26 Mar, 2017
1 commit
-
This patch refactors the num_packets counter of a forw_packet in the
following three ways:1) Removed dual-use of forw_packet::num_packets:
-> now for aggregation purposes only
2) Using forw_packet::skb::cb::num_bcasts instead:
-> for easier access in aggregation code later
3) make access to num_bcasts private to batadv_forw_packet_*()Signed-off-by: Linus Lüssing
[sven@narfation.org: Change num_bcasts to unsigned]
Signed-off-by: Sven Eckelmann
Signed-off-by: Simon Wunderlich
26 Jan, 2017
1 commit
-
Signed-off-by: Sven Eckelmann
Signed-off-by: Simon Wunderlich
09 Nov, 2016
2 commits
-
The batman-adv module has to be loaded to fulfill genl request by the
userspace. When it is not loaded then requests will fail. It is therefore
useful to get the module automatically loaded when such a request is made.Signed-off-by: Sven Eckelmann
Signed-off-by: Simon Wunderlich -
Receiving functions in Linux consume the supplied skbuff. Doing the same in
the batadv_rx_handler functions makes the behavior more similar to the rest
of the Linux network code.Signed-off-by: Sven Eckelmann
Signed-off-by: Simon Wunderlich
19 Oct, 2016
1 commit
-
The batman-adv codebase is using "list" for the list node (prev/next) and
+"_list" for the head of a list. Not using this
naming scheme can up in confusions when reading the code.Signed-off-by: Sven Eckelmann
Signed-off-by: Simon Wunderlich
09 Aug, 2016
3 commits
-
The files provided by batman-adv via debugfs are currently converted to
netlink. Tools which are not yet converted to use the netlink interface may
still rely on the old debugfs files. But systems which already upgraded
their tools can save some space by disabling this feature. The default
configuration of batman-adv on amd64 can reduce the size of the module by
around 11% when this feature is disabled.$ size net/batman-adv/batman-adv.ko*
text data bss dec hex filename
150507 10395 4160 165062 284c6 net/batman-adv/batman-adv.ko.y
137106 7099 2112 146317 23b8d net/batman-adv/batman-adv.ko.nSigned-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Simon Wunderlich -
The translation table (global, local) is usually the part of batman-adv
which has the most dynamical allocated objects. Most of them
(tt_local_entry, tt_global_entry, tt_orig_list_entry, tt_change_node,
tt_req_node, tt_roam_node) are equally sized. So it makes sense to have
them allocated from a kmem_cache for each type.This approach allowed a small wireless router (TP-Link TL-841NDv8; SLUB
allocator) to store 34% more translation table entries compared to the
current implementation.[1] https://open-mesh.org/projects/batman-adv/wiki/Kmalloc-kmem-cache-tests
Reported-by: Linus Lüssing
Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Simon Wunderlich -
The batman-adv module can automatically be loaded when operations over the
rtnl link are triggered. This requires only the correct rtnl link name in
the module header.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Simon Wunderlich
04 Jul, 2016
2 commits
-
The throughput meter module is a simple, kernel-space replacement for
throughtput measurements tool like iperf and netperf. It is intended to
approximate TCP behaviour.It is invoked through batctl: the protocol is connection oriented, with
cumulative acknowledgment and a dynamic-size sliding window.The test *can* be interrupted by batctl. A receiver side timeout avoids
unlimited waitings for sender packets: after one second of inactivity, the
receiver abort the ongoing test.Based on a prototype from Edo Monticelli
Signed-off-by: Antonio Quartulli
Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Simon Wunderlich -
debugfs is currently severely broken virtually everywhere in the kernel
where files are dynamically added and removed (see
http://lkml.iu.edu/hypermail/linux/kernel/1506.1/02196.html for some
details). In addition to that, debugfs is not namespace-aware.Instead of adding new debugfs entries, the whole infrastructure should be
moved to netlink. This will fix the long standing problem of large buffers
for debug tables and hard to parse text files.Signed-off-by: Matthias Schiffer
Signed-off-by: Andrew Lunn
[sven.eckelmann@open-mesh.com: Strip down patch to only add genl family,
add missing kerneldoc]
Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Simon Wunderlich
30 Jun, 2016
5 commits
-
The bat_algo.h had some functions declared which were not part of the
bat_algo.c file. These are instead stored in bat_v.c and bat_iv_ogm.c. The
