Using a timer wheel for timewait sockets was nice ~15 years ago when
memory was expensive and machines had a single processor.

This does not scale, code is ugly and source of huge latencies
(Typically 30 ms have been seen, cpus spinning on death_lock spinlock.)

We can afford to use an extra 64 bytes per timewait sock and spread
timewait load to all cpus to have better behavior.

Tested:

On following test, /proc/sys/net/ipv4/tcp_tw_recycle is set to 1
on the target (lpaa24)

Before patch :

lpaa23:~# ./super_netperf 200 -H lpaa24 -t TCP_CC -l 60 -- -p0,0
419594

lpaa23:~# ./super_netperf 200 -H lpaa24 -t TCP_CC -l 60 -- -p0,0
437171

While test is running, we can observe 25 or even 33 ms latencies.

lpaa24:~# ping -c 1000 -i 0.02 -qn lpaa23
...
1000 packets transmitted, 1000 received, 0% packet loss, time 20601ms
rtt min/avg/max/mdev = 0.020/0.217/25.771/1.535 ms, pipe 2

lpaa24:~# ping -c 1000 -i 0.02 -qn lpaa23
...
1000 packets transmitted, 1000 received, 0% packet loss, time 20702ms
rtt min/avg/max/mdev = 0.019/0.183/33.761/1.441 ms, pipe 2

After patch :

About 90% increase of throughput :

lpaa23:~# ./super_netperf 200 -H lpaa24 -t TCP_CC -l 60 -- -p0,0
810442

lpaa23:~# ./super_netperf 200 -H lpaa24 -t TCP_CC -l 60 -- -p0,0
800992

And latencies are kept to minimal values during this load, even
if network utilization is 90% higher :

lpaa24:~# ping -c 1000 -i 0.02 -qn lpaa23
...
1000 packets transmitted, 1000 received, 0% packet loss, time 19991ms
rtt min/avg/max/mdev = 0.023/0.064/0.360/0.042 ms
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The printed values are all of type unsigned integer, therefore use
%u instead of %d. Otherwise an user can face negative values.
Signed-off-by: Richard Weinberger <richard@nod.at>
Acked-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

The printed values are all of type unsigned integer, therefore use
%u instead of %d. Otherwise an user can face negative values.

Fixes:
$ cat /proc/net/netfilter/nfnetlink_queue
    0  29508   278 2 65531     0 2004213241 -2129885586  1
    1 -27747     0 2 65531     0     0        0  1
    2 -27748     0 2 65531     0     0        0  1
Signed-off-by: Richard Weinberger <richard@nod.at>
Acked-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

The netlink portid is an unsigned integer, use this type
also in netfilter.
Signed-off-by: Richard Weinberger <richard@nod.at>
Acked-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

portid is an unsigned integer. Fix urelease_work to
match all other portid user in the kernel.
Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: David S. Miller <davem@davemloft.net>

portid is an unsigned integer. Fix netlink_notify to
match all other portid user in the kernel.
Signed-off-by: Richard Weinberger <richard@nod.at>
Signed-off-by: David S. Miller <davem@davemloft.net>

Since retransmitted segments are not used for RTT estimation, previously
SACKed segments present in the rtx queue are used. This estimation can be
several times larger than the actual RTT. When a cumulative ack covers both
previously SACKed and retransmitted segments, CC may thus get a bogus RTT.

Such segments previously had an RTT estimation in tcp_sacktag_one(), so it
seems reasonable to not reuse them in tcp_clean_rtx_queue() at all.

Afaik, this has had no effect on SRTT/RTO because of Karn's check.
Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no>
Acked-by: Neal Cardwell <ncardwell@google.com>
Tested-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Even if we make use of classifier and actions from the egress
path, we're going into handle_ing() executing additional code
on a per-packet cost for ingress qdisc, just to realize that
nothing is attached on ingress.

Instead, this can just be blinded out as a no-op entirely with
the use of a static key. On input fast-path, we already make
use of static keys in various places, e.g. skb time stamping,
in RPS, etc. It makes sense to not waste time when we're assured
that no ingress qdisc is attached anywhere.

Enabling/disabling of that code path is being done via two
helpers, namely net_{inc,dec}_ingress_queue(), that are being
invoked under RTNL mutex when a ingress qdisc is being either
initialized or destructed.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@plumgrid.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Thomas Graf says:

====================
Bring sizeof(net_device) down to < 2K bytes

The size of struct net_device crossed the 2K boundary a while ago which
is a waste in combination with many net namespaces. This series brings
the size of struct net_device down to well below 2K in total size with
a typical configuration. Some reserves a several holes leave room for
further expansion.

