Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>

It might be possible that 2 threads access the same data in the same
rcu grace period. The first thread calls call_rcu() to decrement the
refcount and free the data while the second thread increases the
refcount to use the data. To avoid this race condition all refcount
operations have to be atomic.
Reported-by: NSven Eckelmann <sven@narfation.org>
Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>

Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>

When unicast_send_skb() is increasing the orig_node's refcount another
thread might have been freeing this orig_node already. We need to
increase the refcount in the rcu read lock protected area to avoid that.
Signed-off-by: NLinus Lüssing <linus.luessing@ascom.ch>
Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>

It might be possible that 2 threads access the same data in the same
rcu grace period. The first thread calls call_rcu() to decrement the
refcount and free the data while the second thread increases the
refcount to use the data. To avoid this race condition all refcount
operations have to be atomic.
Reported-by: NSven Eckelmann <sven@narfation.org>
Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>

bonding / alternating candidates need to be secured by rcu locks
as well. This patch therefore converts the bonding list
from a plain pointer list to a rcu securable lists and references
the bonding candidates.
Signed-off-by: NSimon Wunderlich <siwu@hrz.tu-chemnitz.de>
Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>

Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>

For a host in the mesh network, the batman layer should be transparent.
However, we had one exception, data packets within the mesh network
which have the same destination as a originator are being routed to
that node, although there is no host that node's bat0 interface and
therefore gets dropped anyway. This commit removes this exception.
Signed-off-by: NLinus Lüssing <linus.luessing@ascom.ch>
Signed-off-by: NSven Eckelmann <sven@narfation.org>

Multiple variable declarations in a single statements over multiple lines can
be split into multiple variable declarations without changing the actual
behavior.
Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>
Signed-off-by: NSven Eckelmann <sven@narfation.org>

The two fragments of an unicast packet must have successive sequence numbers to
allow the receiver side to detect matching fragments and merge them again. The
current implementation doesn't provide that property because a sequence of two
atomic_inc_return may be interleaved with another sequence which also changes
the variable.

The access to the fragment sequence number pool has either to be protected by
correct locking or it has to reserve two sequence numbers in a single fetch.
The latter one can easily be done by increasing the value of the last used
sequence number by 2 in a single step. The generated window of two currently
unused sequence numbers can now be scattered across the two fragments.
Reported-by: NLinus Lüssing <linus.luessing@web.de>
Signed-off-by: NSven Eckelmann <sven@narfation.org>

Signed-off-by: NSven Eckelmann <sven@narfation.org>

The routing algorithm must be able to decide if a fragment can be merged with
the missing part and still be passed to a forwarding interface. The fragments
can only differ by one byte in case that the original payload had an uneven
length. In that situation the sender has to inform all possible receivers that
the tail is one byte longer using the flag UNI_FRAG_LARGETAIL.

The combination of UNI_FRAG_LARGETAIL and UNI_FRAG_HEAD flag makes it possible
to calculate the correct length for even and uneven sized payloads.

The original formula missed to add the unicast header at all and forgot to
remove the fragment header of the second fragment. This made the results highly
unreliable and only useful for machines with large differences between the
configured MTUs.
Reported-by: NRussell Senior <russell@personaltelco.net>
Reported-by: NMarek Lindner <lindner_marek@yahoo.de>
Signed-off-by: NSven Eckelmann <sven@narfation.org>

The routing algorithm must know how large two fragments are to be able to
decide that it is safe to merge them or if it should resubmit without waiting
for the second part. When these two fragments have a too different size, it is
not possible to guess right in every situation.

The user could easily configure the MTU of the attached cards so that one
fragment is forwarded and the other one is added to the fragments table to wait
for the missing part.

For even sized packets, it is possible to split it so that the resulting
packages are equal sized by ignoring the old non-fragment header at the
beginning of the original packet.

This still creates different sized fragments for uneven sized packets.
Reported-by: NRussell Senior <russell@personaltelco.net>
Reported-by: NMarek Lindner <lindner_marek@yahoo.de>
Signed-off-by: NSven Eckelmann <sven@narfation.org>

There's a problem in net/batman-adv/unicast.c::frag_send_skb().
dev_alloc_skb() allocates memory and may fail, thus returning NULL. If
this happens we'll pass a NULL pointer on to skb_split() which in turn
hands it to skb_split_inside_header() from where it gets passed to
skb_put() that lets skb_tail_pointer() play with it and that function
dereferences it. And thus the bat dies.

While I was at it I also moved the call to dev_alloc_skb() above the
assignment to 'unicast_packet' since there's no reason to do that
assignment if the memory allocation fails.
Signed-off-by: NJesper Juhl <jj@chaosbits.net>
Signed-off-by: NSven Eckelmann <sven@narfation.org>

B.A.T.M.A.N. (better approach to mobile ad-hoc networking) is a routing
protocol for multi-hop ad-hoc mesh networks. The networks may be wired or
wireless. See http://www.open-mesh.org/ for more information and user space
tools.
Signed-off-by: NSven Eckelmann <sven@narfation.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

linux/etherdevice.h already provides functions to classify different
ethernet addresses. These inlineable functions should be used instead of
custom functions.

The check for multicast together with multicast can also be replaced
with a single test for multicast because for every ethernet address x
following is always true:

is_broadcast_ether_addr(x) => is_multicast_ether_addr(x)

or when looking more at the implementation:

(FF:FF:FF:FF:FF:FF == x) => [(01:00:00:00:00:00 & x) != 00:00:00:00:00:00]
Reported-by: NTobias Klauser <tklauser@distanz.ch>
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

If the gateway client mode is active batman-adv will send the
broadcasted DHCP requests via unicast to the currently selected best
gateway. Therefore attached clients can profit from batman's knowledge
about the network topology.
Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>
[sven.eckelmann@gmx.de: Rework on top of current version]
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

spin_lock_irqsave disables the IRQs and stores them inside the flags
provided by the caller. This is needed to protect a bottom half handler
or a user context critical section from being interrupted by an
interrupt handler which also tries to acquire the spinlock and locks
forever.

