Introduce ____napi_schedule() helper for callers in irq disabled
contexts. rps_trigger_softirq() becomes a leaf function.

Use container_of() in process_backlog() instead of accessing per_cpu
address.

Use a custom inlined version of __napi_complete() in process_backlog()
to avoid one locked instruction :

 only current cpu owns and manipulates this napi,
 and NAPI_STATE_SCHED is the only possible flag set on backlog.
 we can use a plain write instead of clear_bit(),
 and we dont need an smp_mb() memory barrier, since RPS is on,
 backlog is protected by a spinlock.
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

NET_IP_ALIGN defaults to 2, no need to override.

NET_SKB_PAD is now 64, which is much larger than microblaze's
L1_CACHE_SIZE so no need to override that either.
Signed-off-by: David S. Miller <davem@davemloft.net>

eth_type_trans() & get_rps_cpus() currently need two 64bytes cache
lines in packet to compute rxhash.

Increasing NET_SKB_PAD from 32 to 64 reduces the need to one cache
line only, and makes RPS faster.

NET_IP_ALIGN(2) + ethernet_header(14) + IP_header(20/40) + ports(8)
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Adding addresses and ports to the short packet log message,
like ipv4/udp.c does it, makes these messages a lot more useful:

[  822.182450] UDPv6: short packet: From [2001:db8:ffb4:3::1]:47839 23715/178 to [2001:db8:ffb4:3:5054:ff:feff:200]:1234

This requires us to drop logging in case pskb_may_pull() fails,
which also is consistent with ipv4/udp.c
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Use the resource_size function instead of manually calculating the
resource size.  This reduces the chance of introducing off-by-one
errors.
Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

Reset the PHY before first accessing it.  Doing so, ensure that the PHY is
in a known good state before we read/write PHY registers. This fixes a
driver probe failure.
Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

During Sx->S0 transitions, the interconnect between the MAC and PHY on
82577/82578 can remain in SMBus mode instead of transitioning to the
PCIe-like mode required during normal operation.  Toggling the LANPHYPC
Value bit essentially resets the interconnect forcing it to the correct
mode.
Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

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

Based on Andy's work, but I modified a lot.

Similar to the patch for bridge, this patch does:

1) implement the 2 methods to support netpoll for bonding;

2) modify netpoll during forwarding packets via bonding;

3) disable netpoll support of bonding when a netpoll-unabled device
   is added to bonding;

4) enable netpoll support when all underlying devices support netpoll.

Cc: Andy Gospodarek <gospo@redhat.com>
Cc: Jeff Moyer <jmoyer@redhat.com>
Cc: Matt Mackall <mpm@selenic.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Cc: Jay Vosburgh <fubar@us.ibm.com>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: WANG Cong <amwang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Based on the previous patch, make bridge support netpoll by:

1) implement the 2 methods to support netpoll for bridge;

2) modify netpoll during forwarding packets via bridge;

3) disable netpoll support of bridge when a netpoll-unabled device
   is added to bridge;

4) enable netpoll support when all underlying devices support netpoll.

Cc: David Miller <davem@davemloft.net>
Cc: Neil Horman <nhorman@tuxdriver.com>
Cc: Stephen Hemminger <shemminger@linux-foundation.org>
Cc: Matt Mackall <mpm@selenic.com>
Signed-off-by: WANG Cong <amwang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This whole patchset is for adding netpoll support to bridge and bonding
devices. I already tested it for bridge, bonding, bridge over bonding,
and bonding over bridge. It looks fine now.

To make bridge and bonding support netpoll, we need to adjust
some netpoll generic code. This patch does the following things:

1) introduce two new priv_flags for struct net_device:
   IFF_IN_NETPOLL which identifies we are processing a netpoll;
   IFF_DISABLE_NETPOLL is used to disable netpoll support for a device
   at run-time;

2) introduce one new method for netdev_ops:
   ->ndo_netpoll_cleanup() is used to clean up netpoll when a device is
     removed.

3) introduce netpoll_poll_dev() which takes a struct net_device * parameter;
   export netpoll_send_skb() and netpoll_poll_dev() which will be used later;

4) hide a pointer to struct netpoll in struct netpoll_info, ditto.

5) introduce ->real_dev for struct netpoll.

6) introduce a new status NETDEV_BONDING_DESLAE, which is used to disable
   netconsole before releasing a slave, to avoid deadlocks.

Cc: David Miller <davem@davemloft.net>
Cc: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: WANG Cong <amwang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When the PF acks a message from the VF the VF gets an interrupt.  It
must cache the ack bit so that polling SW will not miss the ack.  Also
avoid reading the message buffer on acks because that also will clear
the ack bit.
Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The driver was calling the set Rx mode function for every multicast
filter set by the VF.  When starting many VMs where each might have
multiple VLAN interfaces this would result in the function being
called hundreds or even thousands of times.  This is unnecessary
for the case of the imperfect filters used in the MTA and has been
streamlined to be more efficient.
Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The driver is unnecessarily writing values to VLAN control registers.
These writes already done elsewhere and are superfluous here.
Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch reduces the number of skb cache misses in the
clean_tx_irq path, and results in an overall increase
in tx packet throughput.
Signed-off-by: Nicholas Nunley <nicholasx.d.nunley@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Accessing ks8851 companion eeprom permits modifying the ks8851 stored
MAC address.

