linux/include/net/ip6_route.h
Wangyang Guo d288a162dd net: dst: Prevent false sharing vs. dst_entry:: __refcnt
dst_entry::__refcnt is highly contended in scenarios where many connections
happen from and to the same IP. The reference count is an atomic_t, so the
reference count operations have to take the cache-line exclusive.

Aside of the unavoidable reference count contention there is another
significant problem which is caused by that: False sharing.

perf top identified two affected read accesses. dst_entry::lwtstate and
rtable::rt_genid.

dst_entry:__refcnt is located at offset 64 of dst_entry, which puts it into
a seperate cacheline vs. the read mostly members located at the beginning
of the struct.

That prevents false sharing vs. the struct members in the first 64
bytes of the structure, but there is also

  dst_entry::lwtstate

which is located after the reference count and in the same cache line. This
member is read after a reference count has been acquired.

struct rtable embeds a struct dst_entry at offset 0. struct dst_entry has a
size of 112 bytes, which means that the struct members of rtable which
follow the dst member share the same cache line as dst_entry::__refcnt.
Especially

  rtable::rt_genid

is also read by the contexts which have a reference count acquired
already.

When dst_entry:__refcnt is incremented or decremented via an atomic
operation these read accesses stall. This was found when analysing the
memtier benchmark in 1:100 mode, which amplifies the problem extremly.

Move the rt[6i]_uncached[_list] members out of struct rtable and struct
rt6_info into struct dst_entry to provide padding and move the lwtstate
member after that so it ends up in the same cache line.

The resulting improvement depends on the micro-architecture and the number
of CPUs. It ranges from +20% to +120% with a localhost memtier/memcached
benchmark.

[ tglx: Rearrange struct ]

Signed-off-by: Wangyang Guo <wangyang.guo@intel.com>
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Link: https://lore.kernel.org/r/20230323102800.042297517@linutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2023-03-28 18:52:22 -07:00

