This is a preparatory patch for converting qdisc linked list into a
hashtable. As we'll need to include hashtable.h in netdevice.h, we first
have to make sure that this will not introduce symbol conflicts for any of
the netdevice.h users.
Reviewed-by: NCong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: NJiri Kosina <jkosina@suse.cz>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Based on RFC3376 5.1 and RFC3810 6.1

   If the per-interface listening change that triggers the new report is
   a filter mode change, then the next [Robustness Variable] State
   Change Reports will include a Filter Mode Change Record.  This
   applies even if any number of source list changes occur in that
   period.

   Old State         New State         State Change Record Sent
   ---------         ---------         ------------------------
   INCLUDE (A)       EXCLUDE (B)       TO_EX (B)
   EXCLUDE (A)       INCLUDE (B)       TO_IN (B)

So we should not send source-list change if there is a filter-mode change.

Here are two scenarios:
1. Group deleted and filter mode is EXCLUDE, which means we need send a
   TO_IN { }.
2. Not group deleted, but has pcm->crcount, which means we need send a
   normal filter-mode-change.

At the same time, if the type is ALLOW or BLOCK, and have psf->sf_crcount,
we stop add records and decrease sf_crcount directly

Reference: https://www.ietf.org/mail-archive/web/magma/current/msg01274.htmlSigned-off-by: NHangbin Liu <liuhangbin@gmail.com>
Acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Using list_move() instead of list_del() + list_add().
Signed-off-by: NWei Yongjun <weiyj.lk@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently lastuse is updated on entry creation and cache hit, but it should
also be updated on entry change. Since both on add and update the ttl array
is updated we can simply update the lastuse in ipmr_update_thresholds.
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
CC: Roopa Prabhu <roopa@cumulusnetworks.com>
CC: Donald Sharp <sharpd@cumulusnetworks.com>
CC: David S. Miller <davem@davemloft.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

After a6127697 ("udp: prevent bugcheck if filter truncates packet
too much"), there followed various other fixes for similar cases such
as f4979fce ("rose: limit sk_filter trim to payload").

Latter introduced a new helper sk_filter_trim_cap(), where we can pass
the trim limit directly to the socket filter handling. Make use of it
here as well with sizeof(struct udphdr) as lower cap limit and drop the
extra skb->len test in UDP's input path.
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Cc: Willem de Bruijn <willemb@google.com>
Acked-by: NWillem de Bruijn <willemb@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Default kernel behavior is to delete IPv6 addresses on link
down, which entails deletion of the multicast and the
subnet-router anycast addresses. These deletions do not
happen with sysctl setting to keep global IPv6 addresses on
link down, so every link down/up causes an increment of the
anycast and multicast refcounts. These bogus refcounts may
stop these addrs from being removed on subsequent calls to
delete them. The solution is to leave the groups for the
multicast and subnet anycast on link down for the callflow
when global IPv6 addresses are kept.

Fixes: f1705ec1 ("net: ipv6: Make address flushing on ifdown optional")
Signed-off-by: NMike Manning <mmanning@brocade.com>
Acked-by: NDavid Ahern <dsa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The dummy ruleset I used to test the original validation change was broken,
most rules were unreachable and were not tested by mark_source_chains().

In some cases rulesets that used to load in a few seconds now require
several minutes.

sample ruleset that shows the behaviour:

echo "*filter"
for i in $(seq 0 100000);do
        printf ":chain_%06x - [0:0]\n" $i
done
for i in $(seq 0 100000);do
   printf -- "-A INPUT -j chain_%06x\n" $i
   printf -- "-A INPUT -j chain_%06x\n" $i
   printf -- "-A INPUT -j chain_%06x\n" $i
done
echo COMMIT

[ pipe result into iptables-restore ]

This ruleset will be about 74mbyte in size, with ~500k searches
though all 500k[1] rule entries. iptables-restore will take forever
(gave up after 10 minutes)

Instead of always searching the entire blob for a match, fill an
array with the start offsets of every single ipt_entry struct,
then do a binary search to check if the jump target is present or not.

After this change ruleset restore times get again close to what one
gets when reverting 36472341 (~3 seconds on my workstation).

