When RSS is enabled, the status block format changes slightly.  The
"rx_jumbo_consumer", "reserved", and "rx_mini_consumer" members get
mapped to the other three rx return ring producer indexes.  This patch
introduces a new per-interrupt member which identifies which location
in the status block a particular vector should look for return ring
updates.
Signed-off-by: NMatt Carlson <mcarlson@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When multivector RSS is enabled, the first interrupt vector is only used
to report link interrupts and error conditions.  This patch changes the
code so that rx and tx ring resources are not allocated for this vector.
Signed-off-by: NMatt Carlson <mcarlson@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch adds code needed to enable RSS.
Signed-off-by: NMatt Carlson <mcarlson@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch adds code to tune the coalescing parameters for the other
msix vectors.
Signed-off-by: NMatt Carlson <mcarlson@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch adds code to enable and disable the rest of the NAPI
instances.
Signed-off-by: NMatt Carlson <mcarlson@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch exposes the additional transmit rings to the kernel and makes
the necessary modifications to transmit, open, and close paths.
Signed-off-by: NMatt Carlson <mcarlson@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch fixes up two spots that need attention now that msix support
has been added.
Signed-off-by: NMatt Carlson <mcarlson@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch adds code to assign status block, tx producer ring and rx
return ring resources needed for the other interrupt vectors.
Signed-off-by: NMatt Carlson <mcarlson@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The 5717 assigns mailbox locations to interrupt vectors in a rather
non-intuitive way.  (Much of the complexity stems from legacy
compatibility issues.)  This patch implements the assignment scheme.
Signed-off-by: NMatt Carlson <mcarlson@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch adds MSI-X support.
Signed-off-by: NMatt Carlson <mcarlson@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch adds code to support multiple interrupt vectors around the
kernel's interrupt API.
Signed-off-by: NMatt Carlson <mcarlson@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch moves most of the chip ring setup logic into a separate
function.  This will make it easier to verify the multi ring setup
changes.
Signed-off-by: NMatt Carlson <mcarlson@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Each interrupt vector has its own bit in the host coalescing register to
force that vector's status block to be updated and generate an
interrupt.  This patch adds a member to the per-interrupt structure
that records which bit belongs to that vector.
Signed-off-by: NMatt Carlson <mcarlson@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch inlines the code of tg3_cond_int() into the function's only
callsite.  This prep work makes the following patch cleaner.
Signed-off-by: NMatt Carlson <mcarlson@broadcom.com>
Reviewed-by: NBenjamin Li <benli@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Conflicts:
	drivers/net/yellowfin.c

qdisc drops should be notified to IP_RECVERR enabled sockets, as done in IPV4.
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The net_dev of backlog napi is NULL, like below:

__get_cpu_var(softnet_data).backlog.dev == NULL

So, we should check it in napi tracepoint's probe function
Acked-by: NNeil Horman <nhorman@tuxdriver.com>
Signed-off-by: NXiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

These are full of unresolved problems, mainly that conversions don't
work 1-1 from hrtimers to tasklet_hrtimers because unlike hrtimers
tasklets can't be killed from softirq context.

And when a qdisc gets reset, that's exactly what we need to do here.

We'll work this out in the net-next-2.6 tree and if warranted we'll
backport that work to -stable.

This reverts the following 3 changesets:

a2cb6a4d
("pkt_sched: Fix bogon in tasklet_hrtimer changes.")

38acce2d
("pkt_sched: Convert CBQ to tasklet_hrtimer.")

ee5f9757
("pkt_sched: Convert qdisc_watchdog to tasklet_hrtimer")
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Noticed by Michal Simek.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

After commit 2b85a34e
(net: No more expensive sock_hold()/sock_put() on each tx)
sk_free() frees socks conditionally and depends
on sk_wmem_alloc being set e.g. in sock_init_data(). But in some
cases sk_free() is called earlier, usually after other alloc errors.

Fix is to move sk_wmem_alloc initialization from sock_init_data()
to sk_alloc() itself.
Signed-off-by: NJarek Poplawski <jarkao2@gmail.com>
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

These tables are never modified at runtime. Move to read-only
section.
Signed-off-by: NStephen Hemminger <shemminger@vyatta.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

aup->mii_bus->irq allocation may fail, prevent a dereference of NULL.
Signed-off-by: NRoel Kluin <roel.kluin@gmail.com>
Acked-by: NFlorian Fainelli <florian@openwrt.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch updates the sysctl documentation concerning the interpretation
of tcp_retries{1,2} and tcp_orphan_retries.
Signed-off-by: NDamian Lukowski <damian@tvk.rwth-aachen.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch affects the retransmits_timed_out() function.

