Commit 35f9c09f (tcp: tcp_sendpages() should call tcp_push() once)
added an internal flag MSG_SENDPAGE_NOTLAST, similar to
MSG_MORE.

algif_hash, algif_skcipher, and udp used MSG_MORE from tcp_sendpages()
and need to see the new flag as identical to MSG_MORE.

This fixes sendfile() on AF_ALG.

v3: also fix udp

Cc: Tom Herbert <therbert@google.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: <stable@vger.kernel.org> # 3.4.x + 3.2.x
Reported-and-tested-by: NShawn Landden <shawnlandden@gmail.com>
Original-patch: Richard Weinberger <richard@nod.at>
Signed-off-by: NShawn Landden <shawn@churchofgit.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

since f9242b6b
inet: Sanitize inet{,6} protocol demux.

there are not pretended hash tables for ipv4 or
ipv6 protocol handler.
Signed-off-by: NBaker Zhang <Baker.kernel@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit bceaa902 ("inet: prevent leakage
of uninitialized memory to user in recv syscalls") conditionally updated
addr_len if the msg_name is written to. The recv_error and rxpmtu
functions relied on the recvmsg functions to set up addr_len before.

As this does not happen any more we have to pass addr_len to those
functions as well and set it to the size of the corresponding sockaddr
length.

This broke traceroute and such.

Fixes: bceaa902 ("inet: prevent leakage of uninitialized memory to user in recv syscalls")
Reported-by: NBrad Spengler <spender@grsecurity.net>
Reported-by: Tom Labanowski
Cc: mpb <mpb.mail@gmail.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

nobody needs it. remove.
Signed-off-by: NGao feng <gaofeng@cn.fujitsu.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch simplifies the checksum verification in tcpX_gro_receive
by reusing the CHECKSUM_COMPLETE code for CHECKSUM_NONE.  All it
does for CHECKSUM_NONE is compute the partial checksum and then
treat it as if it came from the hardware (CHECKSUM_COMPLETE).
Signed-off-by: NHerbert Xu <herbert@gondor.apana.org.au>

Cheers,
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In some cases we may receive IP packets that are longer than
their stated lengths.  Such packets are never merged in GRO.
However, we may end up computing their checksums incorrectly
and end up allowing packets with a bogus checksum enter our
stack with the checksum status set as verified.

Since such packets are rare and not performance-critical, this
patch simply skips the checksum verification for them.
Reported-by: NAlexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: NHerbert Xu <herbert@gondor.apana.org.au>
Acked-by: NAlexander Duyck <alexander.h.duyck@intel.com>

Thanks,
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

CPUs can ask for local route via ip_route_input_noref() concurrently.
if nh_rth_input is not cached yet, CPUs will proceed to allocate
equivalent DSTs on 'lo' and then will try to cache them in nh_rth_input
via rt_cache_route()
Most of the time they succeed, but on occasion the following two lines:
	orig = *p;
	prev = cmpxchg(p, orig, rt);
in rt_cache_route() do race and one of the cpus fails to complete cmpxchg.
But ip_route_input_slow() doesn't check the return code of rt_cache_route(),
so dst is leaking. dst_destroy() is never called and 'lo' device
refcnt doesn't go to zero, which can be seen in the logs as:
	unregister_netdevice: waiting for lo to become free. Usage count = 1
Adding mdelay() between above two lines makes it easily reproducible.
Fix it similar to nh_pcpu_rth_output case.

Fixes: d2d68ba9 ("ipv4: Cache input routes in fib_info nexthops.")
Signed-off-by: NAlexei Starovoitov <ast@plumgrid.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

As suggested by David Miller, make genl_register_family_with_ops()
a macro and pass only the array, evaluating ARRAY_SIZE() in the
macro, this is a little safer.

The openvswitch has some indirection, assing ops/n_ops directly in
that code. This might ultimately just assign the pointers in the
family initializations, saving the struct genl_family_and_ops and
code (once mcast groups are handled differently.)
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

snd_nxt must be updated synchronously with sk_send_head.  Otherwise
tp->packets_out may be updated incorrectly, what may bring a kernel panic.