declaration should therefore be also in bat_v.h and bat_iv_ogm,h to make
them easier to find.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Simon Wunderlich -
There are several places in batman-adv which provide logging related
functions. These should be grouped together in the log.* files to make them
easier to find.Reported-by: Markus Pargmann
Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Simon Wunderlich -
The bat_algo functionality in main.c is mostly unrelated to the rest of the
content. It still takes up a large portion of this source file (~15%, 103
lines). Moving it to a separate file makes it better visible as a main
component of the batman-adv implementation and hides it less in the other
helper functions in main.c.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Simon Wunderlich -
The tvlv functionality in main.c is mostly unrelated to the rest of the
content. It still takes up a large portion of this source file (~45%, 588
lines). Moving it to a separate file makes it better visible as a main
component of the batman-adv implementation and hides it less in the other
helper functions in main.cSigned-off-by: Markus Pargmann
[sven@narfation.org: fix conflicts with current version, fix includes,
rewrote commit message]
Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Simon Wunderlich -
The ogm_emit and ogm_schedule API calls were rather tight to the
B.A.T.M.A.N. IV logic and therefore rather difficult to use
with other algorithm implementations.Remove such calls and move the surrounding logic into the
B.A.T.M.A.N. IV specific code.Signed-off-by: Antonio Quartulli
Signed-off-by: Marek Lindner
Signed-off-by: Sven Eckelmann
Signed-off-by: Simon Wunderlich
10 May, 2016
2 commits
-
batadv_tvlv_container_get requires that tvlv.container_list_lock is held by
the caller. It is therefore not possible that an item in
tvlv.container_list has an reference counter of 0 and is still in the listThe kref_get function instead WARNs (with debug information) when the
reference counter would still be 0. This makes a bug in batman-adv better
visible because kref_get_unless_zero would have ignored this problem.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli -
The receive function may start processing an incoming packet while the
hard_iface is shut down in a different context. All called functions called
with the batadv_hard_iface object belonging to the incoming interface would
have to check whether the reference counter is still > 0.This is rather error-prone because this check can be forgotten easily.
Instead check the reference counter when receiving the object to make sure
that all called functions have a valid reference.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli
04 May, 2016
1 commit
-
Some really long function names in batman-adv require a newline between
return type and the function name. This has lead to some lines starting
with *batadv_...This * belongs to the return type and thus should be on the same line as
the return type.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli
29 Feb, 2016
2 commits
-
This is the initial implementation of the new OGM protocol
(version 2). It has been designed to work on top of the
newly added ELP.In the previous version the OGM protocol was used to both
measure link qualities and flood the network with the metric
information. In this version the protocol is in charge of
the latter task only, leaving the former to ELP.This means being able to decouple the interval used by the
neighbor discovery from the OGM broadcasting, which revealed
to be costly in dense networks and needed to be relaxed so
leading to a less responsive routing protocol.Signed-off-by: Antonio Quartulli
Signed-off-by: Marek Lindner -
The B.A.T.M.A.N. protocol originally only used a single
message type (called OGM) to determine the link qualities to
the direct neighbors and spreading these link quality
information through the whole mesh. This procedure is
summarized on the BATMAN concept page and explained in
details in the RFC draft published in 2008.This approach was chosen for its simplicity during the
protocol design phase and the implementation. However, it
also bears some drawbacks:* Wireless interfaces usually come with some packet loss,
therefore a higher broadcast rate is desirable to allow
a fast reaction on flaky connections.
Other interfaces of the same host might be connected to
Ethernet LANs / VPNs / etc which rarely exhibit packet
loss would benefit from a lower broadcast rate to reduce
overhead.
