[AF_UNIX]: use shift instead of integer division

The patch below replaces a divide by 2 with a shift -- sk_sndbuf is an
integer, so gcc emits an idiv, which takes 10x longer than a shift by 1.
This improves af_unix bandwidth by ~6-10K/s.  Also, tidy up the comment
to fit in 80 columns while we're at it.

Signed-off-by: Benjamin LaHaise <benjamin.c.lahaise@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: Uninline kfree_skb and allow NULL argument

o Uninline kfree_skb, which saves some 15k of object code on my notebook.

o Allow kfree_skb to be called with a NULL argument.

  Subsequent patches can remove conditional from drivers and further
  reduce source and object size.

Signed-off-by: Jörn Engel <joern@wohnheim.fh-wedel.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

[LLC]: Fix sap refcounting

Thanks to Leslie Harlley Watter <leslie@watter.org> for reporting the
problem an testing this patch.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[LLC]: Replace __inline__ with inline

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[LLC]: Fix struct proto .name

Cut'n'paste error from ddp_proto.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET] pktgen: Fix races between control/worker threads.

There's a race in pktgen which can lead to a double
free of a pktgen_dev's skb. If a worker thread is in
the midst of doing fill_packet(), and the controlling
thread gets a "stop" message, the already freed skb
can be freed once again in pktgen_stop_device(). This
patch gives all responsibility for cleaning up a
pktgen_dev's skb to the associated worker thread.

Signed-off-by: Arthur Kepner <akepner@sgi.com>
Acked-by: Robert Olsson <Robert.Olsson@data.slu.se>
Signed-off-by: David S. Miller <davem@davemloft.net>

[XFRM]: Rearrange struct xfrm_aevent_id for better compatibility.

struct xfrm_aevent_id needs to be 32-bit + 64-bit align friendly.

Based upon suggestions from Yoshifuji.

Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP]: Move the IPv4 specific bits from proto.c to ipv4.c

With this patch in place we can break down the complexity by better
compartmentalizing the code that is common to ipv6 and ipv4.

Now we have these modules:
Module                  Size  Used by
dccp_diag               1344  0
inet_diag               9448  1 dccp_diag
dccp_ccid3             15856  0
dccp_tfrc_lib          12320  1 dccp_ccid3
dccp_ccid2              5764  0
dccp_ipv4              16996  2
dccp                   48208  4 dccp_diag,dccp_ccid3,dccp_ccid2,dccp_ipv4

dccp_ipv6 still requires dccp_ipv4 due to dccp_ipv6_mapped, that is
the next target to work on the "hey, ipv4 is legacy, I only want ipv6
dude!" direction.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP]: Rename init_dccp_v4_mibs to dccp_mib_init

And introduce dccp_mib_exit grouping previously open coded sequence.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP]: Move dccp_hashinfo from ipv4.c to the core

As it is used by both ipv4 and ipv6.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP]: Dont use dccp_v4_checksum in dccp_make_response

dccp_make_response is shared by ipv4/6 and the ipv6 code was
recalculating the checksum, not good, so move the dccp_v4_checksum
call to dccp_v4_send_response.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP]: Move dccp_[un]hash from ipv4.c to the core

As this is used by both ipv4 and ipv6 and is not ipv4 specific.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP]: Move dccp_v4_{init,destroy}_sock to the core

Removing one more ipv6 uses ipv4 stuff case in dccp land.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP]: Generalize dccp_v4_send_reset

Renaming it to dccp_send_reset and moving it from the ipv4 specific
code to the core dccp code.

This fixes some bugs in IPV6 where timers would send v4 resets, etc.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP] feat: Introduce sysctls for the default features

[root@qemu ~]# for a in /proc/sys/net/dccp/default/* ; do echo $a ; cat $a ; done
/proc/sys/net/dccp/default/ack_ratio
2
/proc/sys/net/dccp/default/rx_ccid
3
/proc/sys/net/dccp/default/send_ackvec
1
/proc/sys/net/dccp/default/send_ndp
1
/proc/sys/net/dccp/default/seq_window
100
/proc/sys/net/dccp/default/tx_ccid
3
[root@qemu ~]#

