netdev: ehea: locking order correction

Nested locks always need to be taken in the same order.  This change factors
out the ehea_fw_handles.lock to make the locking order consistent.

Signed-off-by: Daniel Walker <dwalker@mvista.com>
Cc: Christoph Raisch <raisch@de.ibm.com>
Cc: Jan-Bernd Themann <themann@de.ibm.com>
Cc: Thomas Klein <tklein@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

netdev: ehea: ehea_fw_handles semaphore to mutex

Converted the ehea_fw_handles.lock to a mutex.

Signed-off-by: Daniel Walker <dwalker@mvista.com>
Cc: Christoph Raisch <raisch@de.ibm.com>
Cc: Jan-Bernd Themann <themann@de.ibm.com>
Cc: Thomas Klein <tklein@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

netdev: ehea: semaphore to mutex

Converted the dlpar_mem_lock.  With a bit of cleanup, I converted to
DEFINE_MUTEX() instead of a runtime init.  I also made the lock static.

Signed-off-by: Daniel Walker <dwalker@mvista.com>
Cc: Christoph Raisch <raisch@de.ibm.com>
Acked-by: Jan-Bernd Themann <themann@de.ibm.com>
Cc: Thomas Klein <tklein@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

drivers/net/tokenring/3c359.c: use time_* macros

The functions time_before, time_before_eq, time_after, and time_after_eq are
more robust for comparing jiffies against other values.

So use the time_after() macro, defined in linux/jiffies.h, which deals with
wrapping correctly.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: S.Caglar Onur <caglar@pardus.org.tr>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

drivers/net/arcnet/arcnet.c: use time_* macros

The functions time_before, time_before_eq, time_after, and time_after_eq are
more robust for comparing jiffies against other values.

So use the time_after() macro, defined in linux/jiffies.h, which deals with
wrapping correctly.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: S.Caglar Onur <caglar@pardus.org.tr>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

3c509: convert to isa_driver and pnp_driver

Convert 3c509 driver to isa_driver and pnp_driver.  The result is that
autoloading using udev and hibernation works with ISA PnP cards.  It also adds
hibernation support for non-PnP ISA cards.

xcvr module parameter was removed as its value was not used.

Tested using 3 ISA cards in various combinations of PnP and non-PnP modes.
EISA and MCA only compile-tested.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

[NET]: Fix dev_alloc_skb() typo.

Noticed by Joe Perches.

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

[LIB]: Drop the pcounter itself.

The knock-out. The pcounter abstraction is not used any longer in the
kernel.

Not sure whether this should go via netdev tree, but as far as I
remember it was added via this one, and besides Eric thinks that
Andrew shouldn't mind this.

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

[SOCK]: Drop inuse pcounter from struct proto (v2).

An uppercut - do not use the pcounter on struct proto.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Acked-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[SOCK]: Drop per-proto inuse init and fre functions (v2).

Constructive part of the set is finished here. We have to remove the
pcounter, so start with its init and free functions.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Acked-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[SOCK]: Introduce a percpu inuse counters array (v2).

And redirect sock_prot_inuse_add and _get to use one.

As far as the dereferences are concerned. Before the patch we made
1 dereference to proto->inuse.add call, the call itself and then
called the __get_cpu_var() on a static variable. After the patch we
make a direct call, then one dereference to proto->inuse_idx and
then the same __get_cpu_var() on a still static variable. So this
patch doesn't seem to produce performance penalty on SMP.

This is not per-net yet, but I will deliberately make NET_NS=y case
separated from NET_NS=n one, since it'll cost us one-or-two more
dereferences to get the struct net and the inuse counter.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Acked-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[SOCK]: Enumerate struct proto-s to facilitate percpu inuse accounting (v2).

The inuse counters are going to become a per-cpu array.  Introduce an
index for this array on the struct proto.

To handle the case of proto register-unregister-register loop the
bitmap is used. All its bits manipulations are protected with
proto_list_lock and a sanity check for the bitmap being exhausted is
also added.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Acked-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: Rename inet_frag.h identifiers COMPLETE, FIRST_IN, LAST_IN to INET_FRAG_*

On Fri, 2008-03-28 at 03:24 -0700, Andrew Morton wrote:
> they should all be renamed.