Changes:
1) Variables have more meaningful names
2) retransmits_timed_out() has an introductionary comment.
3) Small coding style changes.
Signed-off-by: NDamian Lukowski <damian@tvk.rwth-aachen.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

B0_CTST is a 24bit register according to the vendor driver (sk98lin).
A 16bit read on B0_CTST will always return 0 for Y2_VAUX_AVAIL (1<<16),
 so use a 32bit read when testing Y2_VAUX_AVAIL
Signed-off-by: NMike McCormack <mikem@ring3k.org>
Acked-by: NStephen Hemminger <shemminger@vyatta.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Refactor similar two sections of code that free buffers into one.
Only call tx_init if all buffer allocations succeed.
Signed-off-by: NMike McCormack <mikem@ring3k.org>
Acked-by: NStephen Hemminger <shemminger@vyatta.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Observed by Mike McCormack.

The LED bit here is just a software controlled value used to
turn on one of the LED's on some boards. The register value was wrong,
which could have been causing some power control issues.
Get rid of problematic define use the correct mask.
Signed-off-by: NStephen Hemminger <shemminger@vyatta.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

As tun always has an embeedded struct sock,
use sk and sk_receive_queue fields instead of
duplicating them in tun_struct.
Signed-off-by: NMichael S. Tsirkin <mst@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

struct net::ipv6.ip6_dst_ops is separatedly dynamically allocated,
but there is no fundamental reason for it. Embed it directly into
struct netns_ipv6.

For that:
* move struct dst_ops into separate header to fix circular dependencies
	I honestly tried not to, it's pretty impossible to do other way
* drop dynamical allocation, allocate together with netns

For a change, remove struct dst_ops::dst_net, it's deducible
by using container_of() given dst_ops pointer.
Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We can speedup ether addresses compares using compare_ether_addr_64bits()
instead of memcmp(). We make sure all operands are at least 8 bytes long and
16bits aligned (or better, long word aligned if possible)
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Reviewed-by: NJiri Pirko <jpirko@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

To speedup ether addresses compares, we can use compare_ether_addr_64bits()
(all operands are guaranteed to be at least 8 bytes long)
Signed-off-by: NEric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

ixgbe: Patch to fix 82599 multispeed fiber link issues when driver is loaded without any cable and reconnecting it to 1G partner

In 82599 multi speed fiber case when driver is loaded without any
cable and reconnecting the cable with a 1G partner does not bring
up the link in 1Gb mode. When there is no link we first setup the link
at 10G & 1G and then try to re-establish the link at highest speed 10G
and thereby changing autoneg_advertised value to highest speed 10G.
After connecting back the cable to a 1G link partner we never try 1G
as autoneg advertised value is changed to link at 10G only. The
following patch fixes the issue by properly initializing the
autoneg_advertised value just before exiting from link setup routine.
Signed-off-by: NMallikarjuna R Chilakala <mallikarjuna.chilakala@intel.com>
Acked-by: NPeter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When disabling the Rx and Tx data arbiters prior to configuration changes,
the arbiters were not being shut down properly.  This can create a race
in the DCB hardware blocks, and potentially hang the arbiters.  Also, the
Tx descriptor arbiter shouldn't be disabled when applying configuration
changes; disabling this arbiter can cause a Tx hang.
Signed-off-by: NPeter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Link code cleanup: a number of redundant functions and MAC variables are cleaned up,
with some functions being consolidated into a single-purpose code path.
Removed following deprecated link functions and mac variables
 * ixgbe_setup_copper_link_speed_82598
 * ixgbe_setup_mac_link_speed_multispeed_fiber
 * ixgbe_setup_mac_link_speed_82599
 * mac.autoneg, mac.autoneg_succeeded, phy.autoneg_wait_to_complete
Signed-off-by: NMallikarjuna R Chilakala <mallikarjuna.chilakala@intel.com>
Acked-by: NPeter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Enabling VLAN filters (VFE) when the primary interface is brought up
(per commit 78ed11a5) has caused problems for some users who manage
their systems using IPMI over a VLAN. This is because when the driver
enables the VLAN filter, this same filter table is enabled for the
management channel, and the table is initially empty, which means that
the IPMI/VLAN packets are filtered out and not received by the BMC.
This is a problem only on e1000 class adapters, as it is only
on e1000 that the filter table is common to the management and host
streams.

