netfilter: nfnetlink_log: fix silly refcount leak

Quick fix for memory/module refcount leak.
Reference count of listener instance never reaches 0.

Start/stop of ulogd2 is enough to trigger this bug!

Now, refcounting there looks very fishy in particular this code:

  if (!try_module_get(THIS_MODULE)) {
...

and creation of listener instance with refcount 2,
so it may very well be ripped and redone.  :-)

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>

IPVS: ip_vs_lblcr: use list headA

Use list_head rather than a custom list implementation.

Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Patrick McHardy <kaber@trash.net>

netfilter: xtables: reduce arguments to translate_table

Just pass in the entire repl struct. In case of a new table (e.g.
ip6t_register_table), the repldata has been previously filled with
table->name and table->size already (in ip6t_alloc_initial_table).

Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
Signed-off-by: Patrick McHardy <kaber@trash.net>

netfilter: xtables: optimize call flow around xt_ematch_foreach

Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
Signed-off-by: Patrick McHardy <kaber@trash.net>

netfilter: xtables: replace XT_MATCH_ITERATE macro

The macro is replaced by a list.h-like foreach loop. This makes
the code more inspectable.

Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
Signed-off-by: Patrick McHardy <kaber@trash.net>

netfilter: xtables: optimize call flow around xt_entry_foreach

Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
Signed-off-by: Patrick McHardy <kaber@trash.net>

netfilter: xtables: replace XT_ENTRY_ITERATE macro

The macro is replaced by a list.h-like foreach loop. This makes
the code much more inspectable.

Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
Signed-off-by: Patrick McHardy <kaber@trash.net>

netfilter: xt_recent: fix false match

A rule with a zero hit_count will always match.

Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Cc: stable@kernel.org
Signed-off-by: Patrick McHardy <kaber@trash.net>

netfilter: xt_recent: fix buffer overflow

e->index overflows e->stamps[] every ip_pkt_list_tot packets.

Consider the case when ip_pkt_list_tot==1; the first packet received is stored
in e->stamps[0] and e->index is initialized to 1. The next received packet
timestamp is then stored at e->stamps[1] in recent_entry_update(),
a buffer overflow because the maximum e->stamps[] index is 0.

Signed-off-by: Tim Gardner <tim.gardner@canonical.com>
Cc: stable@kernel.org
Signed-off-by: Patrick McHardy <kaber@trash.net>

netfilter: nf_conntrack_reasm: properly handle packets fragmented into a single fragment

When an ICMPV6_PKT_TOOBIG message is received with a MTU below 1280,
all further packets include a fragment header.

Unlike regular defragmentation, conntrack also needs to "reassemble"
those fragments in order to obtain a packet without the fragment
header for connection tracking. Currently nf_conntrack_reasm checks
whether a fragment has either IP6_MF set or an offset != 0, which
makes it ignore those fragments.

Remove the invalid check and make reassembly handle fragment queues
containing only a single fragment.

Reported-and-tested-by: Ulrich Weber <uweber@astaro.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>

netfilter: nf_queue: fix NF_STOLEN skb leak

commit 3bc38712e3a6e059 (handle NF_STOP and unknown verdicts in
nf_reinject) was a partial fix to packet leaks.

If user asks NF_STOLEN status, we must free the skb as well.

Reported-by: Afi Gjermund <afigjermund@gmail.com>
Signed-off-by: Eric DUmazet <eric.dumazet@gmail.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>

netfilter: ctnetlink: fix creation of conntrack with helpers

This patch fixes a bug that triggers an assertion if you create
a conntrack entry with a helper and netfilter debugging is enabled.
Basically, we hit the assertion because the confirmation flag is
set before the conntrack extensions are added. To fix this, we
move the extension addition before the aforementioned flag is
set.

This patch also removes the possibility of setting a helper for
existing conntracks. This operation would also trigger the
assertion since we are not allowed to add new extensions for
existing conntracks. We know noone that could benefit from
this operation sanely.

Thanks to Eric Dumazet for initial posting a preliminary patch
to address this issue.

Reported-by: David Ramblewski <David.Ramblewski@atosorigin.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>

netfilter: restore POST_ROUTING hook in NF_HOOK_COND

Commit 2249065 ("netfilter: get rid of the grossness in netfilter.h")
inverted the logic for conditional hook invocation, breaking the
POST_ROUTING hook invoked by ip_output().