Here is a kernel panic from my host.
[  103.043194] BUG: unable to handle kernel NULL pointer dereference at 0000000000000048
[  103.044025] IP: [<ffffffff815aaaaf>] tcp_rearm_rto+0xcf/0x150
...
[  146.301158] Call Trace:
[  146.301158]  [<ffffffff815ab7f0>] tcp_ack+0xcc0/0x12c0

Before this panic a tcp socket was restored. This socket had sent and
unsent data in the write queue. Sent data was restored in repair mode,
then the socket was switched from reapair mode and unsent data was
restored. After that the socket was switched back into repair mode.

In that moment we had a socket where write queue looks like this:
snd_una    snd_nxt   write_seq
   |_________|________|
             |
	  sk_send_head

After a second switching from repair mode the state of socket was
changed:

snd_una          snd_nxt, write_seq
   |_________ ________|
             |
	  sk_send_head

This state is inconsistent, because snd_nxt and sk_send_head are not
synchronized.

Bellow you can find a call trace, how packets_out can be incremented
twice for one skb, if snd_nxt and sk_send_head are not synchronized.
In this case packets_out will be always positive, even when
sk_write_queue is empty.

tcp_write_wakeup
	skb = tcp_send_head(sk);
	tcp_fragment
		if (!before(tp->snd_nxt, TCP_SKB_CB(buff)->end_seq))
			tcp_adjust_pcount(sk, skb, diff);
	tcp_event_new_data_sent
		tp->packets_out += tcp_skb_pcount(skb);

I think update of snd_nxt isn't required, when a socket is switched from
repair mode.  Because it's initialized in tcp_connect_init. Then when a
write queue is restored, snd_nxt is incremented in tcp_event_new_data_sent,
so it's always is in consistent state.

I have checked, that the bug is not reproduced with this patch and
all tests about restoring tcp connections work fine.

Cc: Pavel Emelyanov <xemul@parallels.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Cc: James Morris <jmorris@namei.org>
Cc: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>
Cc: Patrick McHardy <kaber@trash.net>
Signed-off-by: NAndrey Vagin <avagin@openvz.org>
Acked-by: NPavel Emelyanov <xemul@parallels.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

After searching rt by the vti tunnel dst/src parameter,
if this rt has neither attached to any transformation
nor the transformation is not tunnel oriented, this rt
should be released back to ip layer.

otherwise causing dst memory leakage.
Signed-off-by: NFan Du <fan.du@windriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

A plain read() on a socket does set msg->msg_name to NULL. So check for
NULL pointer first.
Signed-off-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Only update *addr_len when we actually fill in sockaddr, otherwise we
can return uninitialized memory from the stack to the caller in the
recvfrom, recvmmsg and recvmsg syscalls. Drop the the (addr_len == NULL)
checks because we only get called with a valid addr_len pointer either
from sock_common_recvmsg or inet_recvmsg.

If a blocking read waits on a socket which is concurrently shut down we
now return zero and set msg_msgnamelen to 0.
Reported-by: Nmpb <mpb.mail@gmail.com>
Suggested-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When the synproxy_parse_options is called on the client ack the mss
option will not be present. Consequently mss wont be included in the
backend syn packet, which falls back to 536 bytes mss.

Therefore XT_SYNPROXY_OPT_MSS is explicitly flagged when recovering mss
value from cookie.
Signed-off-by: NMartin Topholm <mph@one.com>
Reviewed-by: NJesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

All seq_printf() users are using "%n" for calculating padding size,
convert them to use seq_setwidth() / seq_pad() pair.
Signed-off-by: NTetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: NKees Cook <keescook@chromium.org>
Cc: Joe Perches <joe@perches.com>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

ip4_datagram_connect() being called from process context,
it should use IP_INC_STATS() instead of IP_INC_STATS_BH()
otherwise we can deadlock on 32bit arches, or get corruptions of
SNMP counters.

