This helps tools such as wpa_supplicant can start even if the macsec
module isn't loaded yet.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add support for extended error reporting.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Acked-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add support for extended error reporting.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Acked-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add support for extended error reporting.
Signed-off-by: Matthias Schiffer <mschiffer@universe-factory.net>
Acked-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
It seems like a historic accident that these return unsigned char *,
and in many places that means casts are required, more often than not.
Make these functions return void * and remove all the casts across
the tree, adding a (u8 *) cast only where the unsigned char pointer
was used directly, all done with the following spatch:
@@
expression SKB, LEN;
typedef u8;
identifier fn = { skb_push, __skb_push, skb_push_rcsum };
@@
- *(fn(SKB, LEN))
+ *(u8 *)fn(SKB, LEN)
@@
expression E, SKB, LEN;
identifier fn = { skb_push, __skb_push, skb_push_rcsum };
type T;
@@
- E = ((T *)(fn(SKB, LEN)))
+ E = fn(SKB, LEN)
@@
expression SKB, LEN;
identifier fn = { skb_push, __skb_push, skb_push_rcsum };
@@
- fn(SKB, LEN)[0]
+ *(u8 *)fn(SKB, LEN)
Note that the last part there converts from push(...)[0] to the
more idiomatic *(u8 *)push(...).
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Network devices can allocate reasources and private memory using
netdev_ops->ndo_init(). However, the release of these resources
can occur in one of two different places.
Either netdev_ops->ndo_uninit() or netdev->destructor().
The decision of which operation frees the resources depends upon
whether it is necessary for all netdev refs to be released before it
is safe to perform the freeing.
netdev_ops->ndo_uninit() presumably can occur right after the
NETDEV_UNREGISTER notifier completes and the unicast and multicast
address lists are flushed.
netdev->destructor(), on the other hand, does not run until the
netdev references all go away.
Further complicating the situation is that netdev->destructor()
almost universally does also a free_netdev().
This creates a problem for the logic in register_netdevice().
Because all callers of register_netdevice() manage the freeing
of the netdev, and invoke free_netdev(dev) if register_netdevice()
fails.
If netdev_ops->ndo_init() succeeds, but something else fails inside
of register_netdevice(), it does call ndo_ops->ndo_uninit(). But
it is not able to invoke netdev->destructor().
This is because netdev->destructor() will do a free_netdev() and
then the caller of register_netdevice() will do the same.
However, this means that the resources that would normally be released
by netdev->destructor() will not be.
Over the years drivers have added local hacks to deal with this, by
invoking their destructor parts by hand when register_netdevice()
fails.
Many drivers do not try to deal with this, and instead we have leaks.
Let's close this hole by formalizing the distinction between what
private things need to be freed up by netdev->destructor() and whether
the driver needs unregister_netdevice() to perform the free_netdev().
netdev->priv_destructor() performs all actions to free up the private
resources that used to be freed by netdev->destructor(), except for
free_netdev().
netdev->needs_free_netdev is a boolean that indicates whether
free_netdev() should be done at the end of unregister_netdevice().
Now, register_netdevice() can sanely release all resources after
ndo_ops->ndo_init() succeeds, by invoking both ndo_ops->ndo_uninit()
and netdev->priv_destructor().
And at the end of unregister_netdevice(), we invoke
netdev->priv_destructor() and optionally call free_netdev().
Signed-off-by: David S. Miller <davem@davemloft.net>
The macsec implementation shouldn't account for rx/tx packets that are
dropped in the netdev framework. The netdev framework itself accounts
for such packets by atomically updating struct net_device`rx_dropped and
struct net_device`tx_dropped fields. Later on when the stats for macsec
link is retrieved, the packets dropped in netdev framework will be
included in dev_get_stats() after calling macsec.c`macsec_get_stats64()
Signed-off-by: Girish Moodalbail <girish.moodalbail@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We call skb_cow_data, which is good anyway to ensure we can actually
modify the skb as such (another error from prior). Now that we have the
number of fragments required, we can safely allocate exactly that amount
of memory.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Acked-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
While this may appear as a humdrum one line change, it's actually quite
important. An sk_buff stores data in three places:
1. A linear chunk of allocated memory in skb->data. This is the easiest
one to work with, but it precludes using scatterdata since the memory
must be linear.
2. The array skb_shinfo(skb)->frags, which is of maximum length
MAX_SKB_FRAGS. This is nice for scattergather, since these fragments
can point to different pages.
