[olsrd.git] / CHANGELOG

This file states changes as of version 0.2.4:
$Id: CHANGELOG,v 1.108 2007/11/03 22:56:07 bernd67 Exp $

0.5.5 ---------------------------------------------------------------------

BUGFIX by Hannes Gredler <hannes@gredler.at> fixing not deleted tc entry.

PATCHES and CLEANUPS by Bernd Petrovitsch <bernd@firmix.at>
- Jens Nachtigall <nachtigall@web.de> suggested (and reminded) to use
  olsrd.conf.default.lq as the default installed one.
  It makes much more sense as - to the best of my(bernd's) knowledge - all
  larger networks uses the LQ extensions. And new customers should better
  be pushed in that direction.
- with the help of Patrick McCarty <mccartyp@gmail.com>, we got olsr to build
  on OpenBSD again:
  * OpenBSD'd `sed` is pretty conservative and doesn't understand 's|re|sub|'.
    So we fall back to the old 's/re/sub/' and quote all '/' and '.' correctly
    (which was the main reason to use '|').
    And while I'm at it: Remove the useless $(CCACHE) from the $(CC) for
    dependency generation and beautify (read: "$(strip)") $(CPPFLAGS).