struct dst_ops tracks number of allocated dst in an atomic_t field,
subject to high cache line contention in stress workload.

Switch to a percpu_counter, to reduce number of time we need to dirty a
central location. Place it on a separate cache line to avoid dirtying
read only fields.

Stress test :

(Sending 160.000.000 UDP frames,
IP route cache disabled, dual E5540 @2.53GHz,
32bit kernel, FIB_TRIE, SLUB/NUMA)

Before:

real    0m51.179s
user    0m15.329s
sys     10m15.942s

After:

real	0m45.570s
user	0m15.525s
sys	9m56.669s

With a small reordering of struct neighbour fields, subject of a
following patch, (to separate refcnt from other read mostly fields)

real	0m41.841s
user	0m15.261s
sys	8m45.949s
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

David

This is the first step for RCU conversion of neigh code.

Next patches will convert hash_buckets[] and "struct neighbour" to RCU
protected objects.

Thanks

[PATCH net-next] net neigh: RCU conversion of neigh hash table

Instead of storing hash_buckets, hash_mask and hash_rnd in "struct
neigh_table", a new structure is defined :

struct neigh_hash_table {
       struct neighbour        **hash_buckets;
       unsigned int            hash_mask;
       __u32                   hash_rnd;
       struct rcu_head         rcu;
};

And "struct neigh_table" has an RCU protected pointer to such a
neigh_hash_table.

This means the signature of (*hash)() function changed: We need to add a
third parameter with the actual hash_rnd value, since this is not
anymore a neigh_table field.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In various situations, a device provides a packet to our stack and we
drop it before it enters protocol stack :
- softnet backlog full (accounted in /proc/net/softnet_stat)
- bad vlan tag (not accounted)
- unknown/unregistered protocol (not accounted)

We can handle a per-device counter of such dropped frames at core level,
and automatically adds it to the device provided stats (rx_dropped), so
that standard tools can be used (ifconfig, ip link, cat /proc/net/dev)

This is a generalization of commit 8990f468 (net: rx_dropped
accounting), thus reverting it.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Another exported symbol only used in one file
Signed-off-by: NStephen Hemminger <shemminger@vyatta.com>
Acked-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NMaciej Żenczykowski <maze@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Maintain per_cpu tx_bytes, tx_packets, rx_bytes, rx_packets.

Other seldom used fields are kept in netdev->stats structure, possibly
unsafe.

This is a preliminary work to support lockless transmit path, and
correct RX stats, that are already unsafe.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

SIT tunnels can benefit from lockless xmits, using NETIF_F_LLTX

Bench on a 16 cpus machine (dual E5540 cpus), 16 threads sending
10000000 UDP frames via one sit tunnel (size:220 bytes per frame)

Before patch :

real	3m15.399s
user	0m9.185s
sys	51m55.403s

75029.00 87.5% _raw_spin_lock            vmlinux
 1090.00  1.3% dst_release               vmlinux
  902.00  1.1% dev_queue_xmit            vmlinux
  627.00  0.7% sock_wfree                vmlinux
  613.00  0.7% ip6_push_pending_frames   ipv6.ko
  505.00  0.6% __ip_route_output_key     vmlinux

After patch:

real	1m1.387s
user	0m12.489s
sys	15m58.868s

28239.00 23.3% dst_release               vmlinux
13570.00 11.2% ip6_push_pending_frames   ipv6.ko
13118.00 10.8% ip6_append_data           ipv6.ko
 7995.00  6.6% __ip_route_output_key     vmlinux
 7924.00  6.5% sk_dst_check              vmlinux
 5015.00  4.1% udpv6_sendmsg             ipv6.ko
 3594.00  3.0% sock_alloc_send_pskb      vmlinux
 3135.00  2.6% sock_wfree                vmlinux
 3055.00  2.5% ip6_sk_dst_lookup         ipv6.ko
 2473.00  2.0% ip_finish_output          vmlinux
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

commit 15fc1f70 (sit: percpu stats accounting) forgot the fallback
tunnel case (sit0), and can crash pretty fast.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

AnyIP is the capability to receive packets and establish incoming
connections on IPs we have not explicitly configured on the machine.