3. skb_shinfo(skb)->frag_list, which is a pointer to another sk_buff,
which in turn can have data in either (1) or (2).
The first two are rather easy to deal with, since they're of a fixed
maximum length, while the third one is not, since there can be
potentially limitless chains of fragments. Fortunately dealing with
frag_list is opt-in for drivers, so drivers don't actually have to deal
with this mess. For whatever reason, macsec decided it wanted pain, and
so it explicitly specified NETIF_F_FRAGLIST.
Because dealing with (1), (2), and (3) is insane, most users of sk_buff
doing any sort of crypto or paging operation calls a convenient function
called skb_to_sgvec (which happens to be recursive if (3) is in use!).
This takes a sk_buff as input, and writes into its output pointer an
array of scattergather list items. Sometimes people like to declare a
fixed size scattergather list on the stack; othertimes people like to
allocate a fixed size scattergather list on the heap. However, if you're
doing it in a fixed-size fashion, you really shouldn't be using
NETIF_F_FRAGLIST too (unless you're also ensuring the sk_buff and its
frag_list children arent't shared and then you check the number of
fragments in total required.)
Macsec specifically does this:
size += sizeof(struct scatterlist) * (MAX_SKB_FRAGS + 1);
tmp = kmalloc(size, GFP_ATOMIC);
*sg = (struct scatterlist *)(tmp + sg_offset);
...
sg_init_table(sg, MAX_SKB_FRAGS + 1);
skb_to_sgvec(skb, sg, 0, skb->len);
Specifying MAX_SKB_FRAGS + 1 is the right answer usually, but not if you're
using NETIF_F_FRAGLIST, in which case the call to skb_to_sgvec will
overflow the heap, and disaster ensues.
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Cc: stable@vger.kernel.org
Cc: security@kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
Pass the new extended ACK reporting struct to all of the generic
netlink parsing functions. For now, pass NULL in almost all callers
(except for some in the core.)
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
MACSec test failed when asynchronous crypto operations is used. It
encounters packet validation failed since macsec_skb_cb(skb)->valid
is always 'false'.
This patch adds missing "macsec_skb_cb(skb)->valid = true" in
macsec_decrypt_done() when "err == 0".
Signed-off-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The network device operation for reading statistics is only called
in one place, and it ignores the return value. Having a structure
return value is potentially confusing because some future driver could
incorrectly assume that the return value was used.
Fix all drivers with ndo_get_stats64 to have a void function.
Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Remove first zero for add, and use full attribute name in comments.
Signed-off-by: Zhang Shengju <zhangshengju@cmss.chinamobile.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Mostly simple overlapping changes.
For example, David Ahern's adjacency list revamp in 'net-next'
conflicted with an adjacency list traversal bug fix in 'net'.
Signed-off-by: David S. Miller <davem@davemloft.net>
Even if sending SCIs is explicitly disabled, the code that creates the
Security Tag might still decide to add it (e.g. if multiple RX SCs are
defined on the MACsec interface).
But because the header length so far only depended on the configuration
option the SCI overwrote the original frame's contents (EtherType and
e.g. the beginning of the IP header) and if encrypted did not visibly
end up in the packet, while the SC flag in the TCI field of the Security
Tag was still set, resulting in invalid MACsec frames.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: Tobias Brunner <tobias@strongswan.org>
Acked-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now genl_register_family() is the only thing (other than the
users themselves, perhaps, but I didn't find any doing that)
writing to the family struct.
In all families that I found, genl_register_family() is only
called from __init functions (some indirectly, in which case
I've add __init annotations to clarifly things), so all can
actually be marked __ro_after_init.
This protects the data structure from accidental corruption.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Instead of providing macros/inline functions to initialize
the families, make all users initialize them statically and
get rid of the macros.
This reduces the kernel code size by about 1.6k on x86-64
(with allyesconfig).
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Static family IDs have never really been used, the only
use case was the workaround I introduced for those users
that assumed their family ID was also their multicast
group ID.