Correct the logic and remove an unnecessary initialization.

Reported-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>

netfilter: nf_defrag_ipv4: fix compilation error with NF_CONNTRACK=n

As reported by Randy Dunlap <randy.dunlap@oracle.com>, compilation
of nf_defrag_ipv4 fails with:

include/net/netfilter/nf_conntrack.h:94: error: field 'ct_general' has incomplete type
include/net/netfilter/nf_conntrack.h:178: error: 'const struct sk_buff' has no member named 'nfct'
include/net/netfilter/nf_conntrack.h:185: error: implicit declaration of function 'nf_conntrack_put'
include/net/netfilter/nf_conntrack.h:294: error: 'const struct sk_buff' has no member named 'nfct'
net/ipv4/netfilter/nf_defrag_ipv4.c:45: error: 'struct sk_buff' has no member named 'nfct'
net/ipv4/netfilter/nf_defrag_ipv4.c:46: error: 'struct sk_buff' has no member named 'nfct'

net/nf_conntrack.h must not be included with NF_CONNTRACK=n, add a
few #ifdefs. Long term the header file should be fixed to be usable
even with NF_CONNTRACK=n.

Tested-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>

ipvs: SCTP Trasport Loadbalancing Support

Enhance IPVS to load balance SCTP transport protocol packets. This is done
based on the SCTP rfc 4960. All possible control chunks have been taken
care. The state machine used in this code looks some what lengthy. I tried
to make the state machine easy to understand.

Signed-off-by: Venkata Mohan Reddy Koppula <mohanreddykv@gmail.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Patrick McHardy <kaber@trash.net>

Merge branch 'ebt_config_compat_v4' of git://git.breakpoint.cc/fw/nf-next-2.6

IPv6: convert mc_lock to spinlock

Only used for writing, so convert to spinlock

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/tehuti.c: Reapply use DEFINE_PCI_DEVICE_TABLE()

Commit 865a21a5e3d1b384c559a44c898fcad93e187b82 overwrote
commit a3aa18842a5303fc28fcc4d57dbd16618bd830a0

Fix it.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/yellowfin.c: Use (pr|netdev)_<level> macro helpers

Add #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Remove #define PFX
Use pr_<level>
Use netdev_<level>
Convert formats like %8.8 to %08
Remove periods from formats
Coalesce long formats

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/sky2.c: Use (pr|netdev)_<level> macro helpers

Add #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Remove #define PFX
Use pr_<level>
Use netdev_<level>

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/skge.c: Use (pr|netdev)_<level> macro helpers

Add #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Remove #define PFX
Use pr_<level>
Use netdev_<level>
Checkpatch cleaning

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/skge.c: Use (pr|netdev)_<level> macro helpers

Add #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Remove #define PFX
Use pr_<level>
Use netdev_<level>

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/pci-skeleton.c: Use (pr|netdev|netif)_<level> macro helpers

Add #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Remove #define PFX
Use pr_<level>
Use netdev_<level>
Use netif_<level>
Checkpatch cleaning
Convert formats like 0x%08x to %#08x
Remove periods from formats
Coalesce long formats
Use print_hex_dump

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/cnic.c: Use (pr|netdev)_<level> macro helpers

Add #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Remove #define PFX
Use pr_<level>
Use netdev_<level>
Remove periods from formats
Coalesce long formats
Use __func__

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/cassini.c: Use (pr|netdev|netif)_<level> macro helpers

Add #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Remove #define PFX
Use pr_<level>
Use netdev_<level>
Use netif_<level>
Remove periods from formats
Coalesce long formats

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/bnx2x: Use (pr|netdev|netif)_<level> macro helpers

Add #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Convert struct bnx2x member msglevel to msg_enable for netif_msg_<foo> macros
Remove #define PFX
Use pr_<level>
Use netdev_<level>
Use netif_<level>
Coalesce long formats

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/bnx2.c: Use (pr|netdev|netif)_<level> macro helpers

Add #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Remove #define PFX
Use pr_<level>
Use netdev_<level>
Use netif_<level>
Remove periods from formats
Coalesce long formats
Coalesce some printks

