Add device tree binding documentation for the Broadcom iProc MDIO
bus driver.
Signed-off-by: NArun Parameswaran <arunp@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Santosh Shilimkar says:

====================
RDS: connection scalability and performance improvements

[v4]
Re-sending the same patches from v3 again since my repost of
patch 05/14 from v3 was whitespace damaged.

[v3]
Updated patch "[PATCH v2 05/14] RDS: defer the over_batch work to
send worker" as per David Miller's comment [4] to avoid the magic
value usage. Patch now makes use of already available but unused
send_batch_count module parameter. Rest of the patches are same as
earlier version v2 [3]

[v2]:
Dropped "[PATCH 05/15] RDS: increase size of hash-table to 8K" from
earlier version [1]. I plan to address the hash table scalability using
re-sizable hash tables as suggested by David Laight and David Miller [2]

This series addresses RDS connection bottlenecks on massive workloads and
improve the RDMA performance almost by 3X. RDS TCP also gets a small gain
of about 12%.

RDS is being used in massive systems with high scalability where several
hundred thousand end points and tens of thousands of local processes
are operating in tens of thousand sockets. Being RC(reliable connection),
socket bind and release happens very often and any inefficiencies in
bind hash look ups hurts the overall system performance. RDS bin hash-table
uses global spin-lock which is the biggest bottleneck. To make matter worst,
it uses rcu inside global lock for hash buckets.
This is being addressed by simply using per bucket rw lock which makes the
locking simple and very efficient. The hash table size is still an issue and
I plan to address it by using re-sizable hash tables as suggested on the list.

For RDS RDMA improvement, the completion handling is revamped so that we
can do batch completions. Both send and receive completion handlers are
split logically to achieve the same. RDS 8K messages being one of the
key usecase, mr pool is adapted to have the 8K mrs along with default 1M
mrs. And while doing this, few fixes and couple of bottlenecks seen with
rds_sendmsg() are addressed.

Series applies against 4.3-rc1 as well net-next. Its tested on Oracle
hardware with IB fabric for both bcopy as well as RDMA mode. RDS TCP is
tested with iXGB NIC. Like last time, iWARP transport is untested with
these changes. The patchset is also available at below git repo:

git://git.kernel.org/pub/scm/linux/kernel/git/ssantosh/linux.git net/rds/4.3-v3

As a side note, the IB HCA driver I used for testing misses at least 3
important patches in upstream to see the full blown IB performance and
am hoping to get that in mainline with help of them.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Eric W. Biederman says:

====================
net: Pass net through the output path v2

This is the next installment of my work to pass struct net through the
output path so the code does not need to guess how to figure out which
network namespace it is in, and ultimately routes can have output
devices in another network namespace.

The first patch in this series is a fix for a bug that came in when sk
was passed through the functions in the output path, and as such is
probably a candidate for net.  At the same time my later patches depend
on it so sending the fix separately would be confusing.

The second patch in this series is another fix that for an issue that
came in when sk was passed through the output path.  I don't think it
needs a backport as I don't think anyone uses the path where the code
was incorrect.

The rest of the patchset focuses on the path from xxx_local_out to
dst_output and in the end succeeds in passing sock_net(sk) from the
socket a packet locally originates on to the dst->output function.

Given the size reduction in the code I think this counts as a cleanup as
much as feature work.

There remain a number of helper functions (like ip option processing) to
take care of before the network stack can support destination devices in
other network namespaces but with this set of changes the backbone of
the work is done.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The network namespace is already passed into dst_output pass it into
dst->output lwt->output and friends.
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Compute net once in ipvlan_process_v4_outbound and
ipvlan_process_v6_outbound and store it in a variable so that net does
not need to be recomputed next time it is used.
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Compute net and store it in a variable in pptp_xmit, so that the value
can be reused the next time it is needed.
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Compute net and store it in a variable in the functions
ip_build_and_send_pkt and ip_queue_xmit so that it does not need to be
recomputed next time it is needed.
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Store net in a variable in ip_tunnel_xmit so it does not need
to be recomputed when it is used again.
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Stop hidding the sk parameter with an inline helper function and make
all of the callers pass it, so that it is clear what the function is
doing.
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Only __ip6_local_out_sk has callers so rename __ip6_local_out_sk
__ip6_local_out and remove the previous __ip6_local_out.
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

It is confusing and silly hiding a parameter so modify all of
the callers to pass in the appropriate socket or skb->sk if
no socket is known.
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

For consistency with the other similar methods in the kernel pass a
struct sock into the dst_ops .local_out method.

Simplifying the socket passing case is needed a prequel to passing a
struct net reference into .local_out.
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Replace dst_output_okfn with dst_output
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

After a packet has been encapsulated by a tunnel we should use the
tunnel sockets local multicast loopback flag to control if the
encapsulated packet should be locally loopback back.

Pass sk into ip_local_out_sk so that in the rare case we are dealing
with a tunneled packet whose tunnel destination address is a multicast
address the kernel properly decides to loopback this packet.