* It generally is more desirable to detect local link
quality changes at a faster rate than propagating all
these changes through the entire mesh (the far end of
the mesh does not need to care about local link quality
changes that much). Other optimizations strategies, like
reducing overhead, might be possible if OGMs weren't
used for all tasks in the mesh at the same time.As a result detecting local link qualities shall be handled
by an independent message type, ELP, whereas the OGM message
type remains responsible for flooding the mesh with these
link quality information and determining the overall path
transmit qualities.Developed by Linus during a 6 months trainee study period in
Ascom (Switzerland) AG.Signed-off-by: Linus Luessing
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli
23 Feb, 2016
5 commits
-
The batman-adv source code is the only place in the kernel which uses the
*_free_ref naming scheme for the *_put functions. Changing it to *_put
makes it more consistent and makes it easier to understand the connection
to the *_get functions.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli -
The batman-adv source code is the only place in the kernel which uses the
*_free_ref naming scheme for the *_put functions. Changing it to *_put
makes it more consistent and makes it easier to understand the connection
to the *_get functions.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli -
The batman-adv source code is the only place in the kernel which uses the
*_free_ref naming scheme for the *_put functions. Changing it to *_put
makes it more consistent and makes it easier to understand the connection
to the *_get functions.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli -
The batman-adv source code is the only place in the kernel which uses the
*_free_ref naming scheme for the *_put functions. Changing it to *_put
makes it more consistent and makes it easier to understand the connection
to the *_get functions.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli -
The batman-adv source code is the only place in the kernel which uses the
*_free_ref naming scheme for the *_put functions. Changing it to *_put
makes it more consistent and makes it easier to understand the connection
to the *_get functions.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli
10 Feb, 2016
3 commits
-
batman-adv uses a self-written reference implementation which is just based
on atomic_t. This is less obvious when reading the code than kref and
therefore increases the change that the reference counting will be missed.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli -
batman-adv uses a self-written reference implementation which is just based
on atomic_t. This is less obvious when reading the code than kref and
therefore increases the change that the reference counting will be missed.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli -
The batadv_tvlv_container* functions state in their kernel-doc that they
require tvlv.container_list_lock. Add an assert to automatically detect
when this might have been ignored by the caller.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli
02 Feb, 2016
3 commits
-
Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli -
Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli -
Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli
09 Jan, 2016
3 commits
-
* Let us return directly if a call of the batadv_orig_hash_find() function
returned a null pointer.* Omit the initialisation for the variable "skb" at the beginning.
* Replace an assignment by a call of the kfree_skb() function
and delete the affected variable "ret" then.Signed-off-by: Markus Elfring
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli -
The kfree_skb() function tests whether its argument is NULL and then
returns immediately. Thus the test around the calls is not needed.This issue was detected by using the Coccinelle software.
Signed-off-by: Markus Elfring
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli -
The function handles tlv containers and not tlv handlers. Thus the
lockdep_assert_held has to check for the container_list lock.Fixes: 2c72d655b044 ("batman-adv: Annotate deleting functions with external lock via lockdep")
Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli
16 Dec, 2015
2 commits
-
Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli -
Since the function applies a threshold and also slightly worse
values are accepted, ''equal or better'' does not represent the
intention of the function. ''Similar or better'' represents that better.Signed-off-by: Simon Wunderlich
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli
28 Aug, 2015
1 commit
-
With rcu, the gateway node deleted attribute is not needed anymore. In
fact, it may delay the free of the gateway node and its referenced
structures. Therefore remove it altogether and simplify purging as well.Signed-off-by: Simon Wunderlich
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli
25 Aug, 2015
3 commits
-
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli -
Functions which use (h)list_del* are requiring correct locking when they
operate on global lists. Most of the time the search in the list and the
delete are done in the same function. All other cases should have it
visible that they require a special lock to avoid race conditions.Lockdep asserts can be used to check these problem during runtime when the
lockdep functionality is enabled.Signed-off-by: Sven Eckelmann
Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli -
Since the list's tail is never accessed using a double linked list head
wastes memory.Signed-off-by: Marek Lindner
Signed-off-by: Antonio Quartulli