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/b44.c: Use (pr|netdev|netif)_<level> macro helpers

Add #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Remove #define PFX
Use pr_<level>
Use netdev_<level>
Use netif_<level>
Remove periods from formats
Coalesce long formats

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/8139too.c: Use (pr|netdev|netif)_<level> macro helpers

Add #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Remove #define PFX
Use pr_<level>
Use netdev_<level>
Use netif_<level>
Convert formats like %8.8 to %08
Remove periods from formats
Coalesce long formats
Use print_hex_dump

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/8139cp.c: Use (pr|netdev|netif)_<level> macro helpers

Add #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Remove #define PFX
Use pr_<level>
Use netdev_<level>
Use netif_<level>
Remove periods from formats
Coalesce long formats

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Update version to 3.108

This patch updates the tg3 version to 3.108.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Push phylib definitions to phylib

This patch pushes phylib definitions out to phylib headers.  For phy
IDs, this removes some code duplication.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Rename tg3 phy ID preprocessor definitions

The phylib presents the phy ID in a different format than the one tg3
has traditionally used.  To highlight the distinction, this patch
prepends the tg3 native phy ID format with TG3.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Reformat SSID to phy ID table

This patch reformats the SSID to phy ID table, replacing constants with
preprocessor definitions.  This patch is also prep work for the
following patch, which will push line lengths beyond 80 chars.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Discover phy address once

The phy address will not change after it has been identified.  Move the
discovery code to a location that only gets executed once.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Reduce indent level of tg3_rx_prodring_alloc

This patch adds an inverted "jumbo ring enable" test and jumps to the exit
if it succeeds.  The change reduces the indent level of the remaining
code making it more readable.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Rename TG3_FLG3_RGMII_STD_IBND_DISABLE

The STD part of this preprocessor definition is a bit of a misnomer.
This flag is a coarse control of the RGMII inband status facilities.
This patch renames the definition to be more accurate.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Add more partno entries for fallback path

This patch adds 57765 asic partno entries for the path executed if VPD
is not present in NVRAM.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Allow phylib flowctrl changes anytime

This patch loosens the restriction that the phylib interface must be up
and running to change the flow control parameters.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Add support for 2 new selfboot formats

This patch adds new offsets to the bootcode version extraction code to
support NVRAM format versions 4 and 5.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Make 57791 and 57795 10/100 only

This patch adds the 57791 and 57795 to the list of devices that only
support 10 and 100 Mbps speeds.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Enforce DMA mapping / skb assignment ordering

Michael Chan noted that there is nothing in the code that would prevent
the compiler from delaying the access of the "mapping" member of the
newly arrived packet until much later.  If this happened after the
skb = NULL assignment, it is possible for the driver to pass a bad
dma_addr value to pci_unmap_single().  To enforce this ordering, we need
a write memory barrier.  The pairing read memory barrier already exists
in tg3_rx_prodring_xfer() under the comments starting with
"Ensure that updates to the...".

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tun: socket filter support

This patch adds Linux Socket Filter support to
tun driver.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: export attach/detach filter routines

Export sk_attach_filter/sk_detach_filter routines,
so that tun module can use them.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

chelsio: convert to use netdev_for_each_mc_addr

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

cxgb3: convert to use netdev_for_each_addr

Removed whole t3_rx_mode structure and appropriate helpers cause they are no
longer needed.

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

smsc911x: replace manual phy lookup.

Use phy_find_first() function instead of manual lookup.

Signed-off-by: Denis Kirjanov <kirjanov@gmail.com>
Reviewed-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: bug fix for vlan + gro issue

Traffic (tcp) doesnot start on a vlan interface when gro is enabled.
Even the tcp handshake was not taking place.
This is because, the eth_type_trans call before the netif_receive_skb
in napi_gro_finish() resets the skb->dev to napi->dev from the previously
set vlan netdev interface. This causes the ip_route_input to drop the
incoming packet considering it as a packet coming from a martian source.

I could repro this on 2.6.32.7 (stable) and 2.6.33-rc7.
With this fix, the traffic starts and the test runs fine on both vlan
and non-vlan interfaces.