[1] every user-defined rule gets an implicit RETURN, so we get
300k jumps + 100k userchains + 100k returns -> 500k rule entries

Fixes: 36472341 ("netfilter: x_tables: validate targets of jumps")
Reported-by: NJeff Wu <wujiafu@gmail.com>
Tested-by: NJeff Wu <wujiafu@gmail.com>
Signed-off-by: NFlorian Westphal <fw@strlen.de>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

In preparation for hardware offloading of ipmr/ip6mr we need an
interface that allows to check (and later update) the age of entries.
Relying on stats alone can show activity but not actual age of the entry,
furthermore when there're tens of thousands of entries a lot of the
hardware implementations only support "hit" bits which are cleared on
read to denote that the entry was active and shouldn't be aged out,
these can then be naturally translated into age timestamp and will be
compatible with the software forwarding age. Using a lastuse entry doesn't
affect performance because the members in that cache line are written to
along with the age.
Since all new users are encouraged to use ipmr via netlink, this is
exported via the RTA_EXPIRES attribute.
Also do a minor local variable declaration style adjustment - arrange them
longest to shortest.
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
CC: Roopa Prabhu <roopa@cumulusnetworks.com>
CC: Shrijeet Mukherjee <shm@cumulusnetworks.com>
CC: Satish Ashok <sashok@cumulusnetworks.com>
CC: Donald Sharp <sharpd@cumulusnetworks.com>
CC: David S. Miller <davem@davemloft.net>
CC: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
CC: James Morris <jmorris@namei.org>
CC: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>
CC: Patrick McHardy <kaber@trash.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If socket filter truncates an udp packet below the length of UDP header
in udpv6_queue_rcv_skb() or udp_queue_rcv_skb(), it will trigger a
BUG_ON in skb_pull_rcsum(). This BUG_ON (and therefore a system crash if
kernel is configured that way) can be easily enforced by an unprivileged
user which was reported as CVE-2016-6162. For a reproducer, see
http://seclists.org/oss-sec/2016/q3/8

Fixes: e6afc8ac ("udp: remove headers from UDP packets before queueing")
Reported-by: NMarco Grassi <marco.gra@gmail.com>
Signed-off-by: NMichal Kubecek <mkubecek@suse.cz>
Acked-by: NEric Dumazet <edumazet@google.com>
Acked-by: NWillem de Bruijn <willemb@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

All inet6_netconf_notify_devconf() callers are in process context,
so we can use GFP_KERNEL allocations if we take care of not holding
a rwlock while not needed in ip6mr (we hold RTNL there)

Fixes: d67b8c61 ("netconf: advertise mc_forwarding status")
Fixes: f3a1bfb1 ("rtnl/ipv6: use netconf msg to advertise forwarding status")
Signed-off-by: NEric Dumazet <edumazet@google.com>
Cc: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Acked-by: NNicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Extend the SIT driver to support MPLS over IPv4. This implementation
extends existing support for IPv6 over IPv4 and IPv4 over IPv4.
Signed-off-by: NSimon Horman <simon.horman@netronome.com>
Reviewed-by: NDinan Gunawardena <dinan.gunawardena@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

It was first reported and reproduced by Petr (thanks!) in
https://bugzilla.kernel.org/show_bug.cgi?id=119581

free_percpu(rt->rt6i_pcpu) used to always happen in ip6_dst_destroy().

However, after fixing a deadlock bug in
commit 9c7370a1 ("ipv6: Fix a potential deadlock when creating pcpu rt"),
free_percpu() is not called before setting non_pcpu_rt->rt6i_pcpu to NULL.

It is worth to note that rt6i_pcpu is protected by table->tb6_lock.

kmemleak somehow did not report it.  We nailed it down by
observing the pcpu entries in /proc/vmallocinfo (first suggested
by Hannes, thanks!).
Signed-off-by: NMartin KaFai Lau <kafai@fb.com>
Fixes: 9c7370a1 ("ipv6: Fix a potential deadlock when creating pcpu rt")
Reported-by: NPetr Novopashenniy <pety@rusnet.ru>
Tested-by: NPetr Novopashenniy <pety@rusnet.ru>
Acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Cc: Petr Novopashenniy <pety@rusnet.ru>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

netfilter uses multiple FWINV #defines with identical form that hide a
specific structure variable and dereference it with a invflags member.

$ git grep "#define FWINV"
include/linux/netfilter_bridge/ebtables.h:#define FWINV(bool,invflg) ((bool) ^ !!(info->invflags & invflg))
net/bridge/netfilter/ebtables.c:#define FWINV2(bool, invflg) ((bool) ^ !!(e->invflags & invflg))
net/ipv4/netfilter/arp_tables.c:#define FWINV(bool, invflg) ((bool) ^ !!(arpinfo->invflags & (invflg)))
net/ipv4/netfilter/ip_tables.c:#define FWINV(bool, invflg) ((bool) ^ !!(ipinfo->invflags & (invflg)))
net/ipv6/netfilter/ip6_tables.c:#define FWINV(bool, invflg) ((bool) ^ !!(ip6info->invflags & (invflg)))
net/netfilter/xt_tcpudp.c:#define FWINVTCP(bool, invflg) ((bool) ^ !!(tcpinfo->invflags & (invflg)))

Consolidate these macros into a single NF_INVF macro.

