Some users observed that "least connection" distribution algorithm doesn't
handle well bursts of TCP connections from reconnecting clients after
a node or network failure.

This is because the algorithm counts active connection as worth 256
inactive ones where for TCP, "active" only means TCP connections in
ESTABLISHED state. In case of a connection burst, new connections are
handled before previous ones have finished the three way handshaking so
that all are still counted as "inactive", i.e. cheap ones. The become
"active" quickly but at that time, all of them are already assigned to one
real server (or few), resulting in highly unbalanced distribution.

Address this by counting the "pre-established" states as "active".
Signed-off-by: NMichal Kubecek <mkubecek@suse.cz>
Acked-by: NJulian Anastasov <ja@ssi.bg>
Signed-off-by: NSimon Horman <horms@verge.net.au>

It is hard to unbind nf-logger:

  echo NONE > /proc/sys/net/netfilter/nf_log/0
  bash: echo: write error: No such file or directory

  sysctl -w net.netfilter.nf_log.0=NONE
  sysctl: setting key "net.netfilter.nf_log.0": No such file or directory
  net.netfilter.nf_log.0 = NONE

You need explicitly send '\0', for instance like:

  echo -e "NONE\0" > /proc/sys/net/netfilter/nf_log/0

That seem to be strange, so fix it using proc_dostring.

Now it works fine:
   modprobe nfnetlink_log
   echo nfnetlink_log > /proc/sys/net/netfilter/nf_log/0
   cat /proc/sys/net/netfilter/nf_log/0
   nfnetlink_log
   echo NONE > /proc/sys/net/netfilter/nf_log/0
   cat /proc/sys/net/netfilter/nf_log/0
   NONE

v2: add missed error check for proc_dostring
Signed-off-by: NPavel Tikhomirov <ptikhomirov@virtuozzo.com>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

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>

This option was removed in commit 47dcf0cb ("[NET]: Rethink mark field
in struct flowi").
Signed-off-by: NMoritz Sichert <moritz+linux@sichert.me>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

There are code duplications of a masked ethernet address comparison here
so make it a separate function instead.

Miscellanea:

o Neaten alignment of FWINV macro uses to make it clearer for the reader
Signed-off-by: NJoe Perches <joe@perches.com>
Acked-by: NDavid S. Miller <davem@davemloft.net>
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>

Introduce a new configuration option for this expression, which allows users
to invert the logic of set lookups.

In _init() we will now return EINVAL if NFT_LOOKUP_F_INV is in anyway
related to a map lookup.

The code in the _eval() function has been untangled and updated to sopport the
XOR of options, as we should consider 4 cases:
 * lookup false, invert false -> NFT_BREAK
 * lookup false, invert true -> return w/o NFT_BREAK
 * lookup true, invert false -> return w/o NFT_BREAK
 * lookup true, invert true -> NFT_BREAK
Signed-off-by: NArturo Borrero Gonzalez <arturo.borrero.glez@gmail.com>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

This flag was introduced to restore rulesets from the new netdev
family, but since 5ebe0b0e ("netfilter: nf_tables: destroy
basechain and rules on netdevice removal") the ruleset is released
once the netdev is gone.

This also removes nft_register_basechain() and
nft_unregister_basechain() since they have no clients anymore after
this rework.
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

No need to restrict this to module parameter.

We export a copy of the real hash size -- when user alters the value we
allocate the new table, copy entries etc before we update the real size
to the requested one.

This is also needed because the real size is used by concurrent readers
and cannot be changed without synchronizing the conntrack generation
seqcnt.

We only allow changing this value from the initial net namespace.

Tested using http-client-benchmark vs. httpterm with concurrent

while true;do
 echo $RANDOM > /proc/sys/net/netfilter/nf_conntrack_buckets
done
Signed-off-by: NFlorian Westphal <fw@strlen.de>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

New elements are inactive in the preparation phase, and its
NFT_SET_ELEM_BUSY_MASK flag is set on.