Done for include/net and net

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

[IPV6]: Remove unused method declaration in include/net/addrconf.h.

This patches removes unused declaration of addrconf_forwarding_on() method
in include/net/addrconf.h.

Signed-off-by: Rami Rosen <ramirose@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: Add inline intent commentary to dev_alloc_skb().

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: Protect device namespace inlines with CONFIG_NET

Include sites should not be bothered by whether
CONFIG_NET is set or not when trying to include
benign files like linux/etherdevice.h et al.

From a report by Stephen Rothwell.

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

[IPV6] NEIGH: Optimize is_router check.

Our interest is not the whole entry of proxy neighbor but the
NTF_ROUTER flag.  Let's test it explicitly.

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

[NET] NEIGHBOUR: Extract hash/lookup functions for pneigh entries.

Extract hash function for pneigh entries from pneigh_lookup(),
__pneigh_lookup() and pneigh_delete() as pneigh_hash().
Extract core of pneigh_lookup() and __pneigh_lookup() as
__pneigh_lookup_1().

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

[NET] NEIGHBOUR: Make each EXPORT_SYMBOL{,_GPL}() immediately follow its function/variable.

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

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

Conflicts:

drivers/net/usb/rndis_host.c
drivers/net/wireless/b43/dma.c
net/ipv6/ndisc.c

Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-2.6.26

[SCTP]: Remove sctp_add_cmd_sf wrapper bloat

With a was number of callsites sctp_add_cmd_sf wrapper bloats
kernel by some amount. Due to unlikely tracking allyesconfig,
with the initial result were around ~7kB (thus caught my
attention) while a non-debug config produced only ~2.3kB effect.

I (ij) proposed first a patch to uninline it but Vlad responded
with a patch that removed the only sctp_add_cmd call which is
wrapped by sctp_add_cmd_sf (I wasn't sure if I could do that).
I did minor cleanup to Vlad's patch.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: uninline skb_trim, de-bloats

Allyesconfig (v2.6.24-mm1):
-10976  209 funcs, 123 +, 11099 -, diff: -10976 --- skb_trim

Without number of debug related CONFIGs (v2.6.25-rc2-mm1):
-7360  192 funcs, 131 +, 7491 -, diff: -7360 --- skb_trim
skb_trim                      |  +42

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: uninline dst_release

Codiff stats (allyesconfig, v2.6.24-mm1):
-16420  187 funcs, 103 +, 16523 -, diff: -16420 --- dst_release

Without number of debug related CONFIGs (v2.6.25-rc2-mm1):
-7257  186 funcs, 70 +, 7327 -, diff: -7257 --- dst_release
dst_release                   |  +40

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: uninline skb_push, de-bloats a lot

Allyesconfig (v2.6.24-mm1):

-21593  356 funcs, 2418 +, 24011 -, diff: -21593 --- skb_push

Without many debug related CONFIGs (v2.6.25-rc2-mm1):

-13890  341 funcs, 189 +, 14079 -, diff: -13890 --- skb_push
skb_push                      |  +46

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: uninline dev_alloc_skb, de-bloats a lot

Allyesconfig (v2.6.24-mm1):

-23668  392 funcs, 104 +, 23772 -, diff: -23668 --- dev_alloc_skb

Without many debug CONFIGs (v2.6.25-rc2-mm1):

-12178  382 funcs, 157 +, 12335 -, diff: -12178 --- dev_alloc_skb
dev_alloc_skb                 |  +37

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: uninline skb_pull, de-bloats a lot

Allyesconfig (v2.6.24-mm1):

-28162  354 funcs, 3005 +, 31167 -, diff: -28162 --- skb_pull

Without number of debug related CONFIGs (v2.6.25-rc2-mm1):

-9697  338 funcs, 221 +, 9918 -, diff: -9697 --- skb_pull
skb_pull                      |  +44

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: uninline skb_put, de-bloats a lot

Allyesconfig (v2.6.24-mm1):

~500 files changed
...
 869 funcs, 198 +, 111003 -, diff: -110805 --- skb_put
  skb_put                       | +104

Without number of debug related CONFIGs (v2.6.25-rc2-mm1):

-60744  855 funcs, 861 +, 61605 -, diff: -60744 --- skb_put
  skb_put                       |  +57

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: Use local_irq_{save,restore}() in napi_complete().