With this change, filtering is only enabled when one or more host VLANs
exist, and is disabled when the last host VLAN is removed. VLAN filtering
is always disabled when the primary interface is in promiscuous mode,
and will be (re)enabled if VLANs exist when the interface exits
promiscuous mode.

Note that this does not completely resolve the issue for those using VLAN
management, because if the host adds a VLAN, then the above problem
occurs when that VLAN is enabled. However, it does mean the there is no
problem for configurations where management is on a VLAN and the host is
not.

A complete solution to this issue would require further driver changes.
The driver would need to discover if (and which) management VLANs are
active before enabling VLAN filtering, so that it could ensure that the
managed VLANs are included in the VLAN filter table. This discovery
requires that the BMC identifies its VLAN in registers accessible
to the driver, and at least on Dell PE2850 systems the BMC does not
identify its VLAN to allow such discovery. Intel is pursuing this issue
with the BMC vendor.
Signed-off-by: NDave Graham <david.graham@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Tested-by: NKrzysztof Piotr Oledzki <ole@ans.pl>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

RFC 1122 specifies two threshold values R1 and R2 for connection timeouts,
which may represent a number of allowed retransmissions or a timeout value.
Currently linux uses sysctl_tcp_retries{1,2} to specify the thresholds
in number of allowed retransmissions.

For any desired threshold R2 (by means of time) one can specify tcp_retries2
(by means of number of retransmissions) such that TCP will not time out
earlier than R2. This is the case, because the RTO schedule follows a fixed
pattern, namely exponential backoff.

However, the RTO behaviour is not predictable any more if RTO backoffs can be
reverted, as it is the case in the draft
"Make TCP more Robust to Long Connectivity Disruptions"
(http://tools.ietf.org/html/draft-zimmermann-tcp-lcd).

In the worst case TCP would time out a connection after 3.2 seconds, if the
initial RTO equaled MIN_RTO and each backoff has been reverted.

This patch introduces a function retransmits_timed_out(N),
which calculates the timeout of a TCP connection, assuming an initial
RTO of MIN_RTO and N unsuccessful, exponentially backed-off retransmissions.

Whenever timeout decisions are made by comparing the retransmission counter
to some value N, this function can be used, instead.

The meaning of tcp_retries2 will be changed, as many more RTO retransmissions
can occur than the value indicates. However, it yields a timeout which is
similar to the one of an unpatched, exponentially backing off TCP in the same
scenario. As no application could rely on an RTO greater than MIN_RTO, there
should be no risk of a regression.
Signed-off-by: NDamian Lukowski <damian@tvk.rwth-aachen.de>
Acked-by: NIlpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Here, an ICMP host/network unreachable message, whose payload fits to
TCP's SND.UNA, is taken as an indication that the RTO retransmission has
not been lost due to congestion, but because of a route failure
somewhere along the path.
With true congestion, a router won't trigger such a message and the
patched TCP will operate as standard TCP.

This patch reverts one RTO backoff, if an ICMP host/network unreachable
message, whose payload fits to TCP's SND.UNA, arrives.
Based on the new RTO, the retransmission timer is reset to reflect the
remaining time, or - if the revert clocked out the timer - a retransmission
is sent out immediately.
Backoffs are only reverted, if TCP is in RTO loss recovery, i.e. if
there have been retransmissions and reversible backoffs, already.

Changes from v2:
1) Renaming of skb in tcp_v4_err() moved to another patch.
2) Reintroduced tcp_bound_rto() and __tcp_set_rto().
3) Fixed code comments.
Signed-off-by: NDamian Lukowski <damian@tvk.rwth-aachen.de>
Acked-by: NIlpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This supplementary patch renames skb to icmp_skb in tcp_v4_err() in order to
disambiguate from another sk_buff variable, which will be introduced
in a separate patch.
Signed-off-by: NDamian Lukowski <damian@tvk.rwth-aachen.de>
Acked-by: NIlpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add support for dcbnl_rtnl_ops.setapp/getapp to set or get the current user
priority bitmap for the given application protocol. Currently, 82599 only
supports setapp/getapp for Fiber Channel over Ethernet (FCoE) protocol.
Signed-off-by: NYi Zou <yi.zou@intel.com>
Acked-by: NPeter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Implements the dcbnl netlink setapp/getapp pair. When a setapp/getapp
is received, dcbnl would just pass on to dcbnl_rtnl_op.setapp/getapp
that are supposed to be implemented by the low level drivers.
Signed-off-by: NYi Zou <yi.zou@intel.com>
Acked-by: NPeter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>