So if wanting to test ccid3 as the tx CCID one can just do:

[root@qemu ~]# echo 3 > /proc/sys/net/dccp/default/tx_ccid
[root@qemu ~]# echo 2 > /proc/sys/net/dccp/default/rx_ccid
[root@qemu ~]# cat /proc/sys/net/dccp/default/[tr]x_ccid
2
3
[root@qemu ~]#

Of course we also need the setsockopt for each app to tell its preferences, but
for testing or defining something other than CCID2 as the default for apps that
don't explicitely set their preference the sysctl interface is handy.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP]: Call dccp_feat_init more early in dccp_v4_init_sock

So that dccp_feat_clean doesn't get confused with uninitialized
list_heads.

Noticed when testing with no ccid kernel modules.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP]: Kconfig tidy up

Make CCID2 and CCID3 default to what was selected for DCCP and use the
standard short description for the CCIDs (TCP-Like & TCP-Friendly).

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP]: Make CCID2 be the default

As per the draft. This fixes the build when netfilter dccp components
are built and dccp isn't. Thanks to Reuben Farrelly for reporting
this.

The following changesets will introduce /proc/sys/net/dccp/defaults/
to give more flexibility to DCCP developers and testers while apps
doesn't use setsockopt to specify the desired CCID, etc.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP]: sparse endianness annotations

This also fixes the layout of dccp_hdr short sequence numbers, problem
was not fatal now as we only support long (48 bits) sequence numbers.

Signed-off-by: Andrea Bittau <a.bittau@cs.ucl.ac.uk>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NETFILTER]: Fix skb->nf_bridge lifetime issues

The bridge netfilter code simulates the NF_IP_PRE_ROUTING hook and skips
the real hook by registering with high priority and returning NF_STOP if
skb->nf_bridge is present and the BRNF_NF_BRIDGE_PREROUTING flag is not
set. The flag is only set during the simulated hook.

Because skb->nf_bridge is only freed when the packet is destroyed, the
packet will not only skip the first invocation of NF_IP_PRE_ROUTING, but
in the case of tunnel devices on top of the bridge also all further ones.
Forwarded packets from a bridge encapsulated by a tunnel device and sent
as locally outgoing packet will also still have the incorrect bridge
information from the input path attached.

We already have nf_reset calls on all RX/TX paths of tunnel devices,
so simply reset the nf_bridge field there too. As an added bonus,
the bridge information for locally delivered packets is now also freed
when the packet is queued to a socket.

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

[DCCP] feat: Actually change the CCID upon negotiation

Change the CCID upon successful feature negotiation.

Commiter note: patch mostly rewritten to use the new ccid API.

Signed-off-by: Andrea Bittau <a.bittau@cs.ucl.ac.uk>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP] CCID: Improve CCID infrastructure

1. No need for ->ccid_init nor ->ccid_exit, this is what module_{init,exit}
   does and anynways neither ccid2 nor ccid3 were using it.

2. Rename struct ccid to struct ccid_operations and introduce struct ccid
   with a pointer to ccid_operations and rigth after it the rx or tx
   private state.

3. Remove the pointer to the state of the half connections from struct
   dccp_sock, now its derived thru ccid_priv() from the ccid pointer.

Now we also can implement the setsockopt for changing the CCID easily as
no ccid init routines can affect struct dccp_sock in any way that prevents
other CCIDs from working if a CCID switch operation is asked by apps.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[PKT_SCHED]: Convert sch_red to a classful qdisc

Convert sch_red to a classful qdisc. All qdiscs that maintain accurate
backlog counters are eligible as child qdiscs. When a queue limit larger
than zero is given, a bfifo qdisc is used for backwards compatibility.
Current versions of tc enforce a limit larger than zero, other users
can avoid creating the default qdisc by using zero.

Signed-off-by: Patrick McHardy <kaber@trash.net>
Acked-by: Jamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>

[XFRM]: Add some missing exports.

To fix the case of modular xfrm_user.

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

[XFRM]: Move xfrm_nl to xfrm_state.c from xfrm_user.c

xfrm_user could be modular, and since generic code uses this symbol
now...

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

[XFRM]: Make sure xfrm_replay_timer_handler() is declared early enough.