Based upon a lockdep report.

Since ->poll() can be invoked from netpoll with interrupts
disabled, we must not unconditionally enable interrupts
in napi_complete().

Instead we must use local_irq_{save,restore}().

Noticed by Peter Zijlstra:

<irqs disabled>

  netpoll_poll()
    poll_napi()
      spin_trylock(&napi->poll_lock)
      poll_one_napi()
        napi->poll() := sky2_poll()
          napi_complete()
            local_irq_disable()
            local_irq_enable() <--- *BUG*

  <irq>
    irq_exit()
      do_softirq()
        net_rx_action()
          spin_lock(&napi->poll_lock) <--- Deadlock!

Because we still hold the lock....

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

[IPV6]: Remove three unused method declarations in include/net/ipv6.h

This patch removes three unused method declarations in include/net/ipv6.h:
inet_getfrag_t(), ipv6_build_nfrag_opts() and ipv6_build_frag_opts().

Signed-off-by: Rami Rosen <ramirose@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NETFILTER]: Replate direct proc_fops assignment with proc_create call.

This elliminates infamous race during module loading when one could lookup
proc entry without proc_fops assigned.

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

[PKT_SCHED]: Pass real namespace in net scheduler classifiers.

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

[ESP]: Ensure IV is in linear part of the skb to avoid BUG() due to OOB access

ESP does not account for the IV size when calling pskb_may_pull() to
ensure everything it accesses directly is within the linear part of a
potential fragment. This results in a BUG() being triggered when the
both the IPv4 and IPv6 ESP stack is fed with an skb where the first
fragment ends between the end of the esp header and the end of the IV.

This bug was found by Dirk Nehring <dnehring@gmx.net> .

Signed-off-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NETNS]: Do no include NET related headers if CONFIG_NET is not set.

This fix broken compilation for 'allnoconfig'. This was introduced by
Introduced by commit 1218854afa6f659be90b748cf1bc7badee954a35 ("[NET]
NETNS: Omit seq_net_private->net without CONFIG_NET_NS.")

Signed-off-by: Denis V. Lunev <den@openvz.org>
Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NETNS]: Compile NET /proc support only if CONFIG_NET is set.

This fix broken compilation for 'allnoconfig'. This was introduced by
Introduced by commit 1218854afa6f659be90b748cf1bc7badee954a35 ("[NET]
NETNS: Omit seq_net_private->net without CONFIG_NET_NS.")

Signed-off-by: Denis V. Lunev <den@openvz.org>
Acked-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

libertas: reduce debug output

This patch tries to make dmesg logs between different runs easier
to compare by

* removing the jiffies (use CONFIG_PRINTK_TIME if you need
  timing)
* remove the line numbers, they change with each applied patch

It also changes the deprecated __FUNCTION__ to __func__ to make
checkpatch.pl happy.

Signed-off-by: Holger Schurig <hs4233@mail.mn-solutions.de>
Acked-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

mac80211: reorder fields to make some structures smaller

This patch reorders some fields in various structures to have
less padding within the structures, making them smaller. It
doesn't yet make any type adjustments, but often size_t is used
for example for IE lengths which is total overkill since size_t
will be 8 bytes long on 64-bit yet the length can at most fill
a u8.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

prism54: correct thinko in "prism54: Convert stats_sem in a mutex"

mutex_trylock has different return code semantics than down_trylock...

Signed-off-by: John W. Linville <linville@tuxdriver.com>

mac80211: A-MPDU MLME use dynamic allocation

This patch alters the A-MPDU MLME in sta_info to use dynamic allocation,
thus drastically improving memory usage - from a constant ~2 Kbyte in
the previous (static) allocation to a lower limit of ~200 Byte and an upper
limit of ~2 Kbyte.

Signed-off-by: Ron Rindjunsky <ron.rindjunsky@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

cfg80211: don't export ieee80211_get_channel

This patch makes ieee80211_get_channel a static inline defined in
cfg80211's header file which simply calls __ieee80211_get_channel
to avoid symbol clashes with the ieee80211 code.

