Using a reader-writer lock in fast path is silly, when we can
instead use RCU or a seqlock.

For mlx4 hwstamp clock, a seqlock is the way to go, removing
two atomic operations and false sharing.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Tariq Toukan <tariqt@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The usage count function is based on ndev_running flag that is
updated before calling ndo_open/close, but if ndo is called in
another place, as with suspend/resume, the counter is not changed,
that breaks sus/resume. For common resource no difference which
device is using it, does matter only device count. So, replace
usage count function on var and inc and dec it in ndo_open/close.

Signed-off-by: Ivan Khoronzhuk <ivan.khoronzhuk@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

pch_gbe_get_stats() just returns dev->stats so we can leave it out
altogether and let dev_get_stats() do the job.

Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

hip04_get_stats() just returns dev->stats so we can leave it
out altogether and let dev_get_stats() do the job.

Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

Fixes the following sparse warning:

net/sched/act_api.c:532:5: warning:
 symbol 'nla_memdup_cookie' was not declared. Should it be static?

Signed-off-by: Wei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

remove useless ioport.h include.

Signed-off-by: Manuel Lauss <manuel.lauss@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

remove the unused timer.  I suppose it was intended as a timeout
detector, but never properly implemented.

Signed-off-by: Manuel Lauss <manuel.lauss@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Fixes the following warnings:

kernel/bpf/verifier.c: In function ‘may_access_direct_pkt_data’:
kernel/bpf/verifier.c:702:6: warning: this statement may fall through [-Wimplicit-fallthrough=]
   if (t == BPF_WRITE)
      ^
kernel/bpf/verifier.c:704:2: note: here
  case BPF_PROG_TYPE_SCHED_CLS:
  ^~~~
kernel/bpf/verifier.c: In function ‘reg_set_min_max_inv’:
kernel/bpf/verifier.c:2057:23: warning: this statement may fall through [-Wimplicit-fallthrough=]
   true_reg->min_value = 0;
   ~~~~~~~~~~~~~~~~~~~~^~~
kernel/bpf/verifier.c:2058:2: note: here
  case BPF_JSGT:
  ^~~~
kernel/bpf/verifier.c:2068:23: warning: this statement may fall through [-Wimplicit-fallthrough=]
   true_reg->min_value = 0;
   ~~~~~~~~~~~~~~~~~~~~^~~
kernel/bpf/verifier.c:2069:2: note: here
  case BPF_JSGE:
  ^~~~
kernel/bpf/verifier.c: In function ‘reg_set_min_max’:
kernel/bpf/verifier.c:2009:24: warning: this statement may fall through [-Wimplicit-fallthrough=]
   false_reg->min_value = 0;
   ~~~~~~~~~~~~~~~~~~~~~^~~
kernel/bpf/verifier.c:2010:2: note: here
  case BPF_JSGT:
  ^~~~
kernel/bpf/verifier.c:2019:24: warning: this statement may fall through [-Wimplicit-fallthrough=]
   false_reg->min_value = 0;
   ~~~~~~~~~~~~~~~~~~~~~^~~
kernel/bpf/verifier.c:2020:2: note: here
  case BPF_JSGE:
  ^~~~

Reported-by: David Binderman <dcb314@hotmail.com>
Signed-off-by: Alexander Alemayhu <alexander@alemayhu.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Even though the port autoselection is enabled by default on AM79C970A,
BNC/AUI port does not work because the link is always reported to be
down. The link state reported by the chip belongs only to the TP port
but the driver uses it regardless of the port used. The chip can't
detect BNC/AUI link state.

Disable port autoselection and use TP port by default to keep current
behavior (link detection works on TP port, BNC/AUI port does not work).

Implement ethtool autoneg, port and duplex configuration to allow
using the BNC/AUI port.

Report the TP link state only if the TP port is selected. When the
port autoselection is enabled or AUI port is selected, report the link
as always up.

Move pcnet32_suspend() and pcnet32_clr_suspend() functions to avoid
forward declarations.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Move the code to clear SUSPEND flag to a separate function to simplify
code.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Point back the unregister IPv6 mc table to the bc table.
It is done since IPv6 mcast snooping is not supported for Spectrum yet.