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

[IPSEC]: Sync series - update selinux

Add new netlink messages to selinux framework

Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPSEC]: Sync series - policy expires

This is similar to the SA expire insertion patch - only it inserts
expires for SP.

Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPSEC]: Sync series - SA expires

This patch allows a user to insert SA expires. This is useful to
do on an HA backup for the case of byte counts but may not be very
useful for the case of time based expiry.

Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPSEC]: Sync series - acquire insert

This introduces a feature similar to the one described in RFC 2367:
"
   ... the application needing an SA sends a PF_KEY
   SADB_ACQUIRE message down to the Key Engine, which then either
   returns an error or sends a similar SADB_ACQUIRE message up to one or
   more key management applications capable of creating such SAs.
   ...
   ...
   The third is where an application-layer consumer of security
   associations (e.g.  an OSPFv2 or RIPv2 daemon) needs a security
   association.

        Send an SADB_ACQUIRE message from a user process to the kernel.

        <base, address(SD), (address(P),) (identity(SD),) (sensitivity,)
          proposal>

        The kernel returns an SADB_ACQUIRE message to registered
          sockets.

        <base, address(SD), (address(P),) (identity(SD),) (sensitivity,)
          proposal>

        The user-level consumer waits for an SADB_UPDATE or SADB_ADD
        message for its particular type, and then can use that
        association by using SADB_GET messages.

 "
An app such as OSPF could then use ipsec KM to get keys

Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPSEC]: Sync series - user

Add xfrm as the user of the core changes

Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPSEC]: Sync series - fast path

Fast path sequence updates that will generate ipsec async
events

Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPSEC]: Sync series - core changes

This patch provides the core functionality needed for sync events
for ipsec. Derived work of Krisztian KOVACS <hidden@balabit.hu>

Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>

[PKT_SCHED]: Keep backlog counter in sch_sfq

Keep backlog counter in SFQ qdisc to make it usable as child qdisc
with RED.

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

[PKT_SCHED]: Restore TBF change semantic

When TBF was converted to a classful qdisc, the semantic of the limit
parameter was broken. On initilization an inner bfifo qdisc is created
for backwards compatibility, when changing parameters however the new
limit is ignored and the current child qdisc remains in place.

Always replace the child qdisc by the default bfifo when limit is above
zero, otherwise don't touch the inner qdisc. Current tc version enforce
a limit above zero, other users can avoid creating the inner qdisc by
using zero.

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

[PKT_SCHED]: Dump child qdisc handle in sch_{atm,dsmark}

A qdisc should set tcm_info to the child qdisc handle in its class
dump function.

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

[PKT_SCHED]: Qdisc drop operation is optional

The drop operation is optional and qdiscs must check if childs support it.

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

[IRDA]: pci_register_driver conversion

This patch converts 2 IrDA drivers pci_module_init() calls to
pci_register_driver().

Signed-off-by: Christophe Lucas <clucas@ifrance.com>
Signed-off-by: Domen Puncer <domen@coderock.org>
Signed-off-by: Samuel Ortiz <samuel@sortiz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IRDA]: sti/cli removal from EP7211 IrDA driver

This patch replaces the deprecated sti/cli routines with the corresponding
spin_lock ones.

Signed-off-by: David chosrova <david.chosrova@libertysurf.fr>
Signed-off-by: Samuel Ortiz <samuel.ortiz@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IRDA]: nsc-ircc: support for yet another Thinkpad IrDA chipset

This patch simply adds support for a variation of the nsc-ircc PC8739x
chipset, found in some IBM Thinkpad laptops.

Signed-off-by: Jean Tourrilhes <jt@hpl.hp.com>
Signed-off-by: Samuel Ortiz <samuel.ortiz@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IRDA]: nsc-ircc: PM update

This patch brings the nsc-ircc code to a more up to date power
management scheme, following the current device model.

Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
Signed-off-by: Rudolf Marek <r.marek@sh.cvut.cz>
Signed-off-by: Samuel Ortiz <samuel.ortiz@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IRDA]: nsc-ircc: ISAPnP support

This enables PnP support for the nsc-ircc chipset.
Since we can't fetch the chipset cfg_base from the PnP layer, we just use
the PnP information as one more hint when probing the chip.