This busy flag doesn't allow us to delete it from the same transaction,
following a sequence like:

	begin transaction
	add element X
	delete element X
	end transaction

This sequence is valid and may be triggered by robots. To resolve this
problem, allow deactivating elements that are active in the current
generation (ie. those that has been just added in this batch).
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

set->ops->deactivate() is invoked from nft_del_setelem() that happens
from the transaction path, so we have to check if the object is active
in the next generation, not the current.
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

Similar to ("netfilter: nf_tables: add generation mask to tables").
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

Similar to ("netfilter: nf_tables: add generation mask to tables").
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

This patch addresses two problems:

1) The netlink dump is inconsistent when interfering with an ongoing
   transaction update for several reasons:

1.a) We don't honor the internal NFT_TABLE_INACTIVE flag, and we should
     be skipping these inactive objects in the dump.

1.b) We perform speculative deletion during the preparation phase, that
     may result in skipping active objects.

1.c) The listing order changes, which generates noise when tracking
     incremental ruleset update via tools like git or our own
     testsuite.

2) We don't allow to add and to update the object in the same batch,
   eg. add table x; add table x { flags dormant\; }.

In order to resolve these problems:

1) If the user requests a deletion, the object becomes inactive in the
   next generation. Then, ignore objects that scheduled to be deleted
   from the lookup path, as they will be effectively removed in the
   next generation.

2) From the get/dump path, if the object is not currently active, we
   skip it.

3) Support 'add X -> update X' sequence from a transaction.

After this update, we obtain a consistent list as long as we stay
in the same generation. The userspace side can detect interferences
through the generation counter so it can restart the dumping.
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

Thus, we can reuse these to check the genmask of any object type, not
only rules. This is required now that tables, chain and sets will get a
generation mask field too in follow up patches.
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

li->u.ulog.copy_len is currently ignored by the kernel, we should truncate
the packet to either li->u.ulog.copy_len (if set) or copy_range before
sending it to userspace. 0 is a valid input for copy_len, so add a new
flag to indicate whether this was option was specified by the user or not.

Add two flags to indicate whether nflog-size/copy_len was set or not.
XT_NFLOG_F_COPY_LEN is for XT_NFLOG and NFLOG_F_COPY_LEN for nfnetlink_log

On the userspace side, this was initially represented by the option
nflog-range, this will be replaced by --nflog-size now. --nflog-range would
still exist but does not do anything.
Reported-by: NJoe Dollard <jdollard@akamai.com>
Reviewed-by: NJosh Hunt <johunt@akamai.com>
Signed-off-by: NVishwanath Pai <vpai@akamai.com>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

In iptables, if the user add a rule to send tcp RST and specify the
non-TCP protocol, such as UDP, kernel will reject this request. But
in nftables, this validity check only occurs in nft tool, i.e. only
in userspace.

This means that user can add such a rule like follows via nfnetlink:
  "nft add rule filter forward ip protocol udp reject with tcp reset"

This will generate some confusing tcp RST packets. So we should send
tcp RST only when it is TCP packet.
Signed-off-by: NLiping Zhang <liping.zhang@spreadtrum.com>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

Making this work is a little tricky as it really isn't kosher to
change the xt_owner_match_info in a check function.

Without changing xt_owner_match_info we need to know the user
namespace the uids and gids are specified in.  In the common case
net->user_ns == current_user_ns().  Verify net->user_ns ==
current_user_ns() in owner_check so we can later assume it in
owner_mt.

In owner_check also verify that all of the uids and gids specified are
in net->user_ns and that the expected min/max relationship exists
between the uids and gids in xt_owner_match_info.

In owner_mt get the network namespace from the outgoing socket, as this
must be the same network namespace as the netfilter rules, and use that
network namespace to find the user namespace the uids and gids in
xt_match_owner_info are encoded in.  Then convert from their encoded
from into the kernel internal format for uids and gids and perform the
owner match.

Similar to ping_group_range, this code does not try to detect
noncontiguous UID/GID ranges.
Signed-off-by: N"Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: NKevin Cernekee <cernekee@chromium.org>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

Curently we store zone information as a conntrack extension.
This has one drawback: for every lookup we need to fetch the zone data
from the extension area.