The linux device drivers will receive the packets inside an interrupt
handler and the network infrastructure will process them inside bottom
half. Thus batman-adv will only run in user context and bottom half
handlers. We can conclude that batman-adv doesn't share its own
spinlocks with real interrupt handlers.

This makes it possible to exchange the quite complex spin_lock_irqsave
with spin_lock_bh which only stops bottom halves from running on the
current cpu, but allows interrupt handlers to take over to keep the
interrupt latency low.
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Function pointers cannot be inlined by a compiler and thus always has
the overhead of an call. hashdata_choose_cb's are one of the most often
called function pointers and its overhead must kept relative low.

As first step, every function which uses this function pointer takes it
as parameter instead of storing it inside the hash abstraction
structure.

This not generate any performance gain right now. The called functions
must also be able to be inlined by the calling functions to enable
inlining of the function pointer.
Reported-by: NDavid S. Miller <davem@davemloft.net>
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Function pointers cannot be inlined by a compiler and thus always has
the overhead of an call. hashdata_compare_cb's are one of the most often
called function pointers and its overhead must kept relative low.

As first step, every function which uses this function pointer takes it
as parameter instead of storing it inside the hash abstraction
structure.

This not generate any performance gain right now. The called functions
must also be able to be inlined by the calling functions to enable
inlining of the function pointer.
Reported-by: NDavid S. Miller <davem@davemloft.net>
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Both sysfs entries and variable names shall be as descriptive as
possible while not exceeding a certain length. This patch renames
bat_priv atomics to be equally descriptive with their according sysfs
entries.

Unifying sysfs and bat_priv atomic names also makes it easier to find
each others pendant.

The reduced ("type"-)information which was previously indicated with a
_enabled for booleans got substituted by a comment in bat_priv.

This patch has also been done in regards for the future BAT_ATTR_*
macros (they only need one name argument instead of a file and variable
name).
Signed-off-by: NLinus Lüssing <linus.luessing@web.de>
Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Signed-off-by: NAndreas Langer <an.langer@gmx.de>
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

If a packet is too big to be forwarded over an interface it will be
fragmented on-the-fly (if fragmentation is enabled).
Signed-off-by: NAndreas Langer <an.langer@gmx.de>
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Signed-off-by: NAndreas Langer <an.langer@gmx.de>
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Signed-off-by: NAndreas Langer <an.langer@gmx.de>
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

The unicast_frag_send_skb() function expected 'raw' packets (without any
batman-adv header) to fragment them. This needs to be changed, so that
this function is able to fragment packets that already traveled inside
the mesh but need to be fragmented now.
Signed-off-by: NAndreas Langer <an.langer@gmx.de>
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

55d1666b521cbed95924c8d4775fe272c103f08c incidentally disabled bonding
of packets first entering the mesh along with also disabling interface
alternating regardless of where the packet came from. This re-enables
these options.
Signed-off-by: NChris Lang <clang@gateworks.com>
Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

kmalloc() may fail, if so drop current packet.
Signed-off-by: NVasiliy Kulikov <segooon@gmail.com>
Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>
[sven.eckelmann@gmx.de: Removed new introduced deadlock]
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This patch removes all remaining global variables and includes the
necessary bits into the bat_priv structure. It is the last
remaining piece to allow multiple concurrent mesh clouds on the
same device.
A few global variables have been rendered obsolete during the process
and have been removed entirely.
Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>
[sven.eckelmann@gmx.de: Rework on top of current version]
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

my_skb_push provided an easy way to allocate enough headroom in
situation were we don't have enough space left and move the data pointer
to the new position, but we didn't checked wether we are allowed to
write to the new pushed header. This is for example a problem when the
skb was cloned and thus doesn't have a private data part.

my_skb_head_push now replaces my_skb_push by using skb_cow_head to
provide only a large enough, writable header without testing for the
rest of the (maybe shared) data. It will also move the data pointer
using skb_push when skb_cow_head doesn't fail.

This should give us enough flexibility in situation were skbs will be
queued by underlying layers and still doesn't unnecessarily copy the
data in situations when the skb was consumed right away during
dev_queue_xmit.
Reported-by: NMarek Lindner <lindner_marek@yahoo.de>
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

This patch implements a simple layer2 fragmentation to allow traffic
exchange over network interfaces with a MTU smaller than 1500 bytes. The
fragmentation splits the big packets into two parts and marks the frames
accordingly. The receiving end buffers the packets to reassemble the
orignal packet before passing it to the higher layers. This feature
makes it necessary to modify the batman-adv encapsulation for unicast
packets by adding a sequence number, flags and the originator address.
This modifcation is part of a seperate packet type for fragemented
packets to keep the original overhead as low as possible. This patch
enables the feature by default to ensure the data traffic can travel
through the network. But it also prints a warning to notify the user
about the performance implications.

Note: Fragmentation should be avoided at all costs since it has a
dramatic impact on the performance, especially when it comes wifi
networks. Instead of a single packet, 2 packets have to be sent! Not
only valuable airtime is wasted but also packetloss decreases the
throughput. A link with 50% packetloss and fragmentation enabled is
pretty much unusable.
Signed-off-by: NAndreas Langer <an.langer@gmx.de>
[sven.eckelmann@gmx.de: Rework on top of current version]
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Signed-off-by: NMarek Lindner <lindner_marek@yahoo.de>
[sven.eckelmann@gmx.de: Rework on top of current version]
Signed-off-by: NSven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>