An example use case is to configure a machine to accept all incoming
traffic on eth0, and leave the policy of whether traffic for a given IP
should be delivered to the machine up to the load balancer.

Can be setup as follows:
  ip -6 rule from all iif eth0 lookup 200
  ip -6 route add local default dev lo table 200
(in this case for all IPv6 addresses)
Signed-off-by: NMaciej Żenczykowski <maze@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Maintain per_cpu tx_bytes, tx_packets, rx_bytes, rx_packets.

Other seldom used fields are kept in netdev->stats structure, possibly
unsafe.

This is a preliminary work to support lockless transmit path, and
correct RX stats, that are already unsafe.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

IPv4 and IPv6 have separate neighbour tables, so
the warning messages should be distinguishable.

[ Add a suitable message prefix on the ipv4 side as well -DaveM ]
Signed-off-by: NUlrich Weber <uweber@astaro.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Clean up a missing exit path in the ipv6 module init routines.  In
addrconf_init we call ipv6_addr_label_init which calls register_pernet_subsys
for the ipv6_addr_label_ops structure.  But if module loading fails, or if the
ipv6 module is removed, there is no corresponding unregister_pernet_subsys call,
which leaves a now-bogus address on the pernet_list, leading to oopses in
subsequent registrations.  This patch cleans up both the failed load path and
the unload path.  Tested by myself with good results.
Signed-off-by: NNeil Horman <nhorman@tuxdriver.com>

 include/net/addrconf.h |    1 +
 net/ipv6/addrconf.c    |   11 ++++++++---
 net/ipv6/addrlabel.c   |    5 +++++
 3 files changed, 14 insertions(+), 3 deletions(-)
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Change "return (EXPR);" to "return EXPR;"

return is not a function, parentheses are not required.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Special care should be taken when slow path is hit in ip_fragment() :

When walking through frags, we transfert truesize ownership from skb to
frags. Then if we hit a slow_path condition, we must undo this or risk
uncharging frags->truesize twice, and in the end, having negative socket
sk_wmem_alloc counter, or even freeing socket sooner than expected.

Many thanks to Nick Bowler, who provided a very clean bug report and
test program.

Thanks to Jarek for reviewing my first patch and providing a V2

While Nick bisection pointed to commit 2b85a34e (net: No more
expensive sock_hold()/sock_put() on each tx), underlying bug is older
(2.6.12-rc5)

A side effect is to extend work done in commit b2722b1c
(ip_fragment: also adjust skb->truesize for packets not owned by a
socket) to ipv6 as well.
Reported-and-bisected-by: NNick Bowler <nbowler@elliptictech.com>
Tested-by: NNick Bowler <nbowler@elliptictech.com>
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
CC: Jarek Poplawski <jarkao2@gmail.com>
CC: Patrick McHardy <kaber@trash.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The family parameter xfrm_state_find is used to find a state matching a
certain policy. This value is set to the template's family
(encap_family) right before xfrm_state_find is called.
The family parameter is however also used to construct a temporary state
in xfrm_state_find itself which is wrong for inter-family scenarios
because it produces a selector for the wrong family. Since this selector
is included in the xfrm_user_acquire structure, user space programs
misinterpret IPv6 addresses as IPv4 and vice versa.
This patch splits up the original init_tempsel function into a part that
initializes the selector respectively the props and id of the temporary
state, to allow for differing ip address families whithin the state.
Signed-off-by: NThomas Egerer <thomas.egerer@secunet.com>
Signed-off-by: NSteffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Under load, netif_rx() can drop incoming packets but administrators dont
have a chance to spot which device needs some tuning (RPS activation for
example)

This patch adds rx_dropped accounting in vlans and tunnels.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

As RTNL is held while doing tunnels inserts and deletes, we can remove
ip6_tnl_lock spinlock. My initial RCU conversion was conservative and
converted the rwlock to spinlock, with no RTNL requirement.

Use appropriate rcu annotations and modern lockdep checks as well.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

As RTNL is held while doing tunnels inserts and deletes, we can remove
ipip6_lock spinlock. My initial RCU conversion was conservative and
converted the rwlock to spinlock, with no RTNL requirement.