Fixes: 584bdf8c ("[IPV4]: Fix "ipOutNoRoutes" counter error for TCP and UDP")
Signed-off-by: NEric Dumazet <edumazet@google.com>
Reported-by: NDave Jones <davej@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Now that genl_ops are no longer modified in place when
registering, they can be made const. This patch was done
mostly with spatch:

@@
identifier ops;
@@
+const
 struct genl_ops ops[] = {
 ...
 };

(except the struct thing in net/openvswitch/datapath.c)
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We had some reports of crashes using TCP fastopen, and Dave Jones
gave a nice stack trace pointing to the error.

Issue is that tcp_get_metrics() should not be called with a NULL dst

Fixes: 1fe4c481 ("net-tcp: Fast Open client - cookie cache")
Signed-off-by: NEric Dumazet <edumazet@google.com>
Reported-by: NDave Jones <davej@redhat.com>
Cc: Yuchung Cheng <ycheng@google.com>
Acked-by: NYuchung Cheng <ycheng@google.com>
Tested-by: NDave Jones <davej@fedoraproject.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

After commit c9eeec26 ("tcp: TSQ can use a dynamic limit"), several
users reported throughput regressions, notably on mvneta and wifi
adapters.

802.11 AMPDU requires a fair amount of queueing to be effective.

This patch partially reverts the change done in tcp_write_xmit()
so that the minimal amount is sysctl_tcp_limit_output_bytes.

It also remove the use of this sysctl while building skb stored
in write queue, as TSO autosizing does the right thing anyway.

Users with well behaving NICS and correct qdisc (like sch_fq),
can then lower the default sysctl_tcp_limit_output_bytes value from
128KB to 8KB.

This new usage of sysctl_tcp_limit_output_bytes permits each driver
authors to check how their driver performs when/if the value is set
to a minimum of 4KB.

Normally, line rate for a single TCP flow should be possible,
but some drivers rely on timers to perform TX completion and
too long TX completion delays prevent reaching full throughput.

Fixes: c9eeec26 ("tcp: TSQ can use a dynamic limit")
Signed-off-by: NEric Dumazet <edumazet@google.com>
Reported-by: NSujith Manoharan <sujith@msujith.org>
Reported-by: NArnaud Ebalard <arno@natisbad.org>
Tested-by: NSujith Manoharan <sujith@msujith.org>
Cc: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

commit 06a23fe3
("core/dev: set pkt_type after eth_type_trans() in dev_forward_skb()")
and refactoring 64261f23
("dev: move skb_scrub_packet() after eth_type_trans()")

are forcing pkt_type to be PACKET_HOST when skb traverses veth.

which means that ip forwarding will kick in inside netns
even if skb->eth->h_dest != dev->dev_addr

Fix order of eth_type_trans() and skb_scrub_packet() in dev_forward_skb()
and in ip_tunnel_rcv()

Fixes: 06a23fe3 ("core/dev: set pkt_type after eth_type_trans() in dev_forward_skb()")
CC: Isaku Yamahata <yamahatanetdev@gmail.com>
CC: Maciej Zenczykowski <zenczykowski@gmail.com>
CC: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: NAlexei Starovoitov <ast@plumgrid.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

While testing virtio_net and skb_segment() changes, Hannes reported
that UFO was sending wrong frames.

It appears this was introduced by a recent commit :
8c3a897b ("inet: restore gso for vxlan")

The old condition to perform IP frag was :

tunnel = !!skb->encapsulation;
...
        if (!tunnel && proto == IPPROTO_UDP) {

So the new one should be :

udpfrag = !skb->encapsulation && proto == IPPROTO_UDP;
...
        if (udpfrag) {

Initialization of udpfrag must be done before call
to ops->callbacks.gso_segment(skb, features), as
skb_udp_tunnel_segment() clears skb->encapsulation

(We want udpfrag to be true for UFO, false for VXLAN)

With help from Alexei Starovoitov
Reported-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NEric Dumazet <edumazet@google.com>
Cc: Alexei Starovoitov <ast@plumgrid.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In order to enable lockdep on seqcount/seqlock structures, we
must explicitly initialize any locks.