Additionally, because static family IDs would never be
reserved by the generic netlink code, using a relatively
low ID would only work for built-in families that can be
registered immediately after generic netlink is started,
which is basically only the control family (apart from
the workaround code, which I also had to add code for so
it would reserve those IDs)
Thus, anything other than GENL_ID_GENERATE is flawed and
luckily not used except in the cases I mentioned. Move
those workarounds into a few lines of code, and then get
rid of GENL_ID_GENERATE entirely, making it more robust.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
geneve:
- Merge __geneve_change_mtu back into geneve_change_mtu, set max_mtu
- This one isn't quite as straight-forward as others, could use some
closer inspection and testing
macvlan:
- set min/max_mtu
tun:
- set min/max_mtu, remove tun_net_change_mtu
vxlan:
- Merge __vxlan_change_mtu back into vxlan_change_mtu
- Set max_mtu to IP_MAX_MTU and retain dynamic MTU range checks in
change_mtu function
- This one is also not as straight-forward and could use closer inspection
and testing from vxlan folks
bridge:
- set max_mtu of IP_MAX_MTU and retain dynamic MTU range checks in
change_mtu function
openvswitch:
- set min/max_mtu, remove internal_dev_change_mtu
- note: max_mtu wasn't checked previously, it's been set to 65535, which
is the largest possible size supported
sch_teql:
- set min/max_mtu (note: max_mtu previously unchecked, used max of 65535)
macsec:
- min_mtu = 0, max_mtu = 65535
macvlan:
- min_mtu = 0, max_mtu = 65535
ntb_netdev:
- min_mtu = 0, max_mtu = 65535
veth:
- min_mtu = 68, max_mtu = 65535
8021q:
- min_mtu = 0, max_mtu = 65535
CC: netdev@vger.kernel.org
CC: Nicolas Dichtel <nicolas.dichtel@6wind.com>
CC: Hannes Frederic Sowa <hannes@stressinduktion.org>
CC: Tom Herbert <tom@herbertland.com>
CC: Daniel Borkmann <daniel@iogearbox.net>
CC: Alexander Duyck <alexander.h.duyck@intel.com>
CC: Paolo Abeni <pabeni@redhat.com>
CC: Jiri Benc <jbenc@redhat.com>
CC: WANG Cong <xiyou.wangcong@gmail.com>
CC: Roopa Prabhu <roopa@cumulusnetworks.com>
CC: Pravin B Shelar <pshelar@ovn.org>
CC: Sabrina Dubroca <sd@queasysnail.net>
CC: Patrick McHardy <kaber@trash.net>
CC: Stephen Hemminger <stephen@networkplumber.org>
CC: Pravin Shelar <pshelar@nicira.com>
CC: Maxim Krasnyansky <maxk@qti.qualcomm.com>
Signed-off-by: Jarod Wilson <jarod@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The netdevice type structure for macsec was being defined but never used.
To set the network device type the macro SET_NETDEV_DEVTYPE must be called.
Compile tested only, I don't use macsec.
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Acked-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
The idea for type_check in dev_get_nest_level() was to count the number
of nested devices of the same type (currently, only macvlan or vlan
devices).
This prevented the false positive lockdep warning on configurations such
as:
eth0 <--- macvlan0 <--- vlan0 <--- macvlan1
However, this doesn't prevent a warning on a configuration such as:
eth0 <--- macvlan0 <--- vlan0
eth1 <--- vlan1 <--- macvlan1
In this case, all the locks end up with a nesting subclass of 1, so
lockdep thinks that there is still a deadlock:
- in the first case we have (macvlan_netdev_addr_lock_key, 1) and then
take (vlan_netdev_xmit_lock_key, 1)
- in the second case, we have (vlan_netdev_xmit_lock_key, 1) and then
take (macvlan_netdev_addr_lock_key, 1)
By removing the linktype check in dev_get_nest_level() and always
incrementing the nesting depth, lockdep considers this configuration
valid.
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, trying to setup a vlan over a macsec device, or other
combinations of devices, triggers a lockdep warning.
Use netdev_lockdep_set_classes and ndo_get_lock_subclass, similar to
what macvlan does.
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
macsec_notify() loops over the list of macsec devices configured on the
underlying device when this device is being removed. This list is part
of the rx_handler data.
However, macsec_dellink unregisters the rx_handler and frees the
rx_handler data when the last macsec device is removed from the
underlying device.
Add macsec_common_dellink() to delete macsec devices without
unregistering the rx_handler and freeing the associated data.
Fixes: 960d5848db ("macsec: fix memory leaks around rx_handler (un)registration")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
When creation of a macsec device fails because an identical device
already exists on this link, the current code decrements the refcnt on
the parent link (in ->destructor for the macsec device), but it had not
been incremented yet.
Move the dev_hold(parent_link) call earlier during macsec device
creation.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Following the previous patch, RXSCs are held and properly refcounted in
the RX path (instead of being implicitly held by their SA), so the SA
doesn't need to hold a reference on its parent RXSC.