Signed-off-by: Jean Tourrilhes <jt@hpl.hp.com>
Signed-off-by: Samuel Ortiz <samuel.ortiz@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NETLINK]: Add netlink_has_listeners for avoiding unneccessary event message generation

Keep a bitmask of multicast groups with subscribed listeners to let
netlink users check for listeners before generating multicast
messages.

Queries don't perform any locking, which may result in false
positives, it is guaranteed however that any new subscriptions are
visible before bind() or setsockopt() return.

Signed-off-by: Patrick McHardy <kaber@trash.net>
ACKed-by: Jamal Hadi Salim<hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NETFILTER]: ctnetlink: avoid unneccessary event message generation

Avoid unneccessary event message generation by checking for netlink
listeners before building a message.

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

[NETFILTER]: x_tables: replace IPv4/IPv6 policy match by address family independant version

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

[NETFILTER]: Move ip6_masked_addrcmp to include/net/ipv6.h

Replace netfilter's ip6_masked_addrcmp by a more efficient version
in include/net/ipv6.h to make it usable without module dependencies.

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

[NETFILTER]: x_tables: add xt_{match,target} arguments to match/target functions

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

[NETFILTER]: x_tables: pass registered match/target data to match/target functions

This allows to make decisions based on the revision (and address family
with a follow-up patch) at runtime.

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

[NETFILTER]: Convert x_tables matches/targets to centralized error checking

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

[NETFILTER]: Convert ip6_tables matches/targets to centralized error checking

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

[NETFILTER]: Convert arp_tables targets to centralized error checking

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

[NETFILTER]: Convert ip_tables matches/targets to centralized error checking

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

[NETFILTER]: Change {ip,ip6,arp}_tables to use centralized error checking

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

[NETFILTER]: xt_tables: add centralized error checking

Introduce new functions for common match/target checks (private data
size, valid hooks, valid tables and valid protocols) to get more consistent
error reporting and to avoid each module duplicating them.

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

[NETFILTER]: nf_conntrack: use ipv6_addr_equal in nf_ct_reasm

Signed-off-by: Yasuyuki Kozakai <yasuyuki.kozakai@toshiba.co.jp>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NETFILTER]: Fix CID offset bug in PPTP NAT helper debug message

The recent (kernel 2.6.15.1) fix for PPTP NAT helper introduced a
bug - which only appears if DEBUGP is enabled though.

The calculation of the CID offset into a PPTP request struct is
not correct, so that at least not the correct CID is displayed
if DEBUGP is enabled.

This patch corrects CID offset calculation and introduces a #define
for that.

Signed-off-by: Holger Eitzenberger <heitzenberger@astaro.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP] CCID2: Drop sock reference count on timer expiration and reset.

There was a hybrid use of standard timers and sk_timers.  This caused
the reference count of the sock to be incorrect when resetting the RTO
timer.  The sock reference count should now be correct, enabling its
destruction, and allowing the DCCP module to be unloaded.

Signed-off-by: Andrea Bittau <a.bittau@cs.ucl.ac.uk>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>

[DCCP]: Set the default CCID according to kernel config selection

Now CCID2 is the default, as stated in the RFC drafts, but we allow
a config where just CCID3 is built, where CCID3 becomes the default.

Signed-off-by: Ian McDonald <imcdnzl@gmail.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>

[NETFILTER] nf_conntrack: clean up to reduce size of 'struct nf_conn'

This patch moves all helper related data fields of 'struct nf_conn'
into a separate structure 'struct nf_conn_help'.  This new structure
is only present in conntrack entries for which we actually have a
helper loaded.

Also, this patch cleans up the nf_conntrack 'features' mechanism to
resemble what the original idea was: Just glue the feature-specific
data structures at the end of 'struct nf_conn', and explicitly
re-calculate the pointer to it when needed rather than keeping
pointers around.

Saves 20 bytes per conntrack on my x86_64 box. A non-helped conntrack
is 276 bytes. We still need to save another 20 bytes in order to fit
into to target of 256bytes.