The u64_stats_sync structure, uses a seqcount, and thus we need
to introduce a u64_stats_init() function and use it to initialize
the structure.

This unfortunately adds a lot of fairly trivial initialization code
to a number of drivers. But the benefit of ensuring correctness makes
this worth while.

Because these changes are required for lockdep to be enabled, and the
changes are quite trivial, I've not yet split this patch out into 30-some
separate patches, as I figured it would be better to get the various
maintainers thoughts on how to best merge this change along with
the seqcount lockdep enablement.

Feedback would be appreciated!
Signed-off-by: NJohn Stultz <john.stultz@linaro.org>
Acked-by: NJulian Anastasov <ja@ssi.bg>
Signed-off-by: NPeter Zijlstra <peterz@infradead.org>
Cc: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>
Cc: James Morris <jmorris@namei.org>
Cc: Jesse Gross <jesse@nicira.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Mirko Lindner <mlindner@marvell.com>
Cc: Patrick McHardy <kaber@trash.net>
Cc: Roger Luethi <rl@hellgate.ch>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Simon Horman <horms@verge.net.au>
Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Wensong Zhang <wensong@linux-vs.org>
Cc: netdev@vger.kernel.org
Link: http://lkml.kernel.org/r/1381186321-4906-2-git-send-email-john.stultz@linaro.orgSigned-off-by: NIngo Molnar <mingo@kernel.org>

Sockets marked with IP_PMTUDISC_INTERFACE won't do path mtu discovery,
their sockets won't accept and install new path mtu information and they
will always use the interface mtu for outgoing packets. It is guaranteed
that the packet is not fragmented locally. But we won't set the DF-Flag
on the outgoing frames.

Florian Weimer had the idea to use this flag to ensure DNS servers are
never generating outgoing fragments. They may well be fragmented on the
path, but the server never stores or usees path mtu values, which could
well be forged in an attack.

(The root of the problem with path MTU discovery is that there is
no reliable way to authenticate ICMP Fragmentation Needed But DF Set
messages because they are sent from intermediate routers with their
source addresses, and the IMCP payload will not always contain sufficient
information to identify a flow.)

Recent research in the DNS community showed that it is possible to
implement an attack where DNS cache poisoning is feasible by spoofing
fragments. This work was done by Amir Herzberg and Haya Shulman:
<https://sites.google.com/site/hayashulman/files/fragmentation-poisoning.pdf>

This issue was previously discussed among the DNS community, e.g.
<http://www.ietf.org/mail-archive/web/dnsext/current/msg01204.html>,
without leading to fixes.

This patch depends on the patch "ipv4: fix DO and PROBE pmtu mode
regarding local fragmentation with UFO/CORK" for the enforcement of the
non-fragmentable checks. If other users than ip_append_page/data should
use this semantic too, we have to add a new flag to IPCB(skb)->flags to
suppress local fragmentation and check for this in ip_finish_output.

Many thanks to Florian Weimer for the idea and feedback while implementing
this patch.

Cc: David S. Miller <davem@davemloft.net>
Suggested-by: NFlorian Weimer <fweimer@redhat.com>
Signed-off-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Slow start now increases cwnd by 1 if an ACK acknowledges some packets,
regardless the number of packets. Consequently slow start performance
is highly dependent on the degree of the stretch ACKs caused by
receiver or network ACK compression mechanisms (e.g., delayed-ACK,
GRO, etc).  But slow start algorithm is to send twice the amount of
packets of packets left so it should process a stretch ACK of degree
N as if N ACKs of degree 1, then exits when cwnd exceeds ssthresh. A
follow up patch will use the remainder of the N (if greater than 1)
to adjust cwnd in the congestion avoidance phase.