This also avoids panics on module unload caused by the double layer of
RCU callbacks (call_rcu frees the RXSA, which puts the final reference
on the RXSC and allows to free it in its own call_rcu) that commit
b196c22af5 ("macsec: add rcu_barrier() on module exit") didn't
protect against.
There were also some refcounting bugs in macsec_add_rxsa where I didn't
put the reference on the RXSC on the error paths, which would lead to
memory leaks.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, we lookup the RXSC without taking a reference on it. The
RXSA holds a reference on the RXSC, but the SA and SC could still both
disappear before we take a reference on the SA.
Take a reference on the RXSC in macsec_handle_frame.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
macsec_decrypt() is not called when validation is disabled and so
macsec_skb_cb(skb)->rx_sa is not set; but it is used later in
macsec_post_decrypt(), ensure that it's always initialized.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: Beniamino Galvani <bgalvani@redhat.com>
Acked-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Test the cipher suite initialization in case ICV length has a value
different than its default. If this test fails, creation of a new macsec
link will also fail. This avoids situations where further security
associations can't be added due to failures of crypto_aead_setauthsize(),
caused by unsupported user-provided values of the ICV length.
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
preserve the return value of AEAD functions that are called when a SA is
created, to avoid inappropriate display of "RTNETLINK answers: Cannot
allocate memory" message.
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
IEEE 802.1AE-2006 standard recommends that the ICV element in a MACsec
frame should not exceed 16 octets: add MACSEC_STD_ICV_LEN in uapi
definitions accordingly, and avoid accepting configurations where the ICV
length exceeds the standard value. Leave definition of MACSEC_MAX_ICV_LEN
unchanged for backwards compatibility with userspace programs.
Fixes: dece8d2b78 ("uapi: add MACsec bits")
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use gro_gells to trigger GRO and allow RPS on macsec traffic
after decryption.
Also, be sure to avoid clearing software offload features in
macsec_fix_features().
Overall this increase TCP tput by 30% on recent h/w.
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Avoid recursions of dev_queue_xmit() to the wrong net device when
frames are unprotected, since at that time skb->dev still points to
our own macsec dev and unlike macsec_encrypt_finish() dev pointer
doesn't get updated to real underlying device.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Sabrina Dubroca <sd@queasysnail.net>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
The ASYNC flag prevents initialization on some physical machines.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
For the crypto callbacks to work properly, we cannot have sg and iv on
the stack. Use kmalloc instead, with a single allocation for
aead_request + scatterlist + iv.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Without this, the various uses of call_rcu could cause a kernel panic.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
In my last commit I replaced MACSEC_SA_ATTR_KEYID by
MACSEC_SA_ATTR_KEY.
Fixes: 8acca6aceb ("macsec: key identifier is 128 bits, not 64")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
In netdevice.h we removed the structure in net-next that is being
changes in 'net'. In macsec.c and rtnetlink.c we have overlaps
between fixes in 'net' and the u64 attribute changes in 'net-next'.
The mlx5 conflicts have to do with vxlan support dependencies.
Signed-off-by: David S. Miller <davem@davemloft.net>
The MACsec standard mentions a key identifier for each key, but
doesn't specify anything about it, so I arbitrarily chose 64 bits.
IEEE 802.1X-2010 specifies MKA (MACsec Key Agreement), and defines the
key identifier to be 128 bits (96 bits "member identifier" + 32 bits
"key number").
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Minor overlapping changes in the conflicts.
In the macsec case, the change of the default ID macro
name overlapped with the 64-bit netlink attribute alignment
fixes in net-next.
Signed-off-by: David S. Miller <davem@davemloft.net>
macsec_validate_attr should check IFLA_MACSEC_REPLAY_PROTECT (not
IFLA_MACSEC_PROTECT) to verify that the replay protection and replay
window arguments are correct.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
I accidentally forgot some MACSEC_ prefixes in if_macsec.h.
Fixes: dece8d2b78 ("uapi: add MACsec bits")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
We leak a struct macsec_rxh_data when we unregister the rx_handler in
macsec_dellink.
We also leak a struct macsec_rxh_data in register_macsec_dev if we fail
to register the rx_handler.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use genl_dump_check_consistent in dump_secy.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Suggested-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
The decrypt callback macsec_decrypt_done needs a reference on the rx_sa
and releases it before returning, but macsec_handle_frame already
put that reference after macsec_decrypt returned NULL.
Set rx_sa to NULL when the decrypt callback runs so that
macsec_handle_frame knows it must not release the reference.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