Signed-off-by: Harald Welte <laforge@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[BNX2]: include <linux/vmalloc.h>

Include <linux/vmalloc.h> so that it compiles properly on all archs.

Update version to 1.4.38.

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

[TCP]: MTU probing

Implementation of packetization layer path mtu discovery for TCP, based on
the internet-draft currently found at
<http://www.ietf.org/internet-drafts/draft-ietf-pmtud-method-05.txt>.

Signed-off-by: John Heffner <jheffner@psc.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>

[BNX2]: Update version

Update version to 1.4.37.

Add missing flush_scheduled_work() in bnx2_suspend as noted by Jeff
Garzik.

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

[BNX2]: Support larger rx ring sizes (part 2)

Support bigger rx ring sizes (up to 1020) in the rx fast path.

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

[BNX2]: Support larger rx ring sizes (part 1)

Increase maximum receive ring size from 255 to 1020 by supporting
up to 4 linked pages of receive descriptors. To accomodate the
higher memory usage, each physical descriptor page is allocated
separately and the software ring that keeps track of the SKBs and the
DMA addresses is allocated using vmalloc.

Some of the receive-related fields in the bp structure are re-
organized a bit for better locality of reference.

The max. was reduced to 1020 from 4080 after discussion with David
Miller.

This patch contains ring init code changes only. This next patch
contains rx data path code changes.

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

[BNX2]: Fix bug when rx ring is full

Fix the rx code path that does not handle the full rx ring correctly.

When the rx ring is set to the max. size (i.e. 255), the consumer and
producer indices will be the same when completing an rx packet. Fix
the rx code to handle this condition properly.

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

[BNX2]: Add ethtool -d support

Add ETHTOOL_GREGS support.

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

[BNX2]: Reduce register test size

Eliminate some of the registers in ethtool register test to reduce
driver size.

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

[TG3]: Update version and reldate

Update version to 3.50.

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

[TG3]: Support shutdown WoL.

Support WoL during shutdown by calling
tg3_set_power_state(tp, PCI_D3hot) during tg3_close().

Change the power state parameter to pci_power_t type and use
constants defined in pci.h.

Certain ethtool operations cannot be performed after tg3_close()
because the device will go to low power state. Add return -EAGAIN
in such cases where appropriate.

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

[TG3]: Enable TSO by default

Enable TSO by default on newer chips that support TSO in hardware.
Leave TSO off by default on older chips that do firmware TSO because
performance is slightly lower.

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

[TG3]: Add support for 5714S and 5715S

Add support for 5714S and 5715S.

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

[IPV4] fib_rules.c: make struct fib_rules static again

struct fib_rules became global for no good reason.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPCOMP6]: don't check vfree() argument for NULL.

vfree does it's own NULL checking, so checking a pointer before
handing it to vfree is pointless.

Signed-off-by: Jesper Juhl <jesper.juhl@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP]: Initial feature negotiation implementation

Still needs more work, but boots and doesn't crashes, even
does some negotiation!

18:38:52.174934  127.0.0.1.43458 > 127.0.0.1.5001: request <change_l ack_ratio 2, change_r ccid 2, change_l ccid 2>
18:38:52.218526  127.0.0.1.5001 > 127.0.0.1.43458: response <nop, nop, change_l ack_ratio 2, confirm_r ccid 2 2, confirm_l ccid 2 2, confirm_r ack_ratio 2>
18:38:52.185398  127.0.0.1.43458 > 127.0.0.1.5001: <nop, confirm_r ack_ratio 2, ack_vector0 0x00, elapsed_time 212>

:-)

Signed-off-by: Andrea Bittau <a.bittau@cs.ucl.ac.uk>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP] CCID2: Initial CCID2 (TCP-Like) implementation

Original work by Andrea Bittau, Arnaldo Melo cleaned up and fixed several
issues on the merge process.

For now CCID2 was turned the default for all SOCK_DCCP connections, but this
will be remedied soon with the merge of the feature negotiation code.

Signed-off-by: Andrea Bittau <a.bittau@cs.ucl.ac.uk>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP] CCID3: Set the no_feedback_timer fields near init_timer

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP]: Don't alloc ack vector for the control sock

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP] ackvec: Delete all the ack vector records in dccp_ackvec_free

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP] CCID: Allow ccid_{init,exit} to be NULL

Testing if the ccid being instantiated has these methods in
ccid_init().