In practice I don't think this matters as ip_queue_xmit is used by
tcp, l2tp and sctp none of which I am aware of uses ip level
multicasting as they are all point to point communications protocols.
Let's fix this before someone uses ip_queue_xmit for a tunnel protocol
that does use multicast.

Fixes: aad88724 ("ipv4: add a sock pointer to dst->output() path.")
Fixes: b0270e91 ("ipv4: add a sock pointer to ip_queue_xmit()")
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In the rare case where sk != skb->sk ip_local_out_sk arranges
to call dst->output differently if the skb is queued or not.
This is a bug.

Fix this bug by passing the sk parameter of ip_local_out_sk through
from ip_local_out_sk to __ip_local_out_sk (skipping __ip_local_out).

Fixes: 7026b1dd ("netfilter: Pass socket pointer down through okfn().")
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Jeff Kirsher says:

====================
Intel Wired LAN Driver Updates 2015-10-07

This series contains updates to i40e and i40evf only.

Paul updates i40e to simply increase the amount of time we wait for a
reset to complete since we have seen in some rare occasions the reset
can take longer to complete.

Shannon updates the driver to turn on Wake-on-LAN by default if it is
enabled in the hardware config to begin with, rather than always disable
it and wait for the user to expressly turn it on.  Added new device id's
and support for future devices.  Fixed a possible type compare problem
between a size and possible negative number.  Also fixed a shift value
that was wrong, which ended up with a bad bitmask.  Did general house
cleaning of the driver to cleanup several low lying fruit in the
driver.  Fixed an issue where new unicast address's would be added to
the VSI list and then immediately removed and would never actually
make it down to the hardware.  Resolved the issue by removing the
separation from unicast and multicast in the search for filters to be
deleted.

Mitch fixes an issue where the hardware would continue to access the
memory formerly used by the rings for a VF which have been removed,
causing memory corruption or DMAR errors.  To relieve this condition,
explicitly stop all rings associated with each VF before releasing its
resources.  Also fixed a panic if the driver is unable to enable MSI-X
or its unable to acquire enough vectors, so propagate interrupt
allocation failure information to the calling function.  Cleaned up
opcode that is not required.

Carolyn extends the size of the test available for the interrupt names
so that all the descriptive data available for the Flow Director
interrupts is not truncated.

Catherine fixes an issue where there was a possibility of speed getting
set to 0 if advertised is set to 0 (which is the case when autoneg is
disabled).

Jesse fixes the checksum on big endian machines, so added code to swap
it correctly.  Also fixed a bug in the return from get_link_status()
where only true or false was being returned, but false could mean
multiple things.  So allow the caller to get all the return values
in the call chain bubbled back to the source so that the reason for
the failure does not get lost.

Anjali adds statistics to keep track of how many times we ask the stack
to linearize the SKB because the hardware cannot handle SKBs with more
than 8 frags per segment/single packet.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Merge tag 'regmap-offload-update-bits' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap

regmap: Allow buses to provide a custom update_bits() operation

Some buses provide a native _update_bits() operation which for uncached
registers is faster than doing a read/modify/write cycle as it is a
single bus transaction.  Add support for implementing this to regmap.

This opcode is not required. VFs that program RSS through the firmware
do it by interacting directly with the firmware, and do not need to use
the virtual channel for this functionality.

Change-ID: Iaf17d2600e28ff1b6be8653f2fe9df1facd23b0e
Signed-off-by: NMitch Williams <mitch.a.williams@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Lower level functions are properly reporting errors, and higher-level
functions are correctly responding to errors, but the errors aren't
actually getting through. Typically, the middle-manager function seems
to want to shield its boss from any bad news.

This change fixes a panic if the driver is unable to enable MSI-X or is
unable to acquire enough vectors.

Change-ID: Ifd5787ce92519a5d97e4b465902db930d97b71a1
Signed-off-by: NMitch Williams <mitch.a.williams@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

The firmware has added additional status information to allow software
to determine if the APP priority for FCoE/iSCSI/FIP is valid or not in
CEE DCBX mode.

This patch adds to support those additional checks and will only add
applications to the software table that have oper and sync bits set
without any error.

Change-ID: I0a76c52427dadf97d4dba4538a3068d05e4eb56b
Signed-off-by: NNeerav Parikh <neerav.parikh@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Keep track of how many times we ask the stack to linearize the
skb because the HW cannot handle skbs with more than 8 frags per
segment/single packet.

Change-ID: If455452060963a769bbe6112cba952e79e944b52
Signed-off-by: NAnjali Singhai Jain <anjali.singhai@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

When using something like "ip maddr add ..." to add another unicast mac
address to the netdev, the mac address comes into the set_rx_mode handler
in the multicast list whether it is a unicast or multicast address.
This was confusing the code when it was trying to search for addresses
that needed to be deleted from the VSI, because it was looking for the
VSI unicast address in the netdev unicast list.  The result was that a
new unicast address would get added to the VSI list and then immediately
removed, and would never actually make it down into the hardware.