The problem was pointed out by David Miller, thanks!

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

libertas: the compact flash driver is no longer experimental

Signed-off-by: Holger Schurig <hs4233@mail.mn-solutions.de>
Acked-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

libertas: remove CMD_802_11_PWR_CFG

This has nowhere been used. Note: in the firmware manual this was
documented as CMD_802_11_PA_CFG. If we ever need it, we can/should
re-implement it as a direct command.

Signed-off-by: Holger Schurig <hs4233@mail.mn-solutions.de>
Acked-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

libertas: kill useless #define LBS_MONITOR_OFF 0

Signed-off-by: Holger Schurig <hs4233@mail.mn-solutions.de>
Acked-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlwifi: iwl_priv - clean up in types of members

This patch fix types of is_open and iw_mode members
of iwl_priv sturct

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlwifi: allow a default callback for ASYNC host commands

This patch provides a default callback for ASYNC host commands instead
of calling to BUG_ON. Most of the callbacks are now just empty functions

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

mac80211: fix wrong Rx A-MPDU control via debugfs

This patch eliminate the use of buf_size as a trigger in favor of a new
flag to control Rx A-MPDU sessions through debugfs

Signed-off-by: Ron Rindjunsky <ron.rindjunsky@intel.com>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlwifi: improve NIC i/o debug prints information

This patch gives the function's caller name in case NIC
access reference count was not used by it.

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Ron Rindjunsky <ron.rindjunsky@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlwifi: rename iwl-4965-io.h to iwl-io.h

This patch renames iwl-4965-io.h back to iw-io.h
it also remove 4965 from all functions it supplies

Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlwifi: Add led support

This patch add LEDS support to 3965 and 4965 drivers. It is based on
led trigger and class. For our drivers we needed to avoid two things.
1- We receive led trigger on/off on each Rx\Tx frame. In our driver
    we can not call led command like that. In this driver once driver
    receive a start of traffic it call the led command to start blinking
    then we count all bytes of Tx and Rx frame, after two second we count the
    blink rate of last two second then id blink rate changed we call the led
     commands
2- Since we can call led command very often, we make sure we call the
    led command after we receive the statistics notification so
    we don't need to wake up the ucode id it is in sleep state.
    This patch was tested with 4965 and 3945.

Signed-off-by: Mohamed Abbas <mabbas@linux.intel.com>
Signed-off-by: Ian Schram<ischram@telenet.be>
Signed-off-by: Tomas Winkler <tomas.winkler@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlwifi: do not register bands with no supported channels

Otherwise, b/g-only devices fail in wiphy_register.

Signed-off-by: John W. Linville <linville@tuxdriver.com>

mac80211: silently accept deletion of non-existant key

Otherwise, 'iwconfig wlan0 key off' with no key set results in:

Error for wireless request "Set Encode" (8B2A) :
    SET failed on device wlan0 ; No such file or directory.

Signed-off-by: John W. Linville <linville@tuxdriver.com>

[NETNS][IPV6] flowlabels - make proc per namespace

Make /proc/net/ip6_flowlabel show only flow labels belonging to the
current network namespace.

Signed-off-by: Benjamin Thery <benjamin.thery@bull.net>
Signed-off-by: Daniel Lezcano <dlezcano@fr.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NETNS][IPV6] flowlabels - make flowlabels per namespace

This patch introduces a new member, fl_net, in struct ip6_flowlabel.
This allows to create labels with the same value in different namespaces.

Signed-off-by: Benjamin Thery <benjamin.thery@bull.net>
Signed-off-by: Daniel Lezcano <dlezcano@fr.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NETNS][IPV6] anycast - handle several network namespace

Make use of the network namespace information to have this protocol to
handle several network namespace.

Signed-off-by: Daniel Lezcano <dlezcano@fr.ibm.com>
Signed-off-by: Benjamin Thery <benjamin.thery@bull.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPSEC]: Fix BEET output

The IPv6 BEET output function is incorrectly including the inner
header in the payload to be protected.  This causes a crash as
the packet doesn't actually have that many bytes for a second
header.