Use appropriate rcu annotations and modern lockdep checks as well.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

xfrm4_tunnel_register() & xfrm6_tunnel_register() should
use rcu_assign_pointer() to make sure previous writes
(to handler->next) are committed to memory before chain
insertion.

deregister functions dont need a particular barrier.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

commit 30fff923 introduced in linux-2.6.33 (udp: bind() optimisation)
added a secondary hash on UDP, hashed on (local addr, local port).

Problem is that following sequence :

fd = socket(...)
connect(fd, &remote, ...)

not only selects remote end point (address and port), but also sets
local address, while UDP stack stored in secondary hash table the socket
while its local address was INADDR_ANY (or ipv6 equivalent)

Sequence is :
 - autobind() : choose a random local port, insert socket in hash tables
              [while local address is INADDR_ANY]
 - connect() : set remote address and port, change local address to IP
              given by a route lookup.

When an incoming UDP frame comes, if more than 10 sockets are found in
primary hash table, we switch to secondary table, and fail to find
socket because its local address changed.

One solution to this problem is to rehash datagram socket if needed.

We add a new rehash(struct socket *) method in "struct proto", and
implement this method for UDP v4 & v6, using a common helper.

This rehashing only takes care of secondary hash table, since primary
hash (based on local port only) is not changed.
Reported-by: NKrzysztof Piotr Oledzki <ole@ans.pl>
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Tested-by: NKrzysztof Piotr Oledzki <ole@ans.pl>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Use cmpxchg() to get rid of spinlocks in inet_add_protocol() and
friends.

inet_protos[] & inet6_protos[] are moved to read_mostly section
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

RFC5722 prohibits reassembling IPv6 fragments when some data overlaps.

Bug spotted by Zhang Zuotao <zuotao.zhang@6wind.com>.
Signed-off-by: NNicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

RFC5722 prohibits reassembling fragments when some data overlaps.

Bug spotted by Zhang Zuotao <zuotao.zhang@6wind.com>.
Signed-off-by: NNicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Similar to accepting router advertisement, the IPv6 stack does not send router
solicitations if forwarding is enabled.

This patch enables this behavior to be overruled by setting forwarding to the
special value 2.
Signed-off-by: NThomas Graf <tgraf@infradead.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The current IPv6 behavior is to not accept router advertisements while
forwarding, i.e. configured as router.

This does make sense, a router is typically not supposed to be auto
configured. However there are exceptions and we should allow the
current behavior to be overwritten.

Therefore this patch enables the user to overrule the "if forwarding
enabled then don't listen to RAs" rule by setting accept_ra to the
special value of 2.

An alternative would be to ignore the forwarding switch alltogether
and solely accept RAs based on the value of accept_ra. However, I
found that if not intended, accepting RAs as a router can lead to
strange unwanted behavior therefore we it seems wise to only do so
if the user explicitely asks for this behavior.
Signed-off-by: NThomas Graf <tgraf@infradead.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

tunnel4_handlers, tunnel64_handlers, tunnel6_handlers and
tunnel46_handlers are protected by RCU, but we dont use appropriate rcu
primitives to scan them. rcu_lock() is already held by caller.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

tunnel6_handlers chain being scanned for each incoming packet,
make sure it doesnt share an often dirtied cache line.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

tunnel4_handlers chain being scanned for each incoming packet,
make sure it doesnt share an often dirtied cache line.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

commit f3c5c1bf
(netfilter: xtables: make ip_tables reentrant) forgot to
also compute the jumpstack size in the compat handlers.

Result is that "iptables -I INPUT -j userchain" turns into -j DROP.

Reported by Sebastian Roesner on #netfilter, closes
http://bugzilla.netfilter.org/show_bug.cgi?id=669.

Note: arptables change is compile-tested only.
Signed-off-by: NFlorian Westphal <fw@strlen.de>
Acked-by: NEric Dumazet <eric.dumazet@gmail.com>
Tested-by: NMikael Pettersson <mikpe@it.uu.se>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

SKBs can be "fragmented" in two ways, via a page array (called
skb_shinfo(skb)->frags[]) and via a list of SKBs (called
skb_shinfo(skb)->frag_list).