Before:
/* size: 2176, cachelines: 34, members: 121 */

After:
/* size: 1984, cachelines: 31, members: 120 */
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Some trivial reorders while preserving the RX/TX cache lines
split to fill a couple of holes.
Signed-off-by: Thomas Graf <tgraf@suug.ch>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

e1000e is the only driver requiring pm_qos_req, instead of causing
every device to waste up to 240 bytes. Allocate it for the specific
driver.
Signed-off-by: Thomas Graf <tgraf@suug.ch>
Acked-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When a new rtnl or xfrm command is added, this part of the code is frequently
missing. Let's help the developer with a build time test.
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Jeff Kirsher says:

====================
Intel Wired LAN Driver Updates 2015-04-11

This series contains updates to iflink, ixgbe and ixgbevf.

The entire set of changes come from Vlad Zolotarov to ultimately add
the ethtool ops to VF driver to allow querying the RSS indirection table
and RSS random key.

Currently we support only 82599 and x540 devices.  On those devices, VFs
share the RSS redirection table and hash key with a PF.  Letting the VF
query this information may introduce some security risks, therefore this
feature will be disabled by default.

The new netdev op allows a system administrator to change the default
behaviour with "ip link set" command.  The relevant iproute2 patch has
already been sent and awaits for this series upstream.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Cong Wang says:

====================
fou: some fixes and updates

Patch 1~3 fix some minor bugs in net/ipv4/fou.c, the only
thing I am not sure is if it's too late to change the
byte order of FOU_ATTR_PORT, if so we have to fix iproute2
instead of kernel.

Patch 4~5 add some new features to make it complete.

v2: make fou->port be16 too
====================
Acked-by: Tom Herbert <tom@herbertland.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Cc: Tom Herbert <tom@herbertland.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Also convert the spinlock to a mutex.

Cc: Tom Herbert <tom@herbertland.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

udp_config.local_udp_port is be16. And iproute2 passes
network order for FOU_ATTR_PORT.

This doesn't fix any bug, just for consistency.

Cc: Tom Herbert <tom@herbertland.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Not a big deal, just for corretness.

Cc: Tom Herbert <tom@herbertland.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This fixes the following harmless warning:

./ip/ip fou del port 7777
[  122.907516] udp_del_offload: didn't find offload for port 7777

Cc: Tom Herbert <tom@herbertland.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Nicolas Dichtel says:

====================
selinux: add missing xfrm nl cmd

With this series, xfrm commands are fully synchronized.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

This command is missing.

Fixes: 3a2dfbe8 ("xfrm: Notify changes in UDP encapsulation via netlink")
CC: Martin Willi <martin@strongswan.org>
Reported-by: Stephen Smalley <sds@tycho.nsa.gov>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This command is missing.

Fixes: 5c79de6e ("[XFRM]: User interface for handling XFRM_MSG_MIGRATE")
Reported-by: Stephen Smalley <sds@tycho.nsa.gov>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This command is missing.

Fixes: 97a64b45 ("[XFRM]: Introduce XFRM_MSG_REPORT.")
Reported-by: Stephen Smalley <sds@tycho.nsa.gov>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

With recent adoption of skc_cookie in struct sock_common,
struct tcp_timewait_sock size increased from 192 to 200 bytes
on 64bit arches. SLAB rounds then to 256 bytes.

It is time to drop SLAB_HWCACHE_ALIGN constraint for twsk_slab.

This saves about 12 MB of memory on typical configuration reaching
262144 timewait sockets, and has no noticeable impact on performance.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge tag 'mac80211-next-for-davem-2015-04-10' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac80211-next

Johannes Berg says:

====================
There isn't much left, but we have
 * new mac80211 internal software queue to allow drivers to have
   shorter hardware queues and pull on-demand
 * use rhashtable for mac80211 station table
 * minstrel rate control debug improvements and some refactoring
 * fix noisy message about TX power reduction
 * fix continuous message printing and activity if CRDA doesn't respond
 * fix VHT-related capabilities with "iw connect" or "iwconfig ..."
 * fix Kconfig for cfg80211 wireless extensions compatibility
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

The statistics are grouped by TX and RX errors.
The SQE Test Errors Register indicates problems with TX.
Signed-off-by: Wolfgang Steinwender <wsteinwender@pcs.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Added get_rxfh_indir_size, get_rxfh_key_size and get_rxfh ethtool_ops
callbacks implementations.

This enables the ethtool's "-x" and "--show-rxfh[-indir]" options for VF
devices.

This patch adds the support for 82599 and x540 devices only. Support for
other devices will be added later.
Signed-off-by: Vlad Zolotarov <vladz@cloudius-systems.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Add the ixgbevf_get_rss_key() function that queries the PF for an RSS
Random Key using a new VF-PF channel IXGBE_VF_GET_RSS_KEY command.