CC: Herbert Xu <herbert@gondor.apana.org.au>
CC: Patrick McHardy <kaber@trash.net>
Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

mv643xx_eth: fix missing validate_addr hook

Fix missing validate_addr hook.

Signed-off-by: Denis Kirjanov <kirjanov@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: Add Aeroflex Gaisler 10/100/1G Ethernet MAC driver

Adds device driver for Aeroflex Gaisler 10/100 and 10/100/1G Ethernet
MAC IP cores.

Signed-off-by: Kristoffer Glembo <kristoffer@gaisler.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

via-velocity: Enable scatter/gather IO by default

Reduces CPU utilization significantly with sendfile for example.

Signed-off-by: Simon Kagstrom <simon.kagstrom@netinsight.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipg: Remove device claimed by dl2k from pci id table

This patch removes D-Link DGE-550T PCI ID (1186:4000) from the ipg
driver. The ipg driver is for IP2000-based cards and the DGE-550T is
a DL2000-based card. The driver loads and works for a few moments, but
once a real workload is applied it stops operating. The ipg driver
claimed this ID since it was introduced in 2.6.24 and it's forced many
users to blacklist it.

The correct driver for this hardware is the dl2k driver, which has been
claiming this PCI ID since the 2.4 days.

Signed-off-by: Jeff Mahoney <jeffm@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

xfrm: Revert false event eliding commits.

As reported by Alexey Dobriyan:

--------------------
setkey now takes several seconds to run this simple script
and it spits "recv: Resource temporarily unavailable" messages.

#!/usr/sbin/setkey -f
flush;
spdflush;

add A B ipcomp 44 -m tunnel -C deflate;
add B A ipcomp 45 -m tunnel -C deflate;

spdadd A B any -P in ipsec
        ipcomp/tunnel/192.168.1.2-192.168.1.3/use;
spdadd B A any -P out ipsec
        ipcomp/tunnel/192.168.1.3-192.168.1.2/use;
--------------------

Obviously applications want the events even when the table
is empty.  So we cannot make this behavioral change.

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

SiS190/191 half-duplex initialization fix

Adds half-duplex specific setup code (taken from SiS own GPL driver).
Without those, half-duplex connections are very unreliable, often
working on small transfers and failing after a while.

Signed-off-by: Riccardo Ghetta <birrachiara@tin.it>
Signed-off-by: David S. Miller <davem@davemloft.net>

ethtool: Don't flush n-tuple list from ethtool_reset()

The n-tuple list should be flushed if and only if the ETH_RESET_FILTER
flag is set and the driver is able to reset filtering/flow direction
hardware without also resetting a component whose flag is not set.
This test is best left to the driver.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: version number and date

Updated release version and date: 1.52.1-6 and 2010/02/16

Signed-off-by: Vladislav Zolotarov <vladz@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: Use firmware 5.2.13

Switch to the new firmware version (5.2.13).

Signed-off-by: Vladislav Zolotarov <vladz@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: FW 5.2.13

New firmware version (5.2.13).

Signed-off-by: Vladislav Zolotarov <vladz@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: fix in 57710 self-test

Bug fix: Use the last unicast entry in CAM in bnx2x_test_int(). Relevant for 57710 only.

Author: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Vladislav Zolotarov <vladz@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: Do not allow enabling LRO if disable_tpa=1

Bug fix: Do not allow enabling LRO if disable_tpa=1.

Signed-off-by: Vladislav Zolotarov <vladz@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: clean up in case of error in bnx2x_init_hw()

Bug fix: clean up MCP state in case of error in bnx2x_init_hw().

Author: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Vladislav Zolotarov <vladz@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: Properly release allocated MSI-X/MSI vectors

Bug fix: Properly release allocated MSI-X/MSI vectors if ifup failed
due to lack of memory.

Signed-off-by: Vladislav Zolotarov <vladz@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/usb: Use netif_<level> logging facilities

Convert from:
if (netif_msg_<foo>(priv))
dev_<level>(dev...
to
netif_<level>(priv, foo, dev...

Also convert a few:

if (i < REG_TIMEOUT) {
etc...
return ret;
       }
to
if (i >= REG_TIMEOUT)
goto fail;
etc...
return ret;

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

usbnet: Convert dev(dbg|err|warn|info) macros to netdev_<level>

These macros are too similar to the dev_<level> equivalents
but take a usbnet * argument.  Convert them to the recently
introduced netdev_<level> macros and remove the old macros.