Since skb_has_frags() tests the latter, it's name is confusing
since it sounds more like it's testing the former.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

After commit 24b36f01 (netfilter: {ip,ip6,arp}_tables: dont block
bottom half more than necessary), lockdep can raise a warning
because we attempt to lock a spinlock with BH enabled, while
the same lock is usually locked by another cpu in a softirq context.

Disable again BH to avoid these lockdep warnings.
Reported-by: NLinus Torvalds <torvalds@linux-foundation.org>
Diagnosed-by: NDavid S. Miller <davem@davemloft.net>
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

sysctl output ipv6 gc_elasticity and min_adv_mss as values divided by
HZ. However, they are not in unit of jiffies, since ip6_rt_min_advmss
refers to packet size and ip6_rt_fc_elasticity is used as scaler as in
expire>>ip6_rt_gc_elasticity, so replace the jiffies conversion
handler will regular handler for them.

This has impact on scripts that are currently working assuming the
divide by HZ, will yield different results with this patch in place.
Signed-off-by: NMin Zhang <mzhang@mvista.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We currently disable BH for the whole duration of get_counters()

On machines with a lot of cpus and large tables, this might be too long.

We can disable preemption during the whole function, and disable BH only
while fetching counters for the current cpu.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NPatrick McHardy <kaber@trash.net>

ipv6_hdr(skb)->payload_len is ZERO and can't be used for accounting, if
the payload is a Jumbo Payload specified in RFC2675.
Signed-off-by: NChangli Gao <xiaosuo@gmail.com>
Signed-off-by: NPatrick McHardy <kaber@trash.net>

If the interface has IPv6 disabled, don't add a multicast or
link-local route since we won't be adding a link-local address.
Reported-by: NMahesh Kelkar <maheshkelkar@gmail.com>
Signed-off-by: NBrian Haley <brian.haley@hp.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The input handler for Type 2 Routing Header (mip6_rthdr_input())
checks if the CoA in the packet matches the CoA in the XFRM state.

Current check is buggy: it compares the adddress in the Type 2
Routing Header, i.e. the HoA, against the expected CoA in the state.
The comparison should be made against the address in the destination
field of the IPv6 header.

The bug remained unnoticed because the main (and possibly only current)
user of the code (UMIP MIPv6 Daemon) initializes the XFRM state with the
unspecified address, i.e. explicitly allows everything.

Yoshifuji-san, can you ack that one?
Signed-off-by: NArnaud Ebalard <arno@natisbad.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

a new boolean flag no_autobind is added to structure proto to avoid the autobind
calls when the protocol is TCP. Then sock_rps_record_flow() is called int the
TCP's sendmsg() and sendpage() pathes.
Signed-off-by: NChangli Gao <xiaosuo@gmail.com>
----
 include/net/inet_common.h |    4 ++++
 include/net/sock.h        |    1 +
 include/net/tcp.h         |    8 ++++----
 net/ipv4/af_inet.c        |   15 +++++++++------
 net/ipv4/tcp.c            |   11 +++++------
 net/ipv4/tcp_ipv4.c       |    3 +++
 net/ipv6/af_inet6.c       |    8 ++++----
 net/ipv6/tcp_ipv6.c       |    3 +++
 8 files changed, 33 insertions(+), 20 deletions(-)
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: NJiri Kosina <jkosina@suse.cz>

As the fragments are sent in order in most of OSes, such as Windows, Darwin and
FreeBSD, it is likely the new fragments are at the end of the inet_frag_queue.
In the fast path, we check if the skb at the end of the inet_frag_queue is the
prev we expect.
Signed-off-by: NChangli Gao <xiaosuo@gmail.com>
Signed-off-by: NPatrick McHardy <kaber@trash.net>

While using xfrm by MARK feature in
2.6.34 - 2.6.35 kernels, the mark
is always cleared in flowi structure via memset in
_decode_session4 (net/ipv4/xfrm4_policy.c), so
the policy lookup fails.
IPv6 code is affected by this bug too.
Signed-off-by: NPeter Kosyh <p.kosyh@gmail.com>
Acked-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>