This patch removes the separation from unicast and multicast in the search
for filters to be deleted.  It also simplifies the logic a little with a
jump to the bottom of the loop when an address is found.  Now it doesn't
matter which netdev list the address is hiding in, we'll check them all.

Change-ID: Ie3685a92427ae7d2212bf948919ce295bc7a874c
Signed-off-by: NShannon Nelson <shannon.nelson@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Previously, the driver could call this function and have only true/false
returned, but false could mean multiple things like failure to read
or link was down. This change allows the caller to get all return values
in the call chain bubbled back to the source, which keeps information about
failures from being lost.

Also, in some unlikely scenarios, the firmware can become slow to respond
to admin queue (AQ) queries for link state.  Should the AQ time out,
the driver can detect the state and avoid a link change when there
may have been none.

Change-ID: Ib2ac38407b7880750fb891b392fa77457fe6c21c
Signed-off-by: NJesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

The checksum is not correct on big endian machines so add code to swap it
correctly.

Change-ID: Ic92b886d172a2cbe49f5d7eee1bc78e447023c7b
Signed-off-by: NJesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: NPaul M Stillwell Jr <paul.m.stillwell.jr@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

During early development, we added the function name to all of the error
strings to make debugging simpler. Now that we've released the driver,
our users should have more comprehensible error messages. So tear the
roof off and give up the __func__. Ow.

Change-ID: I7e1766252c7a032b9af6520da6aff536bdfd533c
Signed-off-by: NShannon Nelson <shannon.nelson@intel.com>
Signed-off-by: NMitch Williams <mitch.a.williams@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

In ethtool, there is a possibility of speed getting set to 0
if advertise is set to 0 (which it is when autoneg is disabled).
We never want this to happen as the firmware will actually attempt
to set the speed to 0 sending link down, so add an extra check
to make sure this doesn't happen.

Change-ID: I62e0eeee2cbf043d8e6f5c9c9f0b92794e877f01
Signed-off-by: NCatherine Sullivan <catherine.sullivan@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

This patch extends the size of the text available for the interrupt names.
Without this patch, all the descriptive data available for the Flow
Director interrupts is truncated.

Change-ID: I2ac458f23ac3b4ea8f1edf73edc283b1d3704c7f
Signed-off-by: NCarolyn Wyborny <carolyn.wyborny@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

There was a possibility where the asq_last_status could get through without
update and thus report a previous error.  I don't think we've actually seen
this happen, but this patch will help make sure it doesn't.

Change-ID: I9e33927052a5ee6ea21f80b66d4c4b76c2760b17
Signed-off-by: NShannon Nelson <shannon.nelson@intel.com>
Signed-off-by: NChristopher Pau <christopher.pau@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

i40e_init_pf_fcoe() didn't return anything except 0, it prints enough
error info already, and no driver logic depends on the return value,
so this can be void.

Change-ID: Ie6afad849857d87a7064c42c3cce14c74c2f29d8
Signed-off-by: NShannon Nelson <shannon.nelson@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Fix a shift value that was wrong, ending up with a bad bitmask.  Also add
a blank line between two sets of #defines for better readability.

Change-ID: I3e41fa2a2ab904d3a4e6cbf13972ab0036a10601
Signed-off-by: NShannon Nelson <shannon.nelson@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Rework an if expression to assure there is no type compare problem between
a size and a possible negative number.

Change-ID: I4921fcc96abfcf69490efce020a9e4007f251c99
Reported-by: NHelin Zhang <helin.zhang@intel.com>
Signed-off-by: NShannon Nelson <shannon.nelson@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Early addition of new a device id.

Change-ID: I61a8c8556fdf4f5714be4e4089689e374f30293c
Signed-off-by: NShannon Nelson <shannon.nelson@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Explicitly stop the rings belonging to each VF when disabling SR-IOV.
Even though the VFs were gone, and the associated VSIs were removed, the
rings were not stopped, and in some circumstances the hardware would
continue to access the memory formerly used by the rings, causing memory
corruption or DMAR errors, both of which would lead to general malaise
of the kernel.

To relieve this condition, explicitly stop all the rings associated with
each VF before releasing its resources.

Change-ID: I78c05d562c66e7b594b7e48d67860f49b3e5b6ec
Signed-off-by: NMitch Williams <mitch.a.williams@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

The driver was disabling Wake-on-LAN by default and waiting for the user
to expressly turn it on.  This patch has the driver turning on WoL from
the start if enabled in the hardware config, which matches the behavior
of our other drivers.

Change-ID: I43faedb907f8ba4d1a61b72a7c86072b97af12b1
Signed-off-by: NShannon Nelson <shannon.nelson@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

In some rare cases the reset can take longer to complete so increase the
amount of time we wait.

Change-ID: Ib5628ec54b526a811ee33d1214fe763226406671
Signed-off-by: NPaul M Stillwell Jr <paul.m.stillwell.jr@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Some applications use a listen() backlog of 1.

Prior kernels were silently enforcing a qlen_log of 4, so that we were
sending up to /proc/sys/net/ipv4/tcp_synack_retries SYNACK messages.

Fixes: ef547f2a ("tcp: remove max_qlen_log")
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>