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP] ackvec: Introduce ack vector records

Based on a patch by Andrea Bittau.

Signed-off-by: Andrea Bittau <a.bittau@cs.ucl.ac.uk>
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[LIST]: Introduce list_for_each_entry_from

For iterating over list of given type continuing from existing point.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPV4]: Use RCU locking in fib_rules.

Signed-off-by: Robert Olsson <robert.olsson@its.uu.se>
Signed-off-by: David S. Miller <davem@davemloft.net>

[LIST]: Introduce list_for_each_entry_safe_from

For iterate over list of given type from existing point safe against removal of
list entry.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP] ackvec: Introduce dccp_ackvec_slab

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP]: Fix error handling in dccp_init

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[DCCP] ackvec: Ditch dccpav_buf_len

Simplifying the code a bit as we're always using DCCP_MAX_ACKVEC_LEN.

Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NETFILTER] nfnetlink_log: add sequence numbers for log events

By using a sequence number for every logged netfilter event, we can
determine from userspace whether logging information was lots somewhere
downstream.

The user has a choice of either having per-instance local sequence
counters, or using a global sequence counter, or both.

Signed-off-by: Harald Welte <laforge@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NETFILTER] NAT sequence adjustment: Save eight bytes per conntrack

This patch reduces the size of 'struct ip_conntrack' on systems with NAT
by eight bytes.  The sequence number delta values can be int16_t, since
we only support one sequence number modification per window anyway, and
one such modification is not going to exceed 32kB ;)

Signed-off-by: Harald Welte <laforge@netfilter.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET]: Do not lose accepted socket when -ENFILE/-EMFILE.

Try to allocate the struct file and an unused file
descriptor before we try to pull a newly accepted
socket out of the protocol layer.

Based upon a patch by Prassana Meda.

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

[NET]: Reduce size of struct sk_buff on 64 bit architectures

Move skb->nf_mark next to skb->tc_index to remove a 4 byte hole between
skb->nfmark and skb->nfct and another one between skb->users and skb->head
when CONFIG_NETFILTER, CONFIG_NET_SCHED and CONFIG_NET_CLS_ACT are enabled.
For all other combinations the size stays the same.

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

[VLAN]: translate IF_OPER_DORMANT to netif_dormant_on()

this patch adds support to the VLAN driver to translate IF_OPER_DORMANT of the
underlying device to netif_dormant_on(). Beside clean state forwarding, this
allows running independant userspace supplicants on both the real device and
the stacked VLAN. It depends on my RFC2863 patch.

Signed-off-by: Stefan Rompf <stefan@loplof.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

[NET] core: add RFC2863 operstate

this patch adds a dormant flag to network devices, RFC2863 operstate derived
from these flags and possibility for userspace interaction. It allows drivers
to signal that a device is unusable for user traffic without disabling
queueing (and therefore the possibility for protocol establishment traffic to
flow) and a userspace supplicant (WPA, 802.1X) to mark a device unusable
without changes to the driver.

It is the result of our long discussion. However I must admit that it
represents what Jamal and I agreed on with compromises towards Krzysztof, but
Thomas and Krzysztof still disagree with some parts. Anyway I think it should
be applied.

Signed-off-by: Stefan Rompf <stefan@loplof.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPV6]: ROUTE: Ensure to accept redirects from nexthop for the target.

It is possible to get redirects from nexthop of "more-specific"
routes.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPV6]: ROUTE: Add accept_ra_rt_info_max_plen sysctl.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPV6]: ROUTE: Flag RTF_DEFAULT for Route Infomation for ::/0.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPV6]: ROUTE: Add experimental support for Route Information Option in RA (RFC4191).

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPV6]: ROUTE: Add router_probe_interval sysctl.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPV6]: ROUTE: Add accept_ra_rtr_pref sysctl.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPV6]: ROUTE: Add Router Reachability Probing (RFC4191).

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

[IPV6]: ROUTE: Add support for Router Preference (RFC4191).

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>