netfilter: nf_conntrack: add allocation flag to nf_conntrack_alloc

ctnetlink does not need to allocate the conntrack entries with GFP_ATOMIC
as its code is executed in user context.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

netfilter: Get rid of refrences to no longer existant Fast NAT.

Get rid of refrences to no longer existant Fast NAT.

IP_ROUTE_NAT support was removed in August of 2004, but references to Fast
NAT were left in a couple of config options.

Signed-off-by: Russ Dill <Russ.Dill@gmail.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

netfilter: use correct namespace in ip6table_security

Signed-off-by: Alexey Dobriyan <adobriyan@parallels.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

amd8111e: delete non NAPI code from the driver.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Acked-by: Jeff Garzik <jgarzik@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

starfire: delete non NAPI code from the driver.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Acked-by: Jeff Garzik <jgarzik@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

netdev: remove unused S2IO_NAPI

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Acked-by: Jeff Garzik <jgarzik@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

netdrv: don't truncate VLAN TCI with VLAN stripping

The vlan_hwaccel_{rx,receive_skb} functions expect the full TCI field
for priority mappings, don't truncate the upper 4 bits.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Acked-by: Jeff Garzik <jgarzik@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: fix dev_set_promiscuity() breakage

Commit dad9b335 (netdevice: Fix promiscuity and allmulti overflow) broke
dev_set_promiscuity() by returning on success without reprogramming the
device.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

net-sched: consolidate default fifo qdisc setup

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

net-sched: sch_htb: remove write-only qdisc filter_cnt

The filter_cnt is supposed to count filter references to a class.
Since the qdisc can't be the target of a filter, it doesn't need
a filter_cnt. In fact the counter is never decreased since cls_api
considers a return value of zero a failure and doesn't unbind again.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

net-sched: sch_htb: remove child and sibling lists

Now that the qdisc isn't destroyed in hierarchical order anymore,
the only user of the child lists left is htb_parent_last_child().
This can be easily changed to use a counter of children to save
a few bytes.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

net-sched: sch_htb: use dynamic class hash helpers

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

net-sched: sch_htb: move hash and sibling list removal to htb_delete

Hash list removal currently happens twice (once in htb_delete, once
in htb_destroy_class), which makes it harder to use the dynamically
sized class hash without adding special cases for HTB. The reason is
that qdisc destruction destroys classes in hierarchical order, which
is not necessary if filters are destroyed in a separate iteration
during qdisc destruction.

Adjust qdisc destruction to follow the same scheme as other hierarchical
qdiscs by first performing a filter destruction pass, then destroying
all classes in hash order.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

net-sched: sch_cbq: use dynamic class hash helpers

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

net-sched: sch_hfsc: use dynamic class hash helpers

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

net-sched: add dynamically sized qdisc class hash helpers

Currently all qdiscs which allow to create classes uses a fixed sized hash
table with size 16 to hash the classes. This causes a large bottleneck
when using thousands of classes and unbound filters.

Add helpers for dynamically sized class hashes to fix this. The following
patches will convert the qdiscs to use them.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'master' of /linux/kernel/git/davem/net-2.6

Conflicts:

Documentation/feature-removal-schedule.txt
drivers/net/wan/hdlc_fr.c
drivers/net/wireless/iwlwifi/iwl-4965.c
drivers/net/wireless/iwlwifi/iwl3945-base.c

Merge branch 'master' of /linux/kernel/git/linville/wireless-next-2.6

vlan: Add GVRP support

Add GVRP support for dynamically registering VLANs with switches.

By default GVRP is disabled because we only support the applicant-only
participant model, which means it should not be enabled on vlans that
are members of a bridge. Since there is currently no way to cleanly
determine that, the user is responsible for enabling it.