This change place the zone data directly into the main conntrack object
structure and then removes the zone conntrack extension.

The zone data is just 4 bytes, it fits into a padding hole before
the tuplehash info, so we do not even increase the nf_conn structure size.
Signed-off-by: NFlorian Westphal <fw@strlen.de>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

If 'logger' was NULL, there would be a direct jump to the label 'out',
since it has already been checked for NULL, remove this unnecessary
check.
Signed-off-by: NShivani Bhardwaj <shivanib134@gmail.com>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

Those comparisions are useless in case of ZONES=n; all conntracks
will reside in the same zone by definition.
Signed-off-by: NFlorian Westphal <fw@strlen.de>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

increases struct size by 32 bytes (288 -> 320), but it is the right thing,
else any attempt to (re-)arrange nf_conn members by cacheline won't work.
Signed-off-by: NFlorian Westphal <fw@strlen.de>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

Consider such situation, if nf_log_ipv4 kernel module is not installed,
and the user add a following iptables rule:
  # iptables -t raw -I PREROUTING -j TRACE

There will be no trace log generated until the user install nf_log_ipv4
module manully. So we should add request related nf_log module
appropriately here.
Signed-off-by: NLiping Zhang <liping.zhang@spreadtrum.com>
Acked-by: NFlorian Westphal <fw@strlen.de>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

When we request NFPROTO_INET, it means both NFPROTO_IPV4 and NFPROTO_IPV6.
Signed-off-by: NLiping Zhang <liping.zhang@spreadtrum.com>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

Since we cannot make sure that the 'hook_mask' will always be none
zero here. If it equals to zero, the num_hooks will be zero too,
and then kmalloc() will return ZERO_SIZE_PTR, which is (void *)16.

Then the following error check will fails:
  ops = kmalloc(sizeof(*ops) * num_hooks, GFP_KERNEL);
  if (ops == NULL)
          return ERR_PTR(-ENOMEM);

So this patch will fix this with just doing the zero check before
kmalloc() is called.

Maybe the case above will never happen here, but in theory.
Signed-off-by: NXiubo Li <lixiubo@cmss.chinamobile.com>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

The expectation table is not duplicated per net namespace anymore, so we can move
the expectation table and conntrack table iteration out of the per-net loop.
Signed-off-by: NFlorian Westphal <fw@strlen.de>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

checkpatch produces data type 'checks'.

This patch amends them by changing, for example:
uint8_t -> u8
Signed-off-by: NTobin C Harding <me@tobin.cc>
Signed-off-by: NPablo Neira Ayuso <pablo@netfilter.org>

David Ahern says:

====================
net: vrf: Add support for local traffic to local addresses

Add support for locally originated traffic to VRF-local addresses,
be it addresses on enslaved devices or addresses on the VRF device:

$ ip addr show dev red
33: red: <NOARP,MASTER,UP,LOWER_UP> mtu 65536 qdisc pfifo_fast state UP group default qlen 1000
    link/ether be:00:53:b5:e4:25 brd ff:ff:ff:ff:ff:ff
    inet 1.1.1.1/32 scope global red
       valid_lft forever preferred_lft forever
    inet6 1111:1::1/128 scope global
       valid_lft forever preferred_lft forever

$ ip addr show dev eth1
3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master red state UP group default qlen 1000
    link/ether 02:e0:f9:79:34:bd brd ff:ff:ff:ff:ff:ff
    inet 10.100.1.1/24 brd 10.100.1.255 scope global eth1
       valid_lft forever preferred_lft forever
    inet6 2100:1::1/120 scope global
       valid_lft forever preferred_lft forever
    inet6 fe80::e0:f9ff:fe79:34bd/64 scope link
       valid_lft forever preferred_lft forever

$ ping -c1 -I red 10.100.1.1
    ping: Warning: source address might be selected on device other than red.
    PING 10.100.1.1 (10.100.1.1) from 10.100.1.1 red: 56(84) bytes of data.
    64 bytes from 10.100.1.1: icmp_seq=1 ttl=64 time=0.057 ms