Miscellanea:

o Neaten the alignment around these uses
o A few lines are > 80 columns for intelligibility
Signed-off-by: NJoe Perches <joe@perches.com>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

No need for a special case to handle NF_INET_POST_ROUTING, this is
basically the same handling as for prerouting, input, forward.
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

Some arches have virtually mapped kernel stacks, or will soon have.

tcp_md5_hash_header() uses an automatic variable to copy tcp header
before mangling th->check and calling crypto function, which might
be problematic on such arches.

David says that using percpu storage is also problematic on non SMP
builds.

Just use kmalloc() to allocate scratch areas.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Reported-by: NAndy Lutomirski <luto@amacapital.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This fixes wrong-interface signaling on 32-bit platforms for entries
created when jiffies > 2^31 + MFC_ASSERT_THRESH.
Signed-off-by: NTom Goff <thomas.goff@ll.mit.edu>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This works in exactly the same way as the CIPSO label cache.
The idea is to allow the lsm to cache the result of a secattr
lookup so that it doesn't need to perform the lookup for
every skbuff.

It introduces two sysctl controls:
 calipso_cache_enable - enables/disables the cache.
 calipso_cache_bucket_size - sets the size of a cache bucket.
Signed-off-by: NHuw Davies <huw@codeweavers.com>
Signed-off-by: NPaul Moore <paul@paul-moore.com>

Lengths, checksum and the DOI are checked.  Checking of the
level and categories are left for the socket layer.

CRC validation is performed in the calipso module to avoid
unconditionally linking crc_ccitt() into ipv6.
Signed-off-by: NHuw Davies <huw@codeweavers.com>
Signed-off-by: NPaul Moore <paul@paul-moore.com>

In some cases, the lsm needs to add the label to the skbuff directly.
A NF_INET_LOCAL_OUT IPv6 hook is added to selinux to match the IPv4
behaviour.  This allows selinux to label the skbuffs that it requires.
Signed-off-by: NHuw Davies <huw@codeweavers.com>
Signed-off-by: NPaul Moore <paul@paul-moore.com>

Signed-off-by: NHuw Davies <huw@codeweavers.com>
Signed-off-by: NPaul Moore <paul@paul-moore.com>

Request sockets need to have a label that takes into account the
incoming connection as well as their parent's label.  This is used
for the outgoing SYN-ACK and for their child full-socket.
Signed-off-by: NHuw Davies <huw@codeweavers.com>
Signed-off-by: NPaul Moore <paul@paul-moore.com>

If set, these will take precedence over the parent's options during
both sending and child creation.  If they're not set, the parent's
options (if any) will be used.

This is to allow the security_inet_conn_request() hook to modify the
IPv6 options in just the same way that it already may do for IPv4.
Signed-off-by: NHuw Davies <huw@codeweavers.com>
Signed-off-by: NPaul Moore <paul@paul-moore.com>

CALIPSO is a hop-by-hop IPv6 option.  A lot of this patch is based on
the equivalent CISPO code.  The main difference is due to manipulating
the options in the hop-by-hop header.
Signed-off-by: NHuw Davies <huw@codeweavers.com>
Signed-off-by: NPaul Moore <paul@paul-moore.com>

The functionality is equivalent to ipv6_renew_options() except
that the newopt pointer is in kernel, not user, memory

The kernel memory implementation will be used by the CALIPSO network
labelling engine, which needs to be able to set IPv6 hop-by-hop
options.
Signed-off-by: NHuw Davies <huw@codeweavers.com>
Signed-off-by: NPaul Moore <paul@paul-moore.com>

Remove a specified DOI through the NLBL_CALIPSO_C_REMOVE command.
It requires the attribute:
 NLBL_CALIPSO_A_DOI.
Signed-off-by: NHuw Davies <huw@codeweavers.com>
Signed-off-by: NPaul Moore <paul@paul-moore.com>

Enumerate the DOI list through the NLBL_CALIPSO_C_LISTALL command.
It takes no attributes.
Signed-off-by: NHuw Davies <huw@codeweavers.com>
Signed-off-by: NPaul Moore <paul@paul-moore.com>

Query a specified DOI through the NLBL_CALIPSO_C_LIST command.
It requires the attribute:
 NLBL_CALIPSO_A_DOI.