346 lines
9.9 KiB
C

/* SPDX-License-Identifier: GPL-2.0 */
#ifndef _NET_IP6_ROUTE_H
#define _NET_IP6_ROUTE_H
#include <net/addrconf.h>
#include <net/flow.h>
#include <net/ip6_fib.h>
#include <net/sock.h>
#include <net/lwtunnel.h>
#include <linux/ip.h>
#include <linux/ipv6.h>
#include <linux/route.h>
#include <net/nexthop.h>
struct route_info {
__u8 type;
__u8 length;
__u8 prefix_len;
#if defined(__BIG_ENDIAN_BITFIELD)
__u8 reserved_h:3,
route_pref:2,
reserved_l:3;
#elif defined(__LITTLE_ENDIAN_BITFIELD)
__u8 reserved_l:3,
route_pref:2,
reserved_h:3;
#endif
__be32 lifetime;
__u8 prefix[]; /* 0,8 or 16 */
};
#define RT6_LOOKUP_F_IFACE 0x00000001
#define RT6_LOOKUP_F_REACHABLE 0x00000002
#define RT6_LOOKUP_F_HAS_SADDR 0x00000004
#define RT6_LOOKUP_F_SRCPREF_TMP 0x00000008
#define RT6_LOOKUP_F_SRCPREF_PUBLIC 0x00000010
#define RT6_LOOKUP_F_SRCPREF_COA 0x00000020
#define RT6_LOOKUP_F_IGNORE_LINKSTATE 0x00000040
#define RT6_LOOKUP_F_DST_NOREF 0x00000080
/* We do not (yet ?) support IPv6 jumbograms (RFC 2675)
* Unlike IPv4, hdr->seg_len doesn't include the IPv6 header
*/
#define IP6_MAX_MTU (0xFFFF + sizeof(struct ipv6hdr))
/*
* rt6_srcprefs2flags() and rt6_flags2srcprefs() translate
* between IPV6_ADDR_PREFERENCES socket option values
* IPV6_PREFER_SRC_TMP = 0x1
* IPV6_PREFER_SRC_PUBLIC = 0x2
* IPV6_PREFER_SRC_COA = 0x4
* and above RT6_LOOKUP_F_SRCPREF_xxx flags.
*/
static inline int rt6_srcprefs2flags(unsigned int srcprefs)
{
/* No need to bitmask because srcprefs have only 3 bits. */
return srcprefs << 3;
}
static inline unsigned int rt6_flags2srcprefs(int flags)
{
return (flags >> 3) & 7;
}
static inline bool rt6_need_strict(const struct in6_addr *daddr)
{
return ipv6_addr_type(daddr) &
(IPV6_ADDR_MULTICAST | IPV6_ADDR_LINKLOCAL | IPV6_ADDR_LOOPBACK);
}
/* fib entries using a nexthop object can not be coalesced into
* a multipath route
*/
static inline bool rt6_qualify_for_ecmp(const struct fib6_info *f6i)
{
/* the RTF_ADDRCONF flag filters out RA's */
return !(f6i->fib6_flags & RTF_ADDRCONF) && !f6i->nh &&
f6i->fib6_nh->fib_nh_gw_family;
}
void ip6_route_input(struct sk_buff *skb);
struct dst_entry *ip6_route_input_lookup(struct net *net,
struct net_device *dev,
struct flowi6 *fl6,
const struct sk_buff *skb, int flags);
struct dst_entry *ip6_route_output_flags(struct net *net, const struct sock *sk,
struct flowi6 *fl6, int flags);
static inline struct dst_entry *ip6_route_output(struct net *net,
const struct sock *sk,
struct flowi6 *fl6)
{
return ip6_route_output_flags(net, sk, fl6, 0);
}
/* Only conditionally release dst if flags indicates
* !RT6_LOOKUP_F_DST_NOREF or dst is in uncached_list.
*/
static inline void ip6_rt_put_flags(struct rt6_info *rt, int flags)
{
if (!(flags & RT6_LOOKUP_F_DST_NOREF) ||
!list_empty(&rt->dst.rt_uncached))
ip6_rt_put(rt);
}
struct dst_entry *ip6_route_lookup(struct net *net, struct flowi6 *fl6,
const struct sk_buff *skb, int flags);
struct rt6_info *ip6_pol_route(struct net *net, struct fib6_table *table,
int ifindex, struct flowi6 *fl6,
const struct sk_buff *skb, int flags);
void ip6_route_init_special_entries(void);
int ip6_route_init(void);
void ip6_route_cleanup(void);
int ipv6_route_ioctl(struct net *net, unsigned int cmd,
struct in6_rtmsg *rtmsg);
int ip6_route_add(struct fib6_config *cfg, gfp_t gfp_flags,
struct netlink_ext_ack *extack);
int ip6_ins_rt(struct net *net, struct fib6_info *f6i);
int ip6_del_rt(struct net *net, struct fib6_info *f6i, bool skip_notify);
void rt6_flush_exceptions(struct fib6_info *f6i);
void rt6_age_exceptions(struct fib6_info *f6i, struct fib6_gc_args *gc_args,
unsigned long now);
static inline int ip6_route_get_saddr(struct net *net, struct fib6_info *f6i,
const struct in6_addr *daddr,
unsigned int prefs,
struct in6_addr *saddr)
{
int err = 0;
if (f6i && f6i->fib6_prefsrc.plen) {
*saddr = f6i->fib6_prefsrc.addr;
} else {
struct net_device *dev = f6i ? fib6_info_nh_dev(f6i) : NULL;
err = ipv6_dev_get_saddr(net, dev, daddr, prefs, saddr);
}
return err;
}
struct rt6_info *rt6_lookup(struct net *net, const struct in6_addr *daddr,
const struct in6_addr *saddr, int oif,
const struct sk_buff *skb, int flags);
u32 rt6_multipath_hash(const struct net *net, const struct flowi6 *fl6,
const struct sk_buff *skb, struct flow_keys *hkeys);
struct dst_entry *icmp6_dst_alloc(struct net_device *dev, struct flowi6 *fl6);
void fib6_force_start_gc(struct net *net);
struct fib6_info *addrconf_f6i_alloc(struct net *net, struct inet6_dev *idev,
const struct in6_addr *addr, bool anycast,
gfp_t gfp_flags);
struct rt6_info *ip6_dst_alloc(struct net *net, struct net_device *dev,
int flags);
/*
* support functions for ND
*
*/
struct fib6_info *rt6_get_dflt_router(struct net *net,
const struct in6_addr *addr,
struct net_device *dev);