The IPv4 BEET output on the other hand is broken when it comes
to handling an inner IPv6 header since it always assumes an
inner IPv4 header.

This patch fixes both by making sure that neither BEET output
function touches the inner header at all.  All access is now
done through the protocol-independent cb structure.  Two new
attributes are added to make this work, the IP header length
and the IPv4 option length.  They're filled in by the inner
mode's output function.

Thanks to Joakim Koskela for finding this problem.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPV6]: Fix potential net leak and oops in ipv6 routing code.

The commits f3db4851 ([NETNS][IPV6] ip6_fib - fib6_clean_all handle several
network namespaces) and 69ddb805 ([NETNS][IPV6] route6 - Make proc entry
/proc/net/rt6_stats per namespace) made some proc files per net.

Both of them introduced potential OOPS - get_proc_net can return NULL, but
this check is lost - and a struct net leak - in case single_open() fails the
previously got net is not put.

Kill all these bugs with one patch.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Acked-by: Daniel Lezcano <dlezcano@fr.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[TIPC]: Cosmetic cleanup of TIPC polling logic

This patch eliminates an unnecessary poll-related routine
by merging it into TIPC's main polling routine, and updates
the comments associated with this code.

Signed-off-by: Allan Stephens <allan.stephens@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[VLAN]: Reduce memory consumed by vlan_groups

Currently each vlan_groupd contains 8 pointers on arrays with 512
pointers on struct net_device each  :)  Such a construction "in many
cases ... wastes memory".

My proposal is to allow for some of these arrays pointers be NULL,
meaning that there are no devices in it. When a new device is added
to the vlan_group, the appropriate array is allocated.

The check in vlan_group_get_device's is safe, since the pointer
vg->vlan_devices_arrays[x] can only switch from NULL to not-NULL.
The vlan_group_prealloc_vid() is guarded with rtnl lock and is
also safe.

I've checked (I hope that) all the places, that use these arrays
and found, that the register_vlan_dev is the only place, that can
put a vlan device on an empty vlan_group.

Rough calculations shows, that after the patch a setup with a
single vlan dev (or up to 512 vlans with sequential vids) will
occupy approximately 8 times less memory.

The question I have is - does this patch makes sense, or a totally
new structures are required to store the vlan_devs?

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Patrick McHardy <kaber@trash.net>

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

[ICMP]: Dst entry leak in icmp_send host re-lookup code (v2).

Commit 8b7817f3a959ed99d7443afc12f78a7e1fcc2063 ([IPSEC]: Add ICMP host
relookup support) introduced some dst leaks on error paths: the rt
pointer can be forgotten to be put. Fix it bu going to a proper label.

Found after net namespace's lo refused to unregister :) Many thanks to
Den for valuable help during debugging.

Herbert pointed out, that xfrm_lookup() will put the rtable in case
of error itself, so the first goto fix is redundant.

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

[NETNS][ICMP]: Build fix for NET_NS=n case (dev->nd_net is omitted).

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

[AX25]: Remove obsolete references to BKL from TODO file.

Given that there are no apparent calls to lock_kernel() or
unlock_kernel() under net/ax25, delete the TODO reference related to
that.

Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: Fix multicast device ioctl checks

SIOCADDMULTI/SIOCDELMULTI check whether the driver has a set_multicast_list
method to determine whether it supports multicast. Drivers implementing
secondary unicast support use set_rx_mode however.

Check for both dev->set_multicast_mode and dev->set_rx_mode to determine
multicast capabilities.

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

[NETNS][ICMP]: Use per-net sysctls in ipv4/icmp.c.

This mostly re-uses the net, used in icmp netnsization patches from Denis.

After this ICMP sysctls are completely virtualized.

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

[NETNS][ICMP]: Make ctl tables for ICMP sysctls per-net.

Add some flesh to ipv4_sysctl_init_net and ipv4_sysctl_exit_net,
i.e. copy the table, alter .data pointers and register it per-net.

Other ipv4_table's sysctls are now global, but this is going to
change once sysctl permissions patches migrate from -mm tree to
mainline in 2.6.26 merge window :)

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

[NETNS][ICMP]: Move ICMP sysctls on struct net.