The reply will contain:
 NLBL_CALIPSO_A_MTYPE
Signed-off-by: NHuw Davies <huw@codeweavers.com>
Signed-off-by: NPaul Moore <paul@paul-moore.com>

CALIPSO is a packet labelling protocol for IPv6 which is very similar
to CIPSO.  It is specified in RFC 5570.  Much of the code is based on
the current CIPSO code.

This adds support for adding passthrough-type CALIPSO DOIs through the
NLBL_CALIPSO_C_ADD command.  It requires attributes:

 NLBL_CALIPSO_A_TYPE which must be CALIPSO_MAP_PASS.
 NLBL_CALIPSO_A_DOI.

In passthrough mode the CALIPSO engine will map MLS secattr levels
and categories directly to the packet label.

At this stage, the major difference between this and the CIPSO
code is that IPv6 may be compiled as a module.  To allow for
this the CALIPSO functions are registered at module init time.
Signed-off-by: NHuw Davies <huw@codeweavers.com>
Signed-off-by: NPaul Moore <paul@paul-moore.com>

with the commit 8c14586f ("net: ipv6: Use passed in table for
nexthop lookups"), net hop lookup is first performed on route creation
in the passed-in table.
However device match is not enforced in table lookup, so the found
route can be later discarded due to egress device mismatch and no
global lookup will be performed.
This cause the following to fail:

ip link add dummy1 type dummy
ip link add dummy2 type dummy
ip link set dummy1 up
ip link set dummy2 up
ip route add 2001:db8:8086::/48 dev dummy1 metric 20
ip route add 2001:db8:d34d::/64 via 2001:db8:8086::2 dev dummy1 metric 20
ip route add 2001:db8:8086::/48 dev dummy2 metric 21
ip route add 2001:db8:d34d::/64 via 2001:db8:8086::2 dev dummy2 metric 21
RTNETLINK answers: No route to host

This change fixes the issue enforcing device lookup in
ip6_nh_lookup_table()

v1->v2: updated commit message title

Fixes: 8c14586f ("net: ipv6: Use passed in table for nexthop lookups")
Reported-and-tested-by: NBeniamino Galvani <bgalvani@redhat.com>
Signed-off-by: NPaolo Abeni <pabeni@redhat.com>
Acked-by: NDavid Ahern <dsa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When receiving an ICMPv4 message containing extensions as
defined in RFC 4884, and translating it to ICMPv6 at SIT
or GRE tunnel, we need some extra manipulation in order
to properly forward the extensions.

This patch only takes care of Time Exceeded messages as they
are the ones that typically carry information from various
routers in a fabric during a traceroute session.

It also avoids complex skb logic if the data_len is not
a multiple of 8.

RFC states :

   The "original datagram" field MUST contain at least 128 octets.
   If the original datagram did not contain 128 octets, the
   "original datagram" field MUST be zero padded to 128 octets.

In practice routers use 128 bytes of original datagram, not more.

Initial translation was added in commit ca15a078
("sit: generate icmpv6 error when receiving icmpv4 error")
Signed-off-by: NEric Dumazet <edumazet@google.com>
Cc: Oussama Ghorbel <ghorbel@pivasoftware.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

For better traceroute/mtr support for SIT and GRE tunnels,
we translate IPV4 ICMP ICMP_TIME_EXCEEDED to ICMPV6_TIME_EXCEED

We also have to translate the IPv4 source IP address of ICMP
message to IPv6 v4mapped.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We want to use this helper from GRE as well, so this is
the time to move it in net/ipv6/icmp.c

Also add a @nhs parameter, since SIT and GRE have different
values for the header(s) to skip.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

SIT or GRE tunnels might want to translate an IPV4 address
into a v4mapped one when translating ICMP to ICMPv6.

This patch adds the parameter to icmp6_send() but
does not change icmpv6_send() signature.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

IPv6 version of 3f2fb9a8 ("net: l3mdev: address selection should only
consider devices in L3 domain") and the follow up commit, a17b693cdd876
("net: l3mdev: prefer VRF master for source address selection").

That is, if outbound device is given then the address preference order
is an address from that device, an address from the master device if it
is enslaved, and then an address from a device in the same L3 domain.
Signed-off-by: NDavid Ahern <dsa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

IPv6 source address selection needs to consider the real egress route.
Similar to IPv4 implement a get_saddr6 method which is called if
source address has not been set.  The get_saddr6 method does a full
lookup which means pulling a route from the VRF FIB table and properly
considering linklocal/multicast destination addresses. Lookup failures
(eg., unreachable) then cause the source address selection to fail
which gets propagated back to the caller.
Signed-off-by: NDavid Ahern <dsa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

VRF driver needs access to ip6_route_get_saddr code. Since it does
little beyond ipv6_dev_get_saddr and ipv6_dev_get_saddr is already
exported for modules move ip6_route_get_saddr to the header as an
inline.