The old macros had "\n" appended to the format string.
Add the "\n" to the converted uses.

Some existing uses of the dev<foo> macros in cdc_eem.c
probably mistakenly had trailing "\n".  No "\n" added there.

Fix net1080 this/other log message inversion.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: implement pci shutdown handler

Signed-off-by: Sathya Perla <sathyap@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: fix rx-path to ignore a flush completion

The flush compl (compl with numfrags == 0; no data) is rcvd
from hw to indicate completion of RXQ destory operation. Fix
the RX path to not process it as RX data.

Signed-off-by: Sathya Perla <sathyap@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: don't rearm mcc cq when device is not open

When an MCC cmd is issued (via a netdev/ethtool op)
while the device is not open, the MCC CQ gets processed but the EQ
is not processed (as isr is not registered.) This can cause the EQ
to become full. So, while the device is not open, CQ must not be re-armed
to prevent EQ entries.

Signed-off-by: Sathya Perla <sathyap@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: a mini optimization in rx_compl_process() code

Introduce unlikely() for skb alloc failure and vlanf checks...

Signed-off-by: Sathya Perla <sathyap@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlge: Fix occasional loopback test failure.

On some servers we see the cleaning of the RX queue finish before all
the loopback packets are sent out.  This delay allows the queues to
settle before checking for successful completion.
Also, delay completion so link has time to come back up.

Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlge: Fix bonding mac address bug.

Use local copy of current mac address when initializing.  In bonding
testing we saw cases where dev_addr was out of data causing failover
errors.

Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlge: Fix dropping of large non-TCP/UDP frames.

This was found with pings that were large enough to span multiple rx
buffers.  The split frame logic handles this case nicely.

Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ll_temac: Fix MAC address configuration from userland

A userland command to set the LLTEMAC MAC address,
i.e. "ifconfig eth0 hw addr xx:yy:zz:pp:dd:qq",
results in a device address of 00:01:xx:yy:zz:pp.
Correct this.

Signed-off-by: Steven J. Magnani <steve@digidescorp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ll_temac: Add support for V2 LLTEMAC core

LLTEMAC V1 cores place only received packet length in the app4 word.
V2 cores place additional information in app4.
Mask out the additional information when retrieving the packet length.

Signed-off-by: Steven J. Magnani <steve@digidescorp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: convert multiple drivers to use netdev_for_each_mc_addr

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

cassini: convert to use netdev_for_each_mc_addr

Introduced a new function to do the mc list processing, makes code
clearer after transition.

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: use kasprintf() for socket cache names

kasprintf() makes code smaller.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

xt_hashlimit: fix locking

Commit 2eff25c18c3d332d3c4dd98f2ac9b7114e9771b0
(netfilter: xt_hashlimit: fix race condition and simplify locking)
added a mutex deadlock :
htable_create() is called with hashlimit_mutex already locked

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipmr: remove useless checks from ipmr_device_event

The net being checked there is dev_net(dev) and thus this if
is always false.

Fits both net and net-next trees.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

igb: update tx DMA mapping error handling

This updates the tx DMA mapping error handling code to
resemble e1000e/ixgbe.

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>

igb: change descriptor control thresholds

This change simplifies the code by setting RX_PTHRESH to 8 for
all devices, as it was unlikely that there was any advantage to
set it at 16 for earlier cards. Additionally TX_WTHRESH is
set to 1 for the 82576 NIC to improve performance by enabling
a minimal amount of write combining when writing back descriptors.

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>

igb: use igb_free_q_vectors to cleanup failure in igb_alloc_q_vectors

This change makes it so that igb_free_q_vectors is reused in
igb_alloc_q_vectors to handle the cleanup instead of unwinding
through the allocations.

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>

igb: remove unused vmolr value

In an earlier version of igb_write_mc_addr_list() the vmolr register
was modified. This register is no longer accessed, although the variable
still exists. This patch removes it from the function.

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>

igb: minor type cleanups

This change cleans up some instances where unsigned int and u32
were being used interchangeably, and cleans up hdr_len which
was being set to 0 twice.