Example how to change the MAC address using ethtool, to set the
01:23:45:67:89:AB MAC address:
$ echo "0:AB8976452301" | xxd -r > mac.bin
$ sudo ethtool -E eth0 magic 0x8870 offset 2 < mac.bin
Signed-off-by: Sebastien Jan <s-jan@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Low-level functions provide 16bits words read and write capability
to ks8851 companion eeprom.
Signed-off-by: Sebastien Jan <s-jan@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

CCR register contains information on companion eeprom availability.
Signed-off-by: Sebastien Jan <s-jan@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Repeated calls to nv_rx_process in napi poll routine do not take
portion of budget that has been consumed in previous calls.  Fix by
subtracting the number of packets processed.
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Stop the queue when we add the packet that will fill it instead of dropping the packet
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

We also reduce the high water mark to 1 so skbufs are not stranded for
long periods of time.  Since we are cleaning after each packet, no
need to do it in the transmit path.
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Don't re-read the interrupt status register, clear the exact bits we
will be testing.
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Under heavy load the TX cleanup tasklet and xmit threads would race
and try to free too many buffers.
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The original implementation incorrectly uses netdev->dev_addrs.

Use netdev->uc instead.  Also use netdev_for_each_uc_addr to iterate
over the addresses.  Fix comment.
Signed-off-by: David Daney <ddaney@caviumnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add support for the "ip link set" and "ip link show" commands that allow
configuration of the virtual functions' MAC and port VLAN via user space
command line.
Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add a boolean parameter to ixgbe-set_vmolr so that the caller can
specify whether the pool should accept untagged packets.  Required
for a follow on patch to enable administrative configuration of port
VLAN for virtual functions.
Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Back before e1000-7.3.20, the e1000 driver had a simple algorithm that
managed interrupt moderation.  The driver was updated in 7.3.20 to
have the new "adaptive" interrupt moderation but we have customer
requests to redeploy the old way as an option.  This patch adds the
old functionality back.  The new functionality can be enabled via
module parameter or at runtime via ethtool.
Module parameter: (InterruptThrottleRate=4) to use this new
moderation method.
Ethtool method: ethtool -C ethX rx-usecs 4
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This change increases the RX fifo size to 36K for standard frames and
decreases the TX fifo size to 4K.  The reason for this change is that on
slower systems the RX is much more likely to backfill and need space than
the TX is.  As long as the TX fifo is twice the size of the MTU we should
have more than enough TX fifo.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Set net->devirq to pdev->irq.  This should be consistent with other
drivers.
Signed-off-by: Tom Herbert <therbert@google.com>
Acked-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Remove e_info message printed whenever TSO is enabled or disabled.
This is not very useful and just clutters dmesg.
Signed-off-by: Tom Herbert <therbert@google.com>
Acked-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Reduce number of writes to RX producer pointer.   When alloc'ing RX
buffers, only write the RX producer pointer once every
E1000_RX_BUFFER_WRITE (16) buffers created.
Signed-off-by: Tom Herbert <therbert@google.com>
Acked-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In e1000_tx_map, precompute number of segements and bytecounts which
are derived from fields in skb; these are stored in buffer_info.  When
cleaning tx in e1000_clean_tx_irq use the values in the associated
buffer_info for statistics counting, this eliminates cache misses
on skb fields.
Signed-off-by: Tom Herbert <therbert@google.com>
Acked-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Flag used in check to get rxhash out of the descriptor is incorrect one.
Fix to use the proper features flag.
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'for-davem' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless-next-2.6

Merge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wireless-next-2.6 into for-davem

Conflicts:
	drivers/net/wireless/libertas_tf/cmd.c
	drivers/net/wireless/libertas_tf/main.c

"mac80211: improve IBSS scanning" was missing a hunk.
This adds that hunk as originally intended.
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

Signed-off-by: John W. Linville <linville@tuxdriver.com>
Acked-by: Hin-Tak Leung <htl10@users.sourceforge.net>

With following patch I can reach maximum rate of my pktgen+udpsink
simulator :
- 'old' machine : dual quad core E5450  @3.00GHz
- 64 UDP rx flows (only differ by destination port)
- RPS enabled, NIC interrupts serviced on cpu0
- rps dispatched on 7 other cores. (~130.000 IPI per second)
- SLAB allocator (faster than SLUB in this workload)
- tg3 NIC
- 1.080.000 pps without a single drop at NIC level.

Idea is to add two prefetchw() calls in __alloc_skb(), one to prefetch
first sk_buff cache line, the second to prefetch the shinfo part.

Also using one memset() to initialize all skb_shared_info fields instead
of one by one to reduce number of instructions, using long word moves.

All skb_shared_info fields before 'dataref' are cleared in 
__alloc_skb().
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>