The code is pretty small and low impact, its wrapped in a config
option though because it depends on the GARP implementation and
the STP core.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

vlan: Move device unregistration before lower dev cleanup

Move the unregister_netdevice() call for the VLAN device before cleanup
for the lower device. This is needed by GVRP so it can send a leave
message before the applicant on the lower device is cleaned up.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

vlan: Change vlan_dev_set_vlan_flag() to handle multiple flags at once

Change vlan_dev_set_vlan_flag() to handle multiple flags at once and
rename to vlan_dev_change_flags(). This allows to to use it from the
netlink interface, which in turn allows to handle necessary adjustments
when changing flags centrally.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: Add GARP applicant-only participant

Add an implementation of the GARP (Generic Attribute Registration Protocol)
applicant-only participant. This will be used by the following patch to
add GVRP support to the VLAN code.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

bridge: Use STP demux

Use the STP demux layer for receiving STP PDUs instead of directly
registering with LLC.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: Add STP demux layer

Add small STP demux layer for demuxing STP PDUs based on MAC address.
This is needed to run both GARP and STP in parallel (or even load the
modules) since both use LLC_SAP_BSPAN.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

MIB: add struct net to UDP6_INC_STATS_BH

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

MIB: add struct net to UDP6_INC_STATS_USER

As simple as the patch #1 in this set.

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

MIB: add struct net to UDP_INC_STATS_BH

Two special cases here - one is rxrpc - I put init_net there
explicitly, since we haven't touched this part yet. The second
place is in __udp4_lib_rcv - we already have a struct net there,
but I have to move its initialization above to make it ready
at the "drop" label.

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

MIB: add struct net to UDP_INC_STATS_USER

Nothing special - all the places already have a struct sock
at hands, so use the sock_net() net.

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

Merge branch 'davem-next' of /linux/kernel/git/jgarzik/netdev-2.6

netns: selective flush of rt_cache

dst cache is marked as expired on the per/namespace basis by previous
path. Right now we have to implement selective cache shrinking. This
procedure has been ported from older OpenVz codebase.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

netns: place rt_genid into struct net

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv4: pass current value of rt_genid into rt_hash

Basically, there is no difference to atomic_read internally or pass it as
a parameter as rt_hash is inline.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

netns: add struct net parameter to rt_cache_invalidate

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

netns: make rt_secret_rebuild timer per namespace

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

netns: register net.ipv4.route.flush in each namespace

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv4: remove static flush_delay variable

flush delay is used as an external storage for net.ipv4.route.flush sysctl
entry. It is write-only.

The ctl_table->data for this entry is used once. Fix this case to point
to the stack to remove global variable. Do this to avoid additional
variable on struct net in the next patch.

Possible race (as it was before) accessing this local variable is removed
using flush_mutex.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: add fib_rules_ops to flush_cache method

This is required to pass namespace context into rt_cache_flush called from
->flush_cache.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

netns: add namespace parameter to rt_cache_flush

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

8390: Split 8390 support into a pausing and a non pausing driver core

Only a few ISA controllers need the pausing version of the 8390 core
while PCMCIA, later ISA and PCI do not. More importantly the ISA delays
can break non ISA boxes so we must use a different build of 8390.c for
the two sets of controllers.

No changes since last time as all the points of concerns raised proved to
be invalid

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

WAN: Remove no longer needed pvc_desc in FR code.

Remove no longer needed struct pvc_desc in FR code.
Requires "WAN: convert drivers to use built-in netdev_stats" patch.

Signed-off-by: Krzysztof Hałasa <khc@pm.waw.pl>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

WAN: convert drivers to use built-in netdev_stats

There is no point in using separate net_device_stats structs when
the one in struct net_device is present. Compiles.

Signed-off-by: Krzysztof Hałasa <khc@pm.waw.pl>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

igb: add NAPI Rx queue support

Update the NAPI implementation to use the new napi_struct infrstructure.
This removes the need of multiple net_device structs to implement a
multiqueue NAPI.

Signed-off-by: PJ Waskiewicz <peter.p.waskiewicz.jr.@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

igb: fix init on 82575 with MNG enabled

This patch resolves an issue seen on 82575 adapters with managability
pass-thru enabled, which could cause the system to panic.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

igb: eliminate hw from the hw_dbg macro arguments

Various cosmetic cleanups. Comment fixes. Eliminate the hw part out
of the hw_dbg macro since it's always used.

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

igb: fix parameter options

Currently there are items in the driver to support parameters, but there
are no parameters being added to the driver.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

igb: cleanup function header comments

Function header comments do not match function name.

Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

igb: Remove adapter struct from these function call parameters

Simplify a bunch of calls to these functions by moving the adapter
struct out of the function arguments since that is not needed.

Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

igb: limit EEPROM access

EEPROM access beyond the base eeprom page is unsupported.

Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

ipg: run-time configurable jumbo frame support

Make jumbo frame support configurable via ifconfig mtu option as suggested by
Stephen Hemminger.