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>

igb: move gso_segs into buffer_info structure

This change moves gso_segs into the buffer_info structure to avoid
a possible cache line miss in clean_tx_irq.

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>

igb: inline igb_maybe_stop_tx

igb_maybe_stop_tx() is extremely small and appears in several spots in
the tx hotpath. This change inlines the function for a possible
performance boost.

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>

igb: only read phy specific stats if in internal phy mode

There are a couple statistics registers that are not meant to
be read when in SGMII/serdes mode. This patch adds a check to
verify mode before reading and updating these statistics.

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>

igb: only support SRRCTL_DROP_EN when using multiple queues

The SRRCTL.DROP_EN bit should only be set when we are supporting
multiple queues. This bit is meant to prevent head of line blocking
and is unnecessary in the single queue case.

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>

igb: cap interrupts at 20K per queue when in itr mode 3

In order to maintain similar performance between MSI-X and
legacy/MSI interrupts, this patch reduces the number of interrupts
when receiving small packets to 20K when in interrupt throttle
rate mode 3.

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>

igb: remove adaptive IFS from driver

Adaptive IFS support has been included in the igb driver since its
initial release, but it is not a feature on any igb NICs. This patch
removes it from the driver.

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>

igb: Allocate rings seperately instead of as a block

This change makes it so that the rings are allocated seperately.  As a
result we can allocate them on seperate nodes at some point in the future
if we so desire.

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>

igb: call pci_save_state after pci_restore_state

This patch adds calls to pci_save_state() immediately after
calls to pci_restore_state(). Due to a change in the behavior
of pci_restore_state() it is necessary to call pci_save_state()
to keep the state_saved flag. This patch is based on a similar
patch for ixgbe.

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>

igb: Power down link when interface is down

This changes the behavior of the driver to power down the link
when the associated interface is down, unless management is enabled.

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>

igb: ignore EEPROM APME check when shutting down serdes link

Checking the EEPROM APME state unnecessarily prevents the link from
shutting down. The standard power down routines should be
sufficient to determine whether the serdes link can power down
when going into D3.

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>

igb: Report link status in ethtool when interface is down

With this change ethtool will correctly report link status when
the interface is down. Currently ethtool reports the link as not
detected when the interface is down.

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>

igb: add support for wake-on-link

This adds support for wake-on-link/phy activity to the ethtool
interface.

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>

igb: remove unecessary q_vector declarations and remove itr_shift

This change removes the use of itr_shift since a mac type call can be
used just as easily to identify the only HW that needs to have the itr
shifted.

In addition it removes two unecessary declarations of a q_vector
pointer from the initialization path.

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>

ipv6.h: reassembly: replace calculated magic number with multiplication

On Tue, 2010-02-16 at 16:47 +0100, Patrick McHardy wrote:
> Joe Perches wrote:
> >> @@ -246,6 +246,8 @@ extern int ipv6_opt_accepted(struct sock *sk, struct sk_buff *skb);
> >>  int ip6_frag_nqueues(struct net *net);
> >>  int ip6_frag_mem(struct net *net);
> >>
> >> +#define IPV6_FRAG_HIGH_THRESH 262144 /* == 256*1024 */
> >> +#define IPV6_FRAG_LOW_THRESH 196608 /* == 192*1024 */
> >>  #define IPV6_FRAG_TIMEOUT (60*HZ) /* 60 seconds */
> >
> > 196608 isn't a number I want to remember.
> > Is this better as:
> >
> > #define IPV6_FRAG_HIGH_THRESH (256 * 1024) /* 262144 */
> > #define IPV6_FRAG_LOW_THRESH (192 * 1024) /* 196608 */
>
> Please send a patch, I'll apply it once these patches are in Dave's
> tree.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: remove INIT_RCU_HEAD() usage

call_rcu() will unconditionally reinitialize RCU head anyway.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

b44: Ratelimit timeout error message.

Signed-off-by: Jochen Friedrich <jochen@scram.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

percpu: add __percpu sparse annotations to net drivers

Add __percpu sparse annotations to net drivers.

These annotations are to make sparse consider percpu variables to be
in a different address space and warn if accessed without going
through percpu accessors.  This patch doesn't affect normal builds.

Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: David S. Miller <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>