$ ping -c1 -I red 1.1.1.1
PING 1.1.1.1 (1.1.1.1) from 1.1.1.1 red: 56(84) bytes of data.
64 bytes from 1.1.1.1: icmp_seq=1 ttl=64 time=0.136 ms

--- 1.1.1.1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.136/0.136/0.136/0.000 ms

$ ping6 -c1 -I red  2100:1::1
ping6: Warning: source address might be selected on device other than red.
PING 2100:1::1(2100:1::1) from 2100:1::1 red: 56 data bytes
64 bytes from 2100:1::1: icmp_seq=1 ttl=64 time=0.167 ms

--- 2100:1::1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.167/0.167/0.167/0.000 ms

$ ping6 -c1 -I red 1111::1
PING 1111::1(1111::1) from 1111:1::1 red: 56 data bytes
64 bytes from 1111::1: icmp_seq=1 ttl=64 time=0.187 ms

--- 1111::1 ping statistics ---
1 packets transmitted, 1 received, 0% packet loss, time 0ms
rtt min/avg/max/mdev = 0.187/0.187/0.187/0.000 ms

This change also enables use of loopback address on the VRF device:
$ ip addr add dev red 127.0.0.1/8

$ ping -c1 -I red 127.0.0.1
PING 127.0.0.1 (127.0.0.1) from 127.0.0.1 red: 56(84) bytes of data.
64 bytes from 127.0.0.1: icmp_seq=1 ttl=64 time=0.058 ms
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add support for locally originated traffic to VRF-local IPv6 addresses.
Similar to IPv4 a local dst is set on the skb and the packet is
reinserted with a call to netif_rx. With this patch, ping, tcp and udp
packets to a local IPv6 address are successfully routed:

    $ ip addr show dev eth1
    4: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master red state UP group default qlen 1000
        link/ether 02:e0:f9:1c:b9:74 brd ff:ff:ff:ff:ff:ff
        inet 10.100.1.1/24 brd 10.100.1.255 scope global eth1
           valid_lft forever preferred_lft forever
        inet6 2100:1::1/120 scope global
           valid_lft forever preferred_lft forever
        inet6 fe80::e0:f9ff:fe1c:b974/64 scope link
           valid_lft forever preferred_lft forever

    $ ping6 -c1 -I red 2100:1::1
    ping6: Warning: source address might be selected on device other than red.
    PING 2100:1::1(2100:1::1) from 2100:1::1 red: 56 data bytes
    64 bytes from 2100:1::1: icmp_seq=1 ttl=64 time=0.098 ms

ip6_input is exported so the VRF driver can use it for the dst input
function. The dst_alloc function for IPv4 defaults to setting the input and
output functions; IPv6's does not. VRF does not need to duplicate the Rx path
so just export the ipv6 input function.
Signed-off-by: NDavid Ahern <dsa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add support for locally originated traffic to VRF-local addresses. If
destination device for an skb is the loopback or VRF device then set
its dst to a local version of the VRF cached dst_entry and call netif_rx
to insert the packet onto the rx queue - similar to what is done for
loopback. This patch handles IPv4 support; follow on patch handles IPv6.

With this patch, ping, tcp and udp packets to a local IPv4 address are
successfully routed:

    $ ip addr show dev eth1
    4: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast master red state UP group default qlen 1000
        link/ether 02:e0:f9:1c:b9:74 brd ff:ff:ff:ff:ff:ff
        inet 10.100.1.1/24 brd 10.100.1.255 scope global eth1
           valid_lft forever preferred_lft forever
        inet6 2100:1::1/120 scope global
           valid_lft forever preferred_lft forever
        inet6 fe80::e0:f9ff:fe1c:b974/64 scope link
           valid_lft forever preferred_lft forever

    $ ping -c1 -I red 10.100.1.1
    ping: Warning: source address might be selected on device other than red.
    PING 10.100.1.1 (10.100.1.1) from 10.100.1.1 red: 56(84) bytes of data.
    64 bytes from 10.100.1.1: icmp_seq=1 ttl=64 time=0.057 ms

This patch also enables use of IPv4 loopback address on the VRF device:
    $ ip addr add dev red 127.0.0.1/8

    $ ping -c1 -I red 127.0.0.1
    PING 127.0.0.1 (127.0.0.1) from 127.0.0.1 red: 56(84) bytes of data.
    64 bytes from 127.0.0.1: icmp_seq=1 ttl=64 time=0.058 ms
Signed-off-by: NDavid Ahern <dsa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Move the stripping of the ethernet header from is_ip_tx_frame into the
ipv4 and ipv6 outbound functions. If the packet is destined to a local
address the header is retained since the packet is sent back to netif_rx.