Cc: Stephen Hemminger <stephen.hemminger@vyatta.com>
Tested-by: Andrew Savchenko <Bircoph@list.ru>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

ipg: per-device max_rxframe_size

Add a ->max_rxframe member to struct ipg_nic_private and convert the users of
IPG_MAX_RXFRAME_SIZE to use it instead to enable per-device jumbo frame
configuration.

Tested-by: Andrew Savchenko <Bircoph@list.ru>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

ipg: per-device rxsupport_size

Add a ->max_rxframe member to struct ipg_nic_private and convert the users of
IPG_RXSUPPORT_SIZE to use it instead to enable per-device jumbo frame
configuration.

Tested-by: Andrew Savchenko <Bircoph@list.ru>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

ipg: per-device rxfrag_size

Add a ->max_rxframe member to struct ipg_nic_private and convert the users of
IPG_RXFRAG_SIZE to use it instead to enable per-device jumbo frame
configuration.

Tested-by: Andrew Savchenko <Bircoph@list.ru>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

ipg: remove jumbo frame #ifdef from mtu

Remove JUMBO_FRAME #ifdef from dev->mtu setting in ipg_nic_open() so that we
can make IPG_TXFRAG_SIZE configurable.

Tested-by: Andrew Savchenko <Bircoph@list.ru>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

ipg: always compile in jumbo frame support

Add a ->is_jumbo boolean to struct ipg_nic_private and fix up
ipg_interrupt_handler() to call the jumbo frame version of ipg_nic_rx() if the
boolean is set to true. Also remove the JUMBO_FRAME #ifdefs so we can always
compile in support for jumbo frames.

Tested-by: Andrew Savchenko <Bircoph@list.ru>
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

Merge branch 'r8169-next' of git://git./linux/kernel/git/romieu/netdev-2.6 into upstream-next

fs_enet: restore promiscuous and multicast settings in restart()

The restart() function is called when the link state changes and resets
multicast and promiscuous settings. This patch restores those settings at the
end of restart().

Signed-off-by: Laurent Pinchart <laurentp@cse-semaphore.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

ibm_newemac: Fixes entry of short packets

Short packets has to be discarded by the driver. So this patch addresses the
issue of discarding the short packets of size lesser then ethernet header
size.

Signed-off-by: Sathya Narayanan <sathyan@teamf1.com>
Signed-off-by: Stefan Roese <sr@denx.de>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

ibm_newemac: Fixes kernel crashes when speed of cable connected changes

The descriptor pointers were not initialized to NIL values, so it was
poiniting to some random addresses which was completely invalid. This
fix takes care of initializing the descriptor to NIL values and clearing
the valid descriptors on clean ring operation.

Signed-off-by: Sathya Narayanan <sathyan@teamf1.com>
Signed-off-by: Stefan Roese <sr@denx.de>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

pasemi_mac: Access iph->tot_len with correct endianness

iph->tot_len is stored in network byte order, so access it using
ntohs().  This doesn't have any real world impact on pasemi_mac, since
the device only exists as part of a big-endian system-on-chip, but
fixing this gets rid of a sparse warning and avoids having a bad example
in the tree.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

ehea: Access iph->tot_len with correct endianness

iph->tot_len is stored in network byte order, so access it using
ntohs().  This doesn't have any real world impact on ehea, since ehea
only exists for big-endian platfroms (at the moment at least) but fixing
this gets rid of a sparse warning and avoids having a bad example in the
tree.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

ehea: fix race condition

When ehea_stop is called the function
cancel_work_sync(&port->reset_task) is used to ensure
that the reset task is not running anymore. We need an
additional flag to ensure that it can not be scheduled
after this call again for a certain time.

Signed-off-by: Jan-Bernd Themann <themann@de.ibm.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

ehea: add MODULE_DEVICE_TABLE

Required to allow distros to easily detect when ehea
module needs to be loaded

Signed-off-by: Jan-Bernd Themann <themann@de.ibm.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

ehea: fix might sleep problem

A mutex has to be replaced by spinlocks as it can be called from
a context which does not allow sleeping.
The kzalloc flag GFP_KERNEL has to be replaced by GFP_ATOMIC
for the same reason.