Reported-by: Jiri Pirko <jiri@mellanox.com>
Fixes: 71c365bd ("mlxsw: spectrum: Separate bc and mc floods")
Signed-off-by: Nogah Frankel <nogahf@mellanox.com>
Signed-off-by: Yotam Gigi <yotamg@mellanox.com>
Tested-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Shannon Nelson says:

====================
sunvnet driver updates

The sunvnet ldom virtual network driver was due for some updates and
a bugfix or two.  These patches address a few items left over from
last year's make-over.

v2:
 - changed memory barrier fix to use smp_wmb
 - put NETIF_F_SG back into the advertised ldmvsw hw_features

v3:
 - the sunvnet_common module doesn't need module_init or _exit

v4:
 - dropped the statistics patch
 - fixed up "default" tag for SUNVNET_COMMON
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

The ldmvsw driver is specifically for supporting the ldom virtual
networking by running in the primary ldom and using the LDC to connect
the remaining ldoms to the outside world via a bridge.  With TSO and GSO
supported while connected the bridge, things tend to misbehave as seen
in our case by delayed packets, enough to begin triggering retransmits
and affecting overall throughput.  By turning off advertised support for
TSO and GSO we restore stable traffic flow through the bridge.

Orabug: 23293104

Signed-off-by: Shannon Nelson <shannon.nelson@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

New version and simplify the print code.

Signed-off-by: Shannon Nelson <shannon.nelson@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The RCU read lock is grabbed first thing in sunvnet_start_xmit_common()
so it always needs to be released.  This removes the conditional release
in the dropped packet error path and removes a couple of superfluous
calls in the middle of the code.

Reported-by: Bijan Mottahedeh <bijan.mottahedeh@oracle.com>
Signed-off-by: Shannon Nelson <shannon.nelson@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The use of gotos for handling the incoming events made this code
harder to read and support than it should be.  This patch straightens
out and clears up the logic.

Signed-off-by: Shannon Nelson <shannon.nelson@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In order to allow the underlying LDC and outstanding memory operations
to potentially catch up with the driver's Tx requests, add a memory
barrier before checking again for available tx descriptors.

Signed-off-by: Shannon Nelson <shannon.nelson@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

There have been several changes since the first version of this code, so
we bump the version number.  While we're at it, we can simplify the
version printing a bit and drop a couple lines of code.

Signed-off-by: Shannon Nelson <shannon.nelson@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The vio_dring_state *dr variable is unused in maybe_tx_wakeup().
As the comments indicate, we call maybe_tx_wakeup() whenever we
get a STOPPED LDC message on the port. If the queue is stopped,
we want to wake it up so that we will send another START message
at the next TX and trigger the consumer to drain the dring.

Signed-off-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
Signed-off-by: Shannon Nelson <shannon.nelson@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When the sunvnet_common code was split out for use by both sunvnet
and the newer ldmvsw, it was made into a static kernel library, which
limits the usefulness of sunvnet and ldmvsw as loadables, since most
of the real work is being done in the shared code.  Also, this is
simply dead code in kernels that aren't running the LDoms.

This patch makes the sunvnet_common into a dynamically loadable
module and makes sunvnet and ldmvsw dependent on sunvnet_common.

Signed-off-by: Shannon Nelson <shannon.nelson@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Edward Cree says:

====================
sfc: prevent bogus interrupt-mode fallbacks

EF10 VFs only support MSI-X interrupts, not MSI or legacy.  This series
 stops the probe logic from trying to fallback to those if MSI-X interrupt
 probe fails.  It also prevents selecting them with the interrupt_mode
 module parameter.
This avoids producing messages like "failed to hook legacy IRQ 0" and "IRQ
 handler type mismatch for IRQ 0", and ensures that the relevant error
 (from the attempt to enable MSI-X) is reported to the caller.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

If we fail to probe interrupts with our minimum mode, return that error.

Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add min_interrupt_mode specification per NIC type.
It is a bit confusing because of "highest interrupt mode is less capable".

Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Nikolay Aleksandrov says:

====================
bridge: minor fdb cleanup

These patches aim to simplify the bridge fdb API a little by removing some
redundant functions and converting them into wrappers of a single function.
Also add proper lock checking to avoid future mistakes for the search
functions.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

We can simplify the logic of entries pointing to the bridge by
converging the fdb_delete_by functions, this would allow us to use the
same function for both cases since the fdb's dst is set to NULL if it is
pointing to the bridge thus we can always check for a port match.

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In order to avoid new errors add checks to br_fdb_find and fdb_find_rcu
functions. The first requires hash_lock, the second obviously RCU.

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Before this patch we had 3 different fdb searching functions which was
confusing. This patch reduces all of them to one - fdb_find_rcu(), and
two flavors: br_fdb_find() which requires hash_lock and br_fdb_find_rcu
which requires RCU. This makes it clear what needs to be used, we also
remove two abusers of __br_fdb_get which called it under hash_lock and
replace them with br_fdb_find().

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

mv88e6097 chip requires watchdog_ops to be set.

Signed-off-by: Volodymyr Bendiuga <volodymyr.bendiuga@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

Bring back the goto that was removed by accident.