In addition this patch retires the experimental limited slow start
(LSS) feature. LSS has multiple drawbacks but questionable benefit. The
fractional cwnd increase in LSS requires a loop in slow start even
though it's rarely used. Configuring such an increase step via a global
sysctl on different BDPS seems hard. Finally and most importantly the
slow start overshoot concern is now better covered by the Hybrid slow
start (hystart) enabled by default.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Applications have started to use Fast Open (e.g., Chrome browser has
such an optional flag) and the feature has gone through several
generations of kernels since 3.7 with many real network tests. It's
time to enable this flag by default for applications to test more
conveniently and extensively.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Remove duplicated include.
Signed-off-by: NWei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

On some codepaths the skb does not have a dst entry
when xfrm_decode_session() is called. So check for
a valid skb_dst() before dereferencing the device
interface index. We use 0 as the device index if
there is no valid skb_dst(), or at reverse decoding
we use skb_iif as device interface index.

Bug was introduced with git commit bafd4bd4
("xfrm: Decode sessions with output interface.").
Reported-by: NMeelis Roos <mroos@linux.ee>
Tested-by: NMeelis Roos <mroos@linux.ee>
Signed-off-by: NSteffen Klassert <steffen.klassert@secunet.com>

Fast Open currently has a fall back feature to address SYN-data being
dropped but it requires the middle-box to pass on regular SYN retry
after SYN-data. This is implemented in commit aab48743 ("net-tcp:
Fast Open client - detecting SYN-data drops")

However some NAT boxes will drop all subsequent packets after first
SYN-data and blackholes the entire connections.  An example is in
commit 356d7d88 "netfilter: nf_conntrack: fix tcp_in_window for Fast
Open".

The sender should note such incidents and fall back to use the regular
TCP handshake on subsequent attempts temporarily as well: after the
second SYN timeouts the original Fast Open SYN is most likely lost.
When such an event recurs Fast Open is disabled based on the number of
recurrences exponentially.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

struct esp_data consists of a single pointer, vanishing the need for it
to be a structure. Fold the pointer into 'data' direcly, removing one
level of pointer indirection.
Signed-off-by: NMathias Krause <mathias.krause@secunet.com>
Cc: Steffen Klassert <steffen.klassert@secunet.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: NSteffen Klassert <steffen.klassert@secunet.com>

The padlen member of struct esp_data is always zero. Get rid of it.
Signed-off-by: NMathias Krause <mathias.krause@secunet.com>
Cc: Steffen Klassert <steffen.klassert@secunet.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: NSteffen Klassert <steffen.klassert@secunet.com>

UFO as well as UDP_CORK do not respect IP_PMTUDISC_DO and
IP_PMTUDISC_PROBE well enough.

UFO enabled packet delivery just appends all frags to the cork and hands
it over to the network card. So we just deliver non-DF udp fragments
(DF-flag may get overwritten by hardware or virtual UFO enabled
interface).

UDP_CORK does enqueue the data until the cork is disengaged. At this
point it sets the correct IP_DF and local_df flags and hands it over to
ip_fragment which in this case will generate an icmp error which gets
appended to the error socket queue. This is not reflected in the syscall
error (of course, if UFO is enabled this also won't happen).

Improve this by checking the pmtudisc flags before appending data to the
socket and if we still can fit all data in one packet when IP_PMTUDISC_DO
or IP_PMTUDISC_PROBE is set, only then proceed.

We use (mtu-fragheaderlen) to check for the maximum length because we
ensure not to generate a fragment and non-fragmented data does not need
to have its length aligned on 64 bit boundaries. Also the passed in
ip_options are already aligned correctly.

Maybe, we can relax some other checks around ip_fragment. This needs
more research.
Signed-off-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

commit 6ff50cd5 ("tcp: gso: do not generate out of order packets")
had an heuristic that can trigger a warning in skb_try_coalesce(),
because skb->truesize of the gso segments were exactly set to mss.

This breaks the requirement that

skb->truesize >= skb->len + truesizeof(struct sk_buff);

It can trivially be reproduced by :

ifconfig lo mtu 1500
ethtool -K lo tso off
netperf

As the skbs are looped into the TCP networking stack, skb_try_coalesce()
warns us of these skb under-estimating their truesize.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Reported-by: NAlexei Starovoitov <ast@plumgrid.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

With the removal of the routing cache, we lost the
option to tweak the garbage collector threshold
along with the maximum routing cache size. So git
commit 703fb94e ("xfrm: Fix the gc threshold value
for ipv4") moved back to a static threshold.