Collapse vrf_send_v4_prep into vrf_process_v4_outbound.
Signed-off-by: NDavid Ahern <dsa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Vitaly Kuznetsov says:

====================
hv_netvsc: cleanup after untangling the pointer mess

Changes since v1:
- resend when net-next is open [David Miller]
- rebased to current net-next.

After we made traveling through our internal structures explicit it became
obvious that some functions take arguments they don't need just to do
redundant pointer travel and get to what they really need while their
callers already have the required information.

This is just a cleanup series with no functional changes intended. It
doesn't pretend to be complete, additional cleanup of other functions may
follow.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The only caller rndis_filter_device_add() has 'struct net_device' pointer
already.
Signed-off-by: NVitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We unpack 'struct net_device' in netvsc_set_mac_addr() to get to
'struct hv_device' pointer which we use in rndis_filter_set_device_mac()
to get back to 'struct net_device'.
Signed-off-by: NVitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Both rndis_filter_open()/rndis_filter_close() use struct hv_device to
reach to struct netvsc_device only and all callers have it already.
While on it, rename net_device to nvdev in rndis_filter_open() as
net_device is misleading.
Signed-off-by: NVitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Make it easier to get 'struct netvsc_device' from 'struct net_device' and
'struct hv_device' by introducing inline helpers.
Signed-off-by: NVitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

net_device_ctx is assigned in the very beginning of the function and 'net'
pointer doesn't change.
Signed-off-by: NVitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Before commit 6d7b857d ("net: use lib/percpu_counter API for
fragmentation mem accounting"), setting the reassembly high threshold
to 0 prevented fragment reassembly as first fragment would be always
evicted before second could be added to the queue. While inefficient,
some users apparently relied on this method.

Since the commit mentioned above, a percpu counter is used for
reassembly memory accounting and high batch size avoids taking slow path
in most common scenarios. As a result, a whole full sized packet can be
reassembled without the percpu counter's main counter changing its value
so that even with high_thresh set to 0, fragmented packets can be still
reassembled and processed.

Add explicit check preventing reassembly if high threshold is zero.
Signed-off-by: NMichal Kubecek <mkubecek@suse.cz>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Kejian Yan says:

====================
net: hns: add support of ACPI

This series adds HNS support of acpi. The routine will call some ACPI
helper functions, like acpi_dev_found() and acpi_evaluate_dsm(), which
are not included in other cases. In order to make system compile
successfully in other cases except ACPI, it needs to add relative stub
functions to linux/acpi.h. And we use device property functions instead
of serial helper functions to suport both DT and ACPI cases. And then
add the supports of ACPI for HNS.

change log:
 v3->v4:
  mii-id gets from dev-name instead of address

 v2->v3:
 1. add Review-by: Andy Shevchenko
 2. fix the potential memory leak

 v1 -> v2:
 1. use acpi_dev_found() instead of acpi_match_device_ids() to check if
it is a acpi node.
 2. use is_of_node() instead of IS_ENABLED() to check if it is a DT node.
 3. split the patch("add support of acpi for hns-mdio") into two patches:
    3.1 Move to use fwnode_handle
    3.2 Add ACPI
 4. add the patch which subject is dsaf misc operation method
 5. fix the comments by Andy Shevchenko
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Enet needs to get configration parameter by acpi. This patch
adds support of ACPI for enet. The configuration parameter will
be configed in BIOS.
Signed-off-by: NKejian Yan <yankejian@huawei.com>
Signed-off-by: NYisen Zhuang <Yisen.Zhuang@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>