struct fib6_info *rt6_add_dflt_router(struct net *net,
const struct in6_addr *gwaddr,
struct net_device *dev, unsigned int pref,
u32 defrtr_usr_metric);
void rt6_purge_dflt_routers(struct net *net);
int rt6_route_rcv(struct net_device *dev, u8 *opt, int len,
const struct in6_addr *gwaddr);
void ip6_update_pmtu(struct sk_buff *skb, struct net *net, __be32 mtu, int oif,
u32 mark, kuid_t uid);
void ip6_sk_update_pmtu(struct sk_buff *skb, struct sock *sk, __be32 mtu);
void ip6_redirect(struct sk_buff *skb, struct net *net, int oif, u32 mark,
kuid_t uid);
void ip6_redirect_no_header(struct sk_buff *skb, struct net *net, int oif);
void ip6_sk_redirect(struct sk_buff *skb, struct sock *sk);
struct netlink_callback;
struct rt6_rtnl_dump_arg {
struct sk_buff *skb;
struct netlink_callback *cb;
struct net *net;
struct fib_dump_filter filter;
};
int rt6_dump_route(struct fib6_info *f6i, void *p_arg, unsigned int skip);
void rt6_mtu_change(struct net_device *dev, unsigned int mtu);
void rt6_remove_prefsrc(struct inet6_ifaddr *ifp);
void rt6_clean_tohost(struct net *net, struct in6_addr *gateway);
void rt6_sync_up(struct net_device *dev, unsigned char nh_flags);
void rt6_disable_ip(struct net_device *dev, unsigned long event);
void rt6_sync_down_dev(struct net_device *dev, unsigned long event);
void rt6_multipath_rebalance(struct fib6_info *f6i);
void rt6_uncached_list_add(struct rt6_info *rt);
void rt6_uncached_list_del(struct rt6_info *rt);
static inline const struct rt6_info *skb_rt6_info(const struct sk_buff *skb)
{
const struct dst_entry *dst = skb_dst(skb);
const struct rt6_info *rt6 = NULL;
if (dst)
rt6 = container_of(dst, struct rt6_info, dst);
return rt6;
}
/*
* Store a destination cache entry in a socket
*/
static inline void ip6_dst_store(struct sock *sk, struct dst_entry *dst,
const struct in6_addr *daddr,
const struct in6_addr *saddr)
{
struct ipv6_pinfo *np = inet6_sk(sk);
np->dst_cookie = rt6_get_cookie((struct rt6_info *)dst);
sk_setup_caps(sk, dst);
np->daddr_cache = daddr;
#ifdef CONFIG_IPV6_SUBTREES
np->saddr_cache = saddr;
#endif
}
void ip6_sk_dst_store_flow(struct sock *sk, struct dst_entry *dst,
const struct flowi6 *fl6);
static inline bool ipv6_unicast_destination(const struct sk_buff *skb)
{
struct rt6_info *rt = (struct rt6_info *) skb_dst(skb);
return rt->rt6i_flags & RTF_LOCAL;
}
static inline bool ipv6_anycast_destination(const struct dst_entry *dst,
const struct in6_addr *daddr)
{
struct rt6_info *rt = (struct rt6_info *)dst;
return rt->rt6i_flags & RTF_ANYCAST ||
(rt->rt6i_dst.plen < 127 &&
!(rt->rt6i_flags & (RTF_GATEWAY | RTF_NONEXTHOP)) &&
ipv6_addr_equal(&rt->rt6i_dst.addr, daddr));
}
int ip6_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
int (*output)(struct net *, struct sock *, struct sk_buff *));
static inline unsigned int ip6_skb_dst_mtu(const struct sk_buff *skb)
{
const struct ipv6_pinfo *np = skb->sk && !dev_recursion_level() ?
inet6_sk(skb->sk) : NULL;
const struct dst_entry *dst = skb_dst(skb);
unsigned int mtu;
if (np && np->pmtudisc >= IPV6_PMTUDISC_PROBE) {
mtu = READ_ONCE(dst->dev->mtu);
mtu -= lwtunnel_headroom(dst->lwtstate, mtu);
} else {
mtu = dst_mtu(dst);
}
return mtu;
}
static inline bool ip6_sk_accept_pmtu(const struct sock *sk)
{
return inet6_sk(sk)->pmtudisc != IPV6_PMTUDISC_INTERFACE &&
inet6_sk(sk)->pmtudisc != IPV6_PMTUDISC_OMIT;
}
static inline bool ip6_sk_ignore_df(const struct sock *sk)
{
return inet6_sk(sk)->pmtudisc < IPV6_PMTUDISC_DO ||
inet6_sk(sk)->pmtudisc == IPV6_PMTUDISC_OMIT;
}
static inline const struct in6_addr *rt6_nexthop(const struct rt6_info *rt,
const struct in6_addr *daddr)
{
if (rt->rt6i_flags & RTF_GATEWAY)
return &rt->rt6i_gateway;
else if (unlikely(rt->rt6i_flags & RTF_CACHE))
return &rt->rt6i_dst.addr;
else
return daddr;
}
static inline bool rt6_duplicate_nexthop(struct fib6_info *a, struct fib6_info *b)
{
struct fib6_nh *nha, *nhb;
if (a->nh || b->nh)
return nexthop_cmp(a->nh, b->nh);
nha = a->fib6_nh;
nhb = b->fib6_nh;
return nha->fib_nh_dev == nhb->fib_nh_dev &&
ipv6_addr_equal(&nha->fib_nh_gw6, &nhb->fib_nh_gw6) &&
!lwtunnel_cmp_encap(nha->fib_nh_lws, nhb->fib_nh_lws);
}
static inline unsigned int ip6_dst_mtu_maybe_forward(const struct dst_entry *dst,
bool forwarding)
{
struct inet6_dev *idev;
unsigned int mtu;
if (!forwarding || dst_metric_locked(dst, RTAX_MTU)) {
mtu = dst_metric_raw(dst, RTAX_MTU);
if (mtu)
goto out;
}
mtu = IPV6_MIN_MTU;
rcu_read_lock();
idev = __in6_dev_get(dst->dev);
if (idev)
mtu = idev->cnf.mtu6;
rcu_read_unlock();
out:
return mtu - lwtunnel_headroom(dst->lwtstate, mtu);
}
u32 ip6_mtu_from_fib6(const struct fib6_result *res,
const struct in6_addr *daddr,
const struct in6_addr *saddr);
struct neighbour *ip6_neigh_lookup(const struct in6_addr *gw,
struct net_device *dev, struct sk_buff *skb,
const void *daddr);
#endif