It turned out that the current threshold before we
start garbage collecting is much to small for some
workloads, so increase it from 1024 to 32768. This
means that we start the garbage collector if we have
more than 32768 dst entries in the system and refuse
new allocations if we are above 65536.
Reported-by: NWolfgang Walter <linux@stwm.de>
Signed-off-by: NSteffen Klassert <steffen.klassert@secunet.com>

Alexei reported a performance regression on vxlan, caused
by commit 3347c960 "ipv4: gso: make inet_gso_segment() stackable"

GSO vxlan packets were not properly segmented, adding IP fragments
while they were not expected.

Rename 'bool tunnel' to 'bool encap', and add a new boolean
to express the fact that UDP should be fragmented.
This fragmentation is triggered by skb->encapsulation being set.

Remove a "skb->encapsulation = 1" added in above commit,
as its not needed, as frags inherit skb->frag from original
GSO skb.
Reported-by: NAlexei Starovoitov <ast@plumgrid.com>
Signed-off-by: NEric Dumazet <edumazet@google.com>
Tested-by: NAlexei Starovoitov <ast@plumgrid.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Patch ed08495c "tcp: use RTT from SACK for RTO" always re-arms RTO upon
obtaining a RTT sample from newly sacked data.

But technically RTO should only be re-armed when the data sent before
the last (re)transmission of write queue head are (s)acked. Otherwise
the RTO may continue to extend during loss recovery on data sent
in the future.

Note that RTTs from ACK or timestamps do not have this problem, as the RTT
source must be from data sent before.

The new RTO re-arm policy is
1) Always re-arm RTO if SND.UNA is advanced
2) Re-arm RTO if sack RTT is available, provided the sacked data was
   sent before the last time write_queue_head was sent.
Signed-off-by: NLarry Brakmo <brakmo@google.com>
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Acked-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Patch ed08495c "tcp: use RTT from SACK for RTO" has a bug that
it does not check if the ACK acknowledge new data before taking
the RTT sample from TCP timestamps. This patch adds the check
back as required by the RFC.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Acked-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

tp->lsndtime may not always be the SYNACK timestamp if a passive
Fast Open socket sends data before handshake completes. And if the
remote acknowledges both the data and the SYNACK, the RTT sample
is already taken in tcp_ack(), so no need to call
tcp_update_ack_rtt() in tcp_synack_rtt_meas() aagain.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Acked-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

All fragmentation hash secrets now get initialized by their
corresponding hash function with net_get_random_once. Thus we can
eliminate the initial seeding.

Also provide a comment that hash secret seeding happens at the first
call to the corresponding hashing function.

Cc: David S. Miller <davem@davemloft.net>
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Defer the generation of the first hash secret for the ipv4 fragmentation
cache as late as possible.

ip4_frags.rnd gets initial seeded by inet_frags_init and regulary
reseeded by inet_frag_secret_rebuild. Either we call ipqhashfn directly
from ip_fragment.c in which case we initialize the secret directly.

If we first get called by inet_frag_secret_rebuild we install a new secret
by a manual call to get_random_bytes. This secret will be overwritten
as soon as the first call to ipqhashfn happens. This is safe because we
won't race while publishing the new secrets with anyone else.

Cc: Eric Dumazet <edumazet@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

During kernel stability testing on an SMP ARMv7 system, Yalin Wang
reported the following panic from the netfilter code:

  1fe0: 0000001c 5e2d3b10 4007e779 4009e110 60000010 00000032 ff565656 ff545454
  [<c06c48dc>] (ipt_do_table+0x448/0x584) from [<c0655ef0>] (nf_iterate+0x48/0x7c)
  [<c0655ef0>] (nf_iterate+0x48/0x7c) from [<c0655f7c>] (nf_hook_slow+0x58/0x104)
  [<c0655f7c>] (nf_hook_slow+0x58/0x104) from [<c0683bbc>] (ip_local_deliver+0x88/0xa8)
  [<c0683bbc>] (ip_local_deliver+0x88/0xa8) from [<c0683718>] (ip_rcv_finish+0x418/0x43c)
  [<c0683718>] (ip_rcv_finish+0x418/0x43c) from [<c062b1c4>] (__netif_receive_skb+0x4cc/0x598)
  [<c062b1c4>] (__netif_receive_skb+0x4cc/0x598) from [<c062b314>] (process_backlog+0x84/0x158)
  [<c062b314>] (process_backlog+0x84/0x158) from [<c062de84>] (net_rx_action+0x70/0x1dc)
  [<c062de84>] (net_rx_action+0x70/0x1dc) from [<c0088230>] (__do_softirq+0x11c/0x27c)
  [<c0088230>] (__do_softirq+0x11c/0x27c) from [<c008857c>] (do_softirq+0x44/0x50)
  [<c008857c>] (do_softirq+0x44/0x50) from [<c0088614>] (local_bh_enable_ip+0x8c/0xd0)
  [<c0088614>] (local_bh_enable_ip+0x8c/0xd0) from [<c06b0330>] (inet_stream_connect+0x164/0x298)
  [<c06b0330>] (inet_stream_connect+0x164/0x298) from [<c061d68c>] (sys_connect+0x88/0xc8)
  [<c061d68c>] (sys_connect+0x88/0xc8) from [<c000e340>] (ret_fast_syscall+0x0/0x30)
  Code: 2a000021 e59d2028 e59de01c e59f011c (e7824103)
  ---[ end trace da227214a82491bd ]---
  Kernel panic - not syncing: Fatal exception in interrupt

This comes about because CPU1 is executing xt_replace_table in response
to a setsockopt syscall, resulting in:

	ret = xt_jumpstack_alloc(newinfo);
		--> newinfo->jumpstack = kzalloc(size, GFP_KERNEL);

	[...]

	table->private = newinfo;
	newinfo->initial_entries = private->initial_entries;

Meanwhile, CPU0 is handling the network receive path and ends up in
ipt_do_table, resulting in:

	private = table->private;

	[...]

	jumpstack  = (struct ipt_entry **)private->jumpstack[cpu];

On weakly ordered memory architectures, the writes to table->private
and newinfo->jumpstack from CPU1 can be observed out of order by CPU0.
Furthermore, on architectures which don't respect ordering of address
dependencies (i.e. Alpha), the reads from CPU0 can also be re-ordered.

This patch adds an smp_wmb() before the assignment to table->private
(which is essentially publishing newinfo) to ensure that all writes to
newinfo will be observed before plugging it into the table structure.
A dependent-read barrier is also added on the consumer sides, to ensure
the same ordering requirements are also respected there.

Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reported-by: NWang, Yalin <Yalin.Wang@sonymobile.com>
Tested-by: NWang, Yalin <Yalin.Wang@sonymobile.com>
Signed-off-by: NWill Deacon <will.deacon@arm.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

For passive TCP connections, upon receiving the ACK that completes the
3WHS, make sure we set our pacing rate after we get our first RTT
sample.

On passive TCP connections, when we receive the ACK completing the
3WHS we do not take an RTT sample in tcp_ack(), but rather in
tcp_synack_rtt_meas(). So upon receiving the ACK that completes the
3WHS, tcp_ack() leaves sk_pacing_rate at its initial value.

Originally the initial sk_pacing_rate value was 0, so passive-side
connections defaulted to sysctl_tcp_min_tso_segs (2 segs) in skbuffs
made in the first RTT. With a default initial cwnd of 10 packets, this
happened to be correct for RTTs 5ms or bigger, so it was hard to
see problems in WAN or emulated WAN testing.

Since 7eec4174 ("pkt_sched: fq: fix non TCP flows pacing"), the
initial sk_pacing_rate is 0xffffffff. So after that change, passive
TCP connections were keeping this value (and using large numbers of
segments per skbuff) until receiving an ACK for data.
Signed-off-by: NNeal Cardwell <ncardwell@google.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>