Initialization is moved to icmp_sk_init, all the places, that
refer to them use init_net for now.

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

[NETNS][ICMP]: Register pernet subsys to make ICMP sysctls per-net.

This includes adding pernet_operations, empty init and exit
hooks and a bit of changes in sysctl_ipv4_init just not to
have this part in next patches.

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

[IRDA]: Store irnet_socket termios properly.

It should be a "struct ktermios" not a "struct termios".

Based upon a build warning reported by Stephen Rothwell.

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

[NETNS]: Compilation warnings under CONFIG_NET_NS.

Recent commits from YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
have been introduced a several compilation warnings
'assignment discards qualifiers from pointer target type'
due to extra const modifier in the inline call parameters of
{dev|sock|twsk}_net_set.

Drop it.

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

[NETNS]: Compilation fix for include/linux/netdevice.h.

Commit commit c346dca10840a874240c78efe3f39acf4312a1f2
([NET] NETNS: Omit net_device->nd_net without CONFIG_NET_NS)
breaks compilation with CONFIG_NET_NS set.

Fix the typo.

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

[UML]: uml-net: don't set IFF_ALLMULTI in set_multicast_list

IFF_ALLMULTI is an indication from the network stack to the driver
to disable multicast filters, drivers should never set it directly.

Since the UML networking device doesn't have any filtering capabilites,
it doesn't the set_multicast_list function at all, it is kept so userspace
can still issue SIOCADDMULTI/SIOCDELMULTI ioctls however.

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

[VLAN]: Don't copy ALLMULTI/PROMISC flags from underlying device

Changing these flags requires to use dev_set_allmulti/dev_set_promiscuity
or dev_change_flags. Setting it directly causes two unwanted effects:

- the next dev_change_flags call will notice a difference between
  dev->gflags and the actual flags, enable promisc/allmulti
  mode and incorrectly update dev->gflags

- this keeps the underlying device in promisc/allmulti mode until
  the VLAN device is deleted

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

Merge branch 'upstream-net26' of /linux/kernel/git/jgarzik/netdev-2.6

gianfar: Fix Rx/Tx HW interrupt coalescing counter reset procedure.

- Fix Rx/Tx HW interrupt coalescing counter reset logic. Disabling
is required before resetting the counter.

- Update the Default both Rx and Tx coalescing timer
threshold. Formerly 4 is set which is equal to 1.5 frame at the line
rate of 1GbE interface, and it doesn't match to the coalescing frame
count which is set to 16. Threashold 21 is matched to frame count 16.

Signed-off-by: Dai Haruki <dai.haruki@freescale.com>
Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

gianfar: Only process completed frames

If the LAST bit is not set in the RxBD, it's possible we're processing
an incomplete frame, which is bad.  While we're at it, add a constant
for the error bitmask, so the whole if-clause fits on one line,
and is more legible.

Signed-off-by: Dai Haruki <dai.haruki@freescale.com>
Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

gianfar: Fix frame size calculation when hardware VLAN acceleration is on

In gfar_change_mtu(), the frame size needs to be increased to account
for the extra 4 bytes VLAN adds to the ethernet header.  However,
it was being increased by the length of the whole header (18 bytes),
which is wrong.

Signed-off-by: Dai Haruki <dai.haruki@freescale.com>
Signed-off-by: Andy Fleming <afleming@freescale.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

gianfar: Fix the data buffer stashing amount

- Buffer stashing parameter change to 96 from 64 in order to cover
the Layer 4 header.

Signed-off-by: Dai Haruki <dai.haruki@freescale.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

Use single_open instead of manual manipulations.

The code opening proc entry for each device makes the
same thing, as the single_open does, so remove the
unneeded code.

Signed-off-by: Pavel Emelyanov <xemul@openvz.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

phy/broadcom: add support for BCM5481 PHY

This patch adds support for BCM5481 PHY. Unfortunately it's hard to
get specifications for this PHY, so its special register 0x18 isn't
annotated properly (but we know it's used to set up the delays).