Code move only; no functional change.
Signed-off-by: NDavid Ahern <dsa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Modern C standards expect the '__inline__' keyword to come before the return
type in a declaration, and we get a couple of warnings for this with "make W=1"
in the xfrm{4,6}_policy.c files:

net/ipv6/xfrm6_policy.c:369:1: error: 'inline' is not at beginning of declaration [-Werror=old-style-declaration]
 static int inline xfrm6_net_sysctl_init(struct net *net)
net/ipv6/xfrm6_policy.c:374:1: error: 'inline' is not at beginning of declaration [-Werror=old-style-declaration]
 static void inline xfrm6_net_sysctl_exit(struct net *net)
net/ipv4/xfrm4_policy.c:339:1: error: 'inline' is not at beginning of declaration [-Werror=old-style-declaration]
 static int inline xfrm4_net_sysctl_init(struct net *net)
net/ipv4/xfrm4_policy.c:344:1: error: 'inline' is not at beginning of declaration [-Werror=old-style-declaration]
 static void inline xfrm4_net_sysctl_exit(struct net *net)
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Since 32b8a8e5 ("sit: add IPv4 over IPv4 support")
ipip6_err() may be called for packets whose IP protocol is
IPPROTO_IPIP as well as those whose IP protocol is IPPROTO_IPV6.

In the case of IPPROTO_IPIP packets the correct protocol value is not
passed to ipv4_update_pmtu() or ipv4_redirect().

This patch resolves this problem by using the IP protocol of the packet
rather than a hard-coded value. This appears to be consistent
with the usage of the protocol of a packet by icmp_socket_deliver()
the caller of ipip6_err().

I was able to exercise the redirect case by using a setup where an ICMP
redirect was received for the destination of the encapsulated packet.
However, it appears that although incorrect the protocol field is not used
in this case and thus no problem manifests.  On inspection it does not
appear that a problem will manifest in the fragmentation needed/update pmtu
case either.

In short I believe this is a cosmetic fix. None the less, the use of
IPPROTO_IPV6 seems wrong and confusing.
Reviewed-by: NDinan Gunawardena <dinan.gunawardena@netronome.com>
Signed-off-by: NSimon Horman <simon.horman@netronome.com>
Acked-by: NYOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

1) gre_parse_header() can be called from gre_err()

   At this point transport header points to ICMP header, not the inner
header.

2) We can not really change transport header as ipgre_err() will later
assume transport header still points to ICMP header (using icmp_hdr())

3) pskb_may_pull() logic in gre_parse_header() really works
  if we are interested at zone pointed by skb->data

4) As Jiri explained in commit b7f8fe25 ("gre: do not pull header in
ICMP error processing") we should not pull headers in error handler.

So this fix :

A) changes gre_parse_header() to use skb->data instead of
skb_transport_header()

B) Adds a nhs parameter to gre_parse_header() so that we can skip the
not pulled IP header from error path.
  This offset is 0 for normal receive path.

C) remove obsolete IPV6 includes
Signed-off-by: NEric Dumazet <edumazet@google.com>
Cc: Tom Herbert <tom@herbertland.com>
Cc: Maciej Żenczykowski <maze@google.com>
Cc: Jiri Benc <jbenc@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The algorithm for checksum neutral mapping is incorrect. This problem
was being hidden since we were previously always performing checksum
offload on the translated addresses and only with IPv6 HW csum.
Enabling an ILA router shows the issue.

Corrected algorithm:

old_loc is the original locator in the packet, new_loc is the value
to overwrite with and is found in the lookup table. old_flag is
the old flag value (zero of CSUM_NEUTRAL_FLAG) and new_flag is
then (old_flag ^ CSUM_NEUTRAL_FLAG) & CSUM_NEUTRAL_FLAG.

Need SUM(new_id + new_flag + diff) == SUM(old_id + old_flag) for
checksum neutral translation.

Solving for diff gives:

diff = (old_id - new_id) + (old_flag - new_flag)

compute_csum_diff8(new_id, old_id) gives old_id - new_id

If old_flag is set
   old_flag - new_flag = old_flag = CSUM_NEUTRAL_FLAG
Else
   old_flag - new_flag = -new_flag = ~CSUM_NEUTRAL_FLAG

Tested:
  - Implemented a user space program that creates random addresses
    and random locators to overwrite. Compares the checksum over
    the address before and after translation (must always be equal)
  - Enabled ILA router and showed proper operation.
Signed-off-by: NTom Herbert <tom@herbertland.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>