Reported-by: Colin Ian King <colin.king@canonical.com>
Fixes: 40c81b25 ("sched: check negative err value to safe one level of indent")
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

We had intended to say "sizeof(u32)" but the "u" is missing.
Fortunately, sizeof(32) is also 4, so the original code still works.

Fixes: c4e7beea ("net: qcom/emac: add ethtool support for reading hardware registers")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Timur Tabi <timur@codeaurora.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

There is no need to use an intermediate variable to handle an error code
in this case.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>

'of_node_put(fpi->phy_node)' should also be called if we branch to
'out_deregister_fixed_link' error handling path.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge tag 'rxrpc-rewrite-20170210' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs



David Howells says:

====================
afs: Use system UUID generation

There is now a general function for generating a UUID and AFS should make
use of it.  It's also been recommended to me that I switch to using random
rather than time plus MAC address-based UUIDs which this function does.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

garp_port is only used in net/802/garp.c which is only compiled with
CONFIG_GARP enabled. Same goes for mrp_port which is only used in
net/802/mrp.c with CONFIG_MRP enabled.

Only include the two members in struct net_device if their respective
CONFIG_* is enabled. This saves a few bytes in struct net_device in case
CONFIG_GARP or CONFIG_MRP are not enabled.

Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

Commit 79e7fff4 ("net: remove support for per driver
ndo_busy_poll()") made them obsolete.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Jeff Kirsher says:

====================
40GbE Intel Wired LAN Driver Updates 2017-02-11

This series contains updates to i40e and i40evf only.

Jake makes a minor change to prevent a minor bit of work, if it is not
necessary.  In the case where we do not have a client, there is no need
to check the client params, so move the check till after we have ensured
we have a client.  Correct a code comment which incorrectly implied
that raw_packet buffers were freed in i40e_clean_tx_ring(), so fixed
the code comment to better explain where memory is freed.  Reduce the
severity and frequency of the message notifying we cleared the receive
timestamp register, since the logic has a much better detection scheme
that could detect a stalled receive timestamp register.  The improved
logic was actually causing the notification message to occur more
frequently and was giving the user a false perception that a timestamp
event was missed for a valid packet, so reduce the severity from
dev_warn to dev_dbg and only fire off the message when 3 or 4 of the
RXTIME registers are stalled and get cleared within the same
watchdog event.  Fixed a bug, where we were modifying the mac_filter
outside a lock when handling the addition of broadcast filters.  Fix
this by updating i40e_update_filter_state logic so that it knows to
avoid broadcast filters, which ensures that we do not have to remove
the filter separately and can put it back using the normal flow.
Refactored how we add new filters to firmware to avoid a race condition
that can occur due to removing filters from the hash temporarily.

Mitch adds a sleep (without timeout) so that we wait for a reply from
the PF before we continue, since the iWarp client cannot continue until
the operation is completed.  Fixed up a function which could never
return an error, to be void and cleaned up the checking of the now
null and void return value.

Scott limits the DMA sync to CPU to the actual length of the incoming
packet, versus the syncing of the entire buffer.  Also reduces the
receive buffer struct (by a single pointer) and align the driver to be
more consistent with other Intel drivers with respect to packets that
span buffers.

Sudheer adds a field to track the bus number info and modified log
statements to print bus, device and function information.

Henry adds the ability to store the FEC status bits from the link up
event.  Also adds the ethtool support for FEC capabilities and 25G
link types.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

Using new link mode indices instead deprecated SUPPORTED_/ADVERTISED_
macro.

Added indication for 2500 and 5000mbit link modes (AQtion adapter already
supports these speeds).

Signed-off-by: Pavel Belous <pavel.belous@aquantia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Andrew Lunn says:

====================
mv88e6xxx Watchdog support

The Marvell switches have an in built watchdog over some of the
internal state machine. The watchdog can be configured to raise an
interrupt on error. The problem the watchdog found is then logged to
the kernel log.

The older switches can automagically perform a software reset when the
watchdog triggers. This just resets the internal state machine, but
leaves the switch configuration unchanged.

The 6390 family of switches cannot both raise an interrupt and
automagically perform a software reset. So the interrupt handler has
to perform the switch reset, and then re-enable the watchdog
interrupts.

This has been tested using hacked together debugfs code which allows
the "force" bit to be set, so cause a watchdog interrupt.

v2: Remove g2_prefix
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

Implement the ops needed to support the watchdog for the MV88E6390
family.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

The switch contains a watchdog looking for issues with the internal
gubbins of the switch. Hook the interrupt the watchdog triggers and
log the value of the control register indicating why the watchdog
fired. The watchdog can only be cleared with a switch reset, which
will destroy the current configuration. Rather than doing this, just
disable the interrupt.

The mv88e6390 family has different watchdog registers. So use an ops
structure, so support for the mv88e6390 family can be added later.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>