I've kept the magic numbers, so we'll not forget to fix it at the
first opportunity, and will name that register and its bits correctly.

p.s. also fixed the line with broken indention, introduced by
commit 03157ac31eb4a8883382a212b161d2e6c5059fbf
    PHYLIB: Add BCM5482 PHY support

Signed-off-by: Anton Vorontsov <avorontsov@ru.mvista.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

qla3xxx: convert byte order of constant instead of variable

Convert byte order of constant instead of variable which can be done at
compile time (vs run time)

Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Acked-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

drivers/net/mv643xx_eth.c: Use FIELD_SIZEOF

Robert P.J. Day proposed to use the macro FIELD_SIZEOF in replace of code
that matches its definition.

The modification was made using the following semantic patch
(http://www.emn.fr/x-info/coccinelle/)

// <smpl>
@haskernel@
@@

#include <linux/kernel.h>

@depends on haskernel@
type t;
identifier f;
@@

- (sizeof(((t*)0)->f))
+ FIELD_SIZEOF(t, f)

@depends on haskernel@
type t;
identifier f;
@@

- sizeof(((t*)0)->f)
+ FIELD_SIZEOF(t, f)
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

Add support the Korina (IDT RC32434) Ethernet MAC

This patch adds support for the IDT rc32434 Ethernet MAC
we can find in the IDT boards and the Mikrotik RB500.
Driver references some code from the linux-mips RB500
support.

Signed-off-by: Florian Fainelli <florian.fainelli@telecomint.eu>
Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Philip Rischel <rischelp@idt.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

forcedeth: Use round_jiffies for stats timer

This timer doesn't need to run at precise times, so round it to a whole
second to decrease wakeups.

Signed-off-by: Daniel Drake <dsd@gentoo.org>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

s2io annotations

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

endianness annotations: rndis

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

annotate cxgb3 (ab)uses of skb->priority/skb->csum

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

skfp annotations

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

misc drivers/net annotations

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

ni52: more unbreaking

missed read*/write* plus a bunch of wrong-sized ones...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

ni52: switch to ioremap()

isa_bus_to_virt() is the wrong thing to do here; it happens
to work on i386, but only by accident.  What we want is
normal ioremap/readb/etc. set - it's all in iomem.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

ni52: clean up check586()

take iscp-based testing into helper, kill the loop, stop
wanking with reassignments of priv->iscp

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

ni52: clean up initialization of priv

* initialize spinlock once
* check586() used to be done before we'd allocated ->priv; these days
  it's there from the very beginning, so we don't have to play with
  private copy.  Consequently, we don't need to mess with reinitializing
  ->base, etc. afterwards.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

ni52: make ->base char __iomem *

... and store the virt address where we map the ->mem_addr, while we
are at it.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

ni52: kill the second argument of check586(), get rid of bogus pointer

Passing ISA bus address explicitly cast to char * only to cast it back to
unsigned long is dumb; so's passing it at all when it's always dev->mem_start...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

ni52: unbreak scp_struct

unsigned short != u8, it's u16...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

ni52 trivial iomem annotations

Driver is still broken, though; partially from Alan's checkpatch-induced
fun, partially from layers of ancient mess ;-)

By the end of the series... hell, might be even worth trying to stick
such card into old alpha or ppc with an ISA slot and see if it work -
would be for the first time ever in case of alpha and for the first
time since at least 2.5.3 in case of ppc...

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

ixgb: remove irq_sem

ixgb can remove irq_sem by auditing all the call sites to make sure
that each of them makes sure the adapter is in the correct state
before re-enabling interrupts.  after doing this to all of our other
drivers it is becoming easier.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

e1000: remove irq_sem

irq_sem was just a hack to prevent interrupts from being enabled
unexpectedly in deep call paths.  Simply finding those call paths and
fixing them by hand results in a driver that behaves as we expect and
doesn't need the atomic at all.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

e1000e: remove irq_sem

irq_sem can safely be removed by auditing all irq.*able sites to
make sure that interrupts don't get enabled unexpectedly when the
interface is down.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>

e1000e: remove no longer used e1000e_read_nvm_spi

This function is no longer used now that 82573 uses the eerd
read method as well. Thanks to Adrian Bunk for pointing this out.

Signed-off-by: Auke Kok <auke-jan.h.kok@intel.com>
Signed-off-by: Jeff Garzik <jeff@garzik.org>