Signed-off-by: Jan-Bernd Themann <themann@de.ibm.com>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

forcedeth: fix lockdep warning on ethtool -s

After enabling CONFIG_LOCKDEP and CONFIG_PROVE_LOCKING I get the
following warning when ethtool -s is first called on one of the
forcedeth ports:

=================================
[ INFO: inconsistent lock state ]
2.6.26-rc4 #28
---------------------------------
inconsistent {in-hardirq-W} -> {hardirq-on-W} usage.
ethtool/1985 [HC0[0]:SC0[1]:HE1:SE0] takes:
 (&np->lock){++..}, at: [<ffffffffa000c5fd>] nv_set_settings+0xc8/0x3de [forcedeth]
{in-hardirq-W} state was registered at:
  [<ffffffffffffffff>] 0xffffffffffffffff
irq event stamp: 3606
hardirqs last  enabled at (3605): [<ffffffff8068106f>] _spin_unlock_irqrestore+0x3f/0x68
hardirqs last disabled at (3604): [<ffffffff80680d38>] _spin_lock_irqsave+0x13/0x46
softirqs last  enabled at (3534): [<ffffffff80246ba5>] __do_softirq+0xbc/0xc5
softirqs last disabled at (3606): [<ffffffff80680b33>] _spin_lock_bh+0x11/0x41

other info that might help us debug this:
2 locks held by ethtool/1985:
 #0:  (rtnl_mutex){--..}, at: [<ffffffff80596072>] rtnl_lock+0x12/0x14
 #1:  (_xmit_ETHER){-+..}, at: [<ffffffffa000c5e8>] nv_set_settings+0xb3/0x3de [forcedeth]
stack backtrace:
Pid: 1985, comm: ethtool Not tainted 2.6.26-rc4 #28
Call Trace:
 [<ffffffff8025f190>] print_usage_bug+0x162/0x173
 [<ffffffff8025fa8b>] mark_lock+0x231/0x41f
 [<ffffffff802607cf>] __lock_acquire+0x4e7/0xcac
 [<ffffffff8025fe64>] ? trace_hardirqs_on+0xf1/0x115
 [<ffffffff80272c3a>] ? disable_irq_nosync+0x6f/0x7b
 [<ffffffff80261375>] lock_acquire+0x55/0x6e
 [<ffffffffa000c5fd>] ? :forcedeth:nv_set_settings+0xc8/0x3de
 [<ffffffff80680b15>] _spin_lock+0x2f/0x3c
 [<ffffffffa000c5fd>] :forcedeth:nv_set_settings+0xc8/0x3de
 [<ffffffff8058f8bb>] dev_ethtool+0x186/0xea3
 [<ffffffff8067f446>] ? mutex_lock_nested+0x243/0x275
 [<ffffffff8025df2b>] ? debug_mutex_free_waiter+0x46/0x4a
 [<ffffffff8067f469>] ? mutex_lock_nested+0x266/0x275
 [<ffffffff8058e1ce>] dev_ioctl+0x4eb/0x600
 [<ffffffff8068106f>] ? _spin_unlock_irqrestore+0x3f/0x68
 [<ffffffff80580f91>] sock_ioctl+0x1f5/0x202
 [<ffffffff802a322e>] vfs_ioctl+0x2a/0x77
 [<ffffffff802a34d6>] do_vfs_ioctl+0x25b/0x270
 [<ffffffff806807b6>] ? trace_hardirqs_on_thunk+0x35/0x3a
 [<ffffffff802a352d>] sys_ioctl+0x42/0x65
 [<ffffffff8021fffb>] system_call_after_swapgs+0x7b/0x80

This is caused by the following snippet in nv_set_settings:

netif_carrier_off(dev);
if (netif_running(dev)) {
nv_disable_irq(dev);
netif_tx_lock_bh(dev);
spin_lock(&np->lock);
/* stop engines */
nv_stop_rxtx(dev);
spin_unlock(&np->lock);
netif_tx_unlock_bh(dev);
}

Because of nv_disable_irq this is probably not really a problem
though (I guess) and replacing the spin_lock with spin_lock_irqsave
could keep interrupts disabled for a longer period of time because
of delays in nv_stop_rx and nv_stop_tx.

Signed-off-by: Tobias Diedrich <ranma+kernel@tdiedrich.de>
Cc: Ayaz Abdulla <aabdulla@nvidia.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

Add missing skb->dev assignment in Frame Relay RX code

Commit 4c13eb6657fe9ef7b4dc8f1a405c902e9e5234e0 ([ETH]: Make
eth_type_trans set skb->dev like the other *_type_trans) removed
skb->dev assignment from hdlc_fr.c:fr_rx(). Unfortunately it was also
needed for cases other than eth_type_trans().