This patch adds the support for 82599 and x540 devices only. Support for
other devices will be added later.
Signed-off-by: Vlad Zolotarov <vladz@cloudius-systems.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

For 82599 and x540 VFs and PF share the same RSS Key. Therefore we will
return the same RSS key for all VFs.

Support for other devices will be added later.
Signed-off-by: Vlad Zolotarov <vladz@cloudius-systems.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

We will currently support only 82599 and x540 devices. Support for other
devices will be added later.

   - Added a new API version support.
   - Added the query implementation in the ixgbevf.
Signed-off-by: Vlad Zolotarov <vladz@cloudius-systems.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Add this new command for 82599 and x540 devices only. Support for other
devices will be added later.

82599 and x540 VFs and PF share the same RSS redirection table (RETA).
Therefore we just return it for all VFs.

For 82599 and x540 RETA table is an array of 32 registers (128 bytes) and
the maximum number of registers that may be delivered in a single VF-PF
channel command is 15. On the other hand VFs of these devices can be
configured to have up to 4 RSS queues. Therefore we will "compress" the
RETA by transferring only 2 bits per entry and thereby it will take only 8
registers (DWORDS) to transfer the whole VF RETA.

Thus this patch does the following:

  - Adds a new API version (to specify a new commands set).
  - Adds the IXGBE_VF_GET_RETA command to the VF-PF commands set.
Signed-off-by: Vlad Zolotarov <vladz@cloudius-systems.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Implements the new netdev op to allow user to enable/disable the ability
of a specific VF to query its RSS Indirection Table and an RSS Hash Key.

This patch limits the new feature support to 82599 and x540 devices only.
Support for other devices will be added later.
Signed-off-by: Vlad Zolotarov <vladz@cloudius-systems.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Add configuration setting for drivers to allow/block an RSS Redirection
Table and a Hash Key querying for discrete VFs.

On some devices VF share the mentioned above information with PF and
querying it may adduce a theoretical security risk. We want to let a
system administrator to decide if he/she wants to take this risk or not.
Signed-off-by: Vlad Zolotarov <vladz@cloudius-systems.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Added get_rxfh_indir_size, get_rxfh_key_size and get_rxfh ethtool_ops
callbacks implementations.

This enables the ethtool's "-x" and "--show-rxfh[-indir]" options.
Signed-off-by: Vlad Zolotarov <vladz@cloudius-systems.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

This patch is a preparation for enablement of ethtool RSS indirection
table and hash key querying. We don't want to read registers every time
the RSS info is queried. Therefore we will store its current content in the
arrays in the adapter struct and will read it from there (instead of from
registers) when requested.

Will change the code that writes the indirection table and hash key into
the HW registers to take its content from these arrays. This will also
simplify the indirection table updating ethtool callback implementation
in the future.
Signed-off-by: Vlad Zolotarov <vladz@cloudius-systems.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Jeff Kirsher says:

====================
Intel Wired LAN Driver Updates 2015-04-10

This series contains updates to ixgbe and documentation for igb,
ixgbe and ixgb.

Stephen cleans up documentation to igb, ixgbe and ixgb.

Don updates how bridge mode is stored to minimize obfuscation and
makes updates for future silicon easier.  Adds a new bridge mode
support function which gathers all the logic needed to configure
bridge modes.  Adds Source Address Prunning for VEPA bridge mode
for x550 devices.

Vasu adds specific FCoE offloads for x550 for DDP context programming
and increased DDP exchanges.

Alex Duyck cleans up the use of HW_VLAN_CTAG_FILTER in hw_features,
where the driver was actually ignoring the value of the bit and was
just assuming it was always set.  Also cleans up the use of rcu_barrier()
since the driver has not used call_rcu() to free the rings for some
time now.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

The ixgbe driver hasn't used call_rcu to free the rings for some time now.
Since that is the case the call to rcu_barrier can be dropped since calls
to kfree_rcu don't require it.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

This change makes it so that the HW_VLAN_CTAG_FILTER bit is not falsely
advertised as being a feature that can be toggled on ixgbe parts.  The
driver was setting the bit in features and letting it be inherited by
hw_features, however the driver was actually ignoring the value of the bit
and just assuming it was always set.  As a result VLAN filtering was always
enabled which is a requirement for SR-IOV, VMDq, DCB, FCoE, and possibly
other features within the adapters.
Signed-off-by: Alexander Duyck <alexander.h.duyck@redhat.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Adds x550 specific FCoE offloads for DDP context programming and
increased DDP exchanges.
Signed-off-by: Vasu Dev <vasu.dev@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>