) nhoff;
|
|
|
|
|
2011-11-28 13:22:18 +08:00
|
|
|
return true;
|
|
|
|
}
|
2014-08-24 03:13:41 +08:00
|
|
|
EXPORT_SYMBOL(__skb_flow_dissect);
|
2013-01-21 08:39:24 +08:00
|
|
|
|
|
|
|
static u32 hashrnd __read_mostly;
|
2013-10-24 02:06:00 +08:00
|
|
|
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);
|
|
|
|
}
|
|
|
|
|
2014-07-02 12:32:05 +08:00
|
|
|
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);
|
|
|
|
|
2013-01-21 08:39:24 +08:00
|
|
|
/*
|
2013-12-16 14:12:06 +08:00
|
|
|
* __skb_get_hash: calculate a flow hash based on src/dst addresses
|
2014-03-25 06:34:47 +08:00
|
|
|
* 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
|
2013-01-21 08:39:24 +08:00
|
|
|
* if hash is a canonical 4-tuple hash over transport ports.
|
|
|
|
*/
|
2013-12-16 14:12:06 +08:00
|
|
|
void __skb_get_hash(struct sk_buff *skb)
|
2013-01-21 08:39:24 +08:00
|
|
|
{
|
|
|
|
struct flow_keys keys;
|
|
|
|
|
|
|
|
if (!skb_flow_dissect(skb, &keys))
|
|
|
|
return;
|
|
|
|
|
|
|
|
if (keys.ports)
|
2014-03-25 06:34:47 +08:00
|
|
|
skb->l4_hash = 1;
|
2013-01-21 08:39:24 +08:00
|
|
|
|
2014-07-02 12:33:17 +08:00
|
|
|
skb->sw_hash = 1;
|
|
|
|
|
2014-07-02 12:32:05 +08:00
|
|
|
skb->hash = __flow_hash_from_keys(&keys);
|
2013-01-21 08:39:24 +08:00
|
|
|
}
|
2013-12-16 14:12:06 +08:00
|
|
|
EXPORT_SYMBOL(__skb_get_hash);
|
2013-01-21 08:39:24 +08:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Returns a Tx hash based on the given packet descriptor a Tx queues' number
|
|
|
|
* to be used as a distribution range.
|
|
|
|
*/
|
2014-07-02 12:32:27 +08:00
|
|
|
u16 __skb_tx_hash(const struct net_device *dev, struct sk_buff *skb,
|
2013-01-21 08:39:24 +08:00
|
|
|
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;
|
|
|
|
}
|
|
|
|
|
2014-08-24 02:58:54 +08:00
|
|
|
return (u16) reciprocal_scale(skb_get_hash(skb), qcount) + qoffset;
|
2013-01-21 08:39:24 +08:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL(__skb_tx_hash);
|
|
|
|
|
2013-03-19 14:39:30 +08:00
|
|
|
/* __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;
|
|
|
|
u32 poff = 0;
|
|
|
|
|
|
|
|
if (!skb_flow_dissect(skb, &keys))
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
poff += keys.thoff;
|
|
|
|
switch (keys.ip_proto) {
|
|
|
|
case IPPROTO_TCP: {
|
|
|
|
const struct tcphdr *tcph;
|
|
|
|
struct tcphdr _tcph;
|
|
|
|
|
|
|
|
tcph = skb_header_pointer(skb, poff, sizeof(_tcph), &_tcph);
|
|
|
|
if (!tcph)
|
|
|
|
return poff;
|
|
|
|
|
|
|
|
poff += max_t(u32, sizeof(struct tcphdr), tcph->doff * 4);
|
|
|
|
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;
|
|
|
|
}
|
|
|
|
|
2013-01-21 08:39:24 +08:00
|
|
|
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[raw_smp_processor_id()]);
|
|
|
|
if (map) {
|
|
|
|
if (map->len == 1)
|
|
|
|
queue_index = map->queues[0];
|
2014-07-02 12:32:27 +08:00
|
|
|
else
|
2014-08-24 02:58:54 +08:00
|
|
|
queue_index = map->queues[reciprocal_scale(skb_get_hash(skb),
|
|
|
|
map->len)];
|
2013-01-21 08:39:24 +08:00
|
|
|
if (unlikely(queue_index >= dev->real_num_tx_queues))
|
|
|
|
queue_index = -1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
rcu_read_unlock();
|
|
|
|
|
|
|
|
return queue_index;
|
|
|
|
#else
|
|
|
|
return -1;
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2014-02-16 22:55:20 +08:00
|
|
|
static u16 __netdev_pick_tx(struct net_device *dev, struct sk_buff *skb)
|
2013-01-21 08:39:24 +08:00
|
|
|
{
|
|
|
|
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);
|
|
|
|
|
2013-08-29 09:10:43 +08:00
|
|
|
if (queue_index != new_index && sk &&
|
|
|
|
rcu_access_pointer(sk->sk_dst_cache))
|
2013-09-08 03:02:57 +08:00
|
|
|
sk_tx_queue_set(sk, new_index);
|
2013-01-21 08:39:24 +08:00
|
|
|
|
|
|
|
queue_index = new_index;
|
|
|
|
}
|
|
|
|
|
|
|
|
return queue_index;
|
|
|
|
}
|
|
|
|
|
|
|
|
struct netdev_queue *netdev_pick_tx(struct net_device *dev,
|
2014-01-10 16:18:26 +08:00
|
|
|
struct sk_buff *skb,
|
|
|
|
void *accel_priv)
|
2013-01-21 08:39:24 +08:00
|
|
|
{
|
|
|
|
int queue_index = 0;
|
|
|
|
|
|
|
|
if (dev->real_num_tx_queues != 1) {
|
|
|
|
const struct net_device_ops *ops = dev->netdev_ops;
|
|
|
|
if (ops->ndo_select_queue)
|
2014-02-16 22:55:20 +08:00
|
|
|
queue_index = ops->ndo_select_queue(dev, skb, accel_priv,
|
|
|
|
__netdev_pick_tx);
|
2013-01-21 08:39:24 +08:00
|
|
|
else
|
|
|
|
queue_index = __netdev_pick_tx(dev, skb);
|
2014-01-10 16:18:26 +08:00
|
|
|
|
|
|
|
if (!accel_priv)
|
2014-02-16 22:55:21 +08:00
|
|
|
queue_index = netdev_cap_txqueue(dev, queue_index);
|
2013-01-21 08:39:24 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
skb_set_queue_mapping(skb, queue_index);
|
|
|
|
return netdev_get_tx_queue(dev, queue_index);
|
|
|
|
}
|