Adding it back.

It's quite serious and may be a security risk as it causes a wrong
input interface indication (the physical hdlcX instead of logical
pvcX). Probably -stable class fix.

Signed-off-by: Krzysztof Halasa <khc@pm.waw.pl>
Signed-off-by: Jeff Garzik <jgarzik@redhat.com>

bridge: fix use-after-free in br_cleanup_bridges()

Unregistering a bridge device may cause virtual devices stacked on the
bridge, like vlan or macvlan devices, to be unregistered as well.
br_cleanup_bridges() uses for_each_netdev_safe() to iterate over all
devices during cleanup. This is not enough however, if one of the
additionally unregistered devices is next in the list to the bridge
device, it will get freed as well and the iteration continues on
the freed element.

Restart iteration after each bridge device removal from the beginning to
fix this, similar to what rtnl_link_unregister() does.

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

tun: Allow GSO using virtio_net_hdr

Add a IFF_VNET_HDR flag.  This uses the same ABI as virtio_net
(ie. prepending struct virtio_net_hdr to packets) to indicate GSO and
checksum information.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Acked-by: Max Krasnyansky <maxk@qualcomm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tun: TUNSETFEATURES to set gso features.

ethtool is useful for setting (some) device fields, but it's
root-only.  Finer feature control is available through a tun-specific
ioctl.

(Includes Mark McLoughlin <markmc@redhat.com>'s fix to hold rtnl sem).

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Acked-by: Max Krasnyansky <maxk@qualcomm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tun: Interface to query tun/tap features.

The problem with introducing checksum offload and gso to tun is they
need to set dev->features to enable GSO and/or checksumming, which is
supposed to be done before register_netdevice(), ie. as part of
TUNSETIFF.

Unfortunately, TUNSETIFF has always just ignored flags it doesn't
understand, so there's no good way of detecting whether the kernel
supports new IFF_ flags.

This patch implements a TUNGETFEATURES ioctl which returns all the valid IFF
flags.  It could be extended later to include other features.

Here's an example program which uses it:

#include <linux/if_tun.h>
#include <sys/types.h>
#include <sys/ioctl.h>
#include <sys/stat.h>
#include <fcntl.h>
#include <err.h>
#include <stdio.h>

static struct {
unsigned int flag;
const char *name;
} known_flags[] = {
{ IFF_TUN, "TUN" },
{ IFF_TAP, "TAP" },
{ IFF_NO_PI, "NO_PI" },
{ IFF_ONE_QUEUE, "ONE_QUEUE" },
};

int main()
{
unsigned int features, i;

int netfd = open("/dev/net/tun", O_RDWR);
if (netfd < 0)
err(1, "Opening /dev/net/tun");

if (ioctl(netfd, TUNGETFEATURES, &features) != 0) {
printf("Kernel does not support TUNGETFEATURES, guessing\n");
features = (IFF_TUN|IFF_TAP|IFF_NO_PI|IFF_ONE_QUEUE);
}
printf("Available features are: ");
for (i = 0; i < sizeof(known_flags)/sizeof(known_flags[0]); i++) {
if (features & known_flags[i].flag) {
features &= ~known_flags[i].flag;
printf("%s ", known_flags[i].name);
}
}
if (features)
printf("(UNKNOWN %#x)", features);
printf("\n");
return 0;
}

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Acked-by: Max Krasnyansky <maxk@qualcomm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tcp: fix a size_t < 0 comparison in tcp_read_sock

<used> should be of type int (not size_t) since recv_actor can return
negative values and it is also used in a < 0 comparison.

Signed-off-by: Octavian Purdila <opurdila@ixiacom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tcp: net/ipv4/tcp.c needs linux/scatterlist.h

alpha:

net/ipv4/tcp.c: In function 'tcp_calc_md5_hash':
net/ipv4/tcp.c:2479: error: implicit declaration of function 'sg_init_table'    net/ipv4/tcp.c:2482: error: implicit declaration of function 'sg_set_buf'
net/ipv4/tcp.c:2507: error: implicit declaration of function 'sg_mark_end'

Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'net-next-2.6-v6ready-20080703' of git://git.linux-ipv6.org/gitroot/yoshfuji/linux-2.6-next

ipv4,ipv6 mroute: Add some helper inline functions to remove ugly ifdefs.

ip{,v6}_mroute_{set,get}sockopt() should not matter by optimization but
it would be better not to depend on optimization semantically.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>

ipv4: Do cleanup for ip_mr_init

Same as ip6_mr_init(), make ip_mr_init() return errno if fails.
But do not do error handling in inet_init(), just print a msg.

Signed-off-by: Wang Chen <wangchen@cn.fujitsu.com>
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>

ipv6: Do cleanup for ip6_mr_init.

If do not do it, we will get following issues:
1. Leaving junks after inet6_init failing halfway.
2. Leaving proc and notifier junks after ipv6 modules unloading.

Signed-off-by: Wang Chen <wangchen@cn.fujitsu.com>
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>

ipv6 route: Prefer outgoing interface with source address assigned.

Outgoing interface is selected by the route decision if unspecified.
Let's prefer routes via interface(s) with the address assigned if we
have multiple routes with same cost.
With help from Naohiro Ooiwa <nooiwa@miraclelinux.com>.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>

ipv6 addrconf: add accept_dad sysctl to control DAD operation.

- If 0, disable DAD.
- If 1, perform DAD (default).
- If >1, perform DAD and disable IPv6 operation if DAD for MAC-based
  link-local address has been failed (RFC4862 5.4.5).

We do not follow RFC4862 by default.  Refer to the netdev thread entitled
"Linux IPv6 DAD not full conform to RFC 4862 ?"
http://www.spinics.net/lists/netdev/msg52027.html

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>

ipv6: Add disable_ipv6 sysctl to disable IPv6 operaion on specific interface.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>

ipv6: Assume the loopback address in link-local scope.

Handle interface property strictly when looking up a route
for the loopback address (RFC4291 2.5.3).

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>

ipv6: Do not forward packets with the unspecified source address.

RFC4291 2.5.2.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>

ipv6: Do not assign non-valid address on interface.

Check the type of the address when adding a new one on interface.
- the unspecified address (::) is always disallowed (RFC4291 2.5.2)
- the loopback address is disallowed unless the interface is (one of)
  loopback (RFC4291 2.5.3).
- multicast addresses are disallowed.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>

tcp: de-bloat a bit with factoring NET_INC_STATS_BH out

There are some places in TCP that select one MIB index to
bump snmp statistics like this:

if (<something>)
NET_INC_STATS_BH(<some_id>);
else if (<something_else>)
NET_INC_STATS_BH(<some_other_id>);
...
else
NET_INC_STATS_BH(<default_id>);

or in a more tricky but still similar way.

On the other hand, this NET_INC_STATS_BH is a camouflaged
increment of percpu variable, which is not that small.

Factoring those cases out de-bloats 235 bytes on non-preemptible
i386 config and drives parts of the code into 80 columns.

add/remove: 0/0 grow/shrink: 0/7 up/down: 0/-235 (-235)
function                                     old     new   delta
tcp_fastretrans_alert                       1437    1424     -13
tcp_dsack_set                                137     124     -13
tcp_xmit_retransmit_queue                    690     676     -14
tcp_try_undo_recovery                        283     265     -18
tcp_sacktag_write_queue                     1550    1515     -35
tcp_update_reordering                        162     106     -56
tcp_retransmit_timer                         990     904     -86

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

Merge branch 'master' of /linux/kernel/git/linville/wireless-2.6

Merge branch 'for-linus' of git://git./linux/kernel/git/jbarnes/pci-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci-2.6:
  PCI: acpiphp: cleanup notify handler on all root bridges
  PCI: Limit VPD read/write lengths for Broadcom 5706, 5708, 5709 rev.
  PCI: Restrict VPD read permission to root

Merge branch 'i2c-fix' of git://aeryn.fluff.org.uk/bjdooks/linux

* 'i2c-fix' of git://aeryn.fluff.org.uk/bjdooks/linux:
  I2C: S3C2410: Add MODULE_ALIAS() for s3c2440 device.
  I2C: S3C2410: Fixup error codes returned rom a transfer.
  I2C: S3C2410: Check ACK on byte transmission

Merge branch 'for-2.6.26' of git://git.kernel.dk/linux-2.6-block

* 'for-2.6.26' of git://git.kernel.dk/linux-2.6-block:
  Properly notify block layer of sync writes
  block: Fix the starving writes bug in the anticipatory IO scheduler

Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6:
  [IA64] export account_system_vtime
  [IA64] Bugfix for system with 32 cpus

Merge git://git./linux/kernel/git/mchehab/v4l-dvb

* git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/v4l-dvb:
  V4L/DVB (8178): uvc: Fix compilation breakage for the other drivers, if uvc is selected
  V4L/DVB (8145a): USB Video Class driver

Merge git://git./linux/kernel/git/bart/ide-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/bart/ide-2.6:
  ide: fix /proc/ide/ide?/mate reporting
  Revert "BAST: Remove old IDE driver"

Merge master.kernel.org:/home/rmk/linux-2.6-arm

* master.kernel.org:/home/rmk/linux-2.6-arm:
  [ARM] 5131/1: Annotate platform_secondary_init with trace_hardirqs_off
  [ARM] 5117/1: pxafb: fix __devinit/exit annotations
  [ARM] Export dma_sync_sg_for_device()
  [ARM] 5109/1: Mark rtc sa1100 driver as wakeup source before registering it
  [ARM] 5116/1: pxafb: cleanup and fix order of failure handling
  [ARM] 5115/1: pxafb: fix ifdef for command line option handling
  ARM: OMAP: Correcting the gpmc prefetch control register address
  ARM: OMAP: DMA: Don't mark channel active in omap_enable_channel_irq

tty: Fix inverted logic in send_break

Not sure how this came to get inverted but it appears to have been my
mess up.

Signed-off-by: Alan Cox <alan@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Merge branch 'sched-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  sched: fix divide error when trying to configure rt_period to zero

Merge branch 'i2c-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6

* 'i2c-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6:
  i2c: Fix bad hint about irqs in i2c.h
  i2c: Documentation: fix device matching description

Merge branch 'core-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  rcu: fix hotplug vs rcu race

Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86: fix NODES_SHIFT Kconfig range

Merge git://git./linux/kernel/git/jejb/scsi-rc-fixes-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6:
  [SCSI] esp: tidy up target reference counting
  [SCSI] esp: Fix OOPS in esp_reset_cleanup().
  [SCSI] ses: Fix timeout

Merge git://git./linux/kernel/git/agk/linux-2.6-dm

* git://git.kernel.org/pub/scm/linux/kernel/git/agk/linux-2.6-dm:
  dm crypt: use cond_resched

Merge branch 'for-2.6.26' of git://neil.brown.name/md

* 'for-2.6.26' of git://neil.brown.name/md:
  Fix error paths if md_probe fails.
  Don't acknowlege that stripe-expand is complete until it really is.
  Ensure interrupted recovery completed properly (v1 metadata plus bitmap)

Merge branch 'merge' of git://git./linux/kernel/git/paulus/powerpc

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc:
  powerpc/mpc5200: Fix lite5200b suspend/resume
  powerpc/legacy_serial: Bail if reg-offset/shift properties are present
  powerpc/bootwrapper: update for initrd with simpleImage

Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (55 commits)
  net: fib_rules: fix error code for unsupported families
  netdevice: Fix wrong string handle in kernel command line parsing
  net: Tyop of sk_filter() comment
  netlink: Unneeded local variable
  net-sched: fix filter destruction in atm/hfsc qdisc destruction
  net-sched: change tcf_destroy_chain() to clear start of filter list
  ipv4: fix sysctl documentation of time related values
  mac80211: don't accept WEP keys other than WEP40 and WEP104
  hostap: fix sparse warnings
  hostap: don't report useless WDS frames by default
  textsearch: fix Boyer-Moore text search bug
  netfilter: nf_conntrack_tcp: fixing to check the lower bound of valid ACK
  ipv6 route: Convert rt6_device_match() to use RT6_LOOKUP_F_xxx flags.
  netlabel: Fix a problem when dumping the default IPv6 static labels
  net/inet_lro: remove setting skb->ip_summed when not LRO-able
  inet fragments: fix race between inet_frag_find and inet_frag_secret_rebuild
  CONNECTOR: add a proc entry to list connectors
  netlink: Fix some doc comments in net/netlink/attr.c
  tcp: /proc/net/tcp rto,ato values not scaled properly (v2)
  include/linux/netdevice.h: don't export MAX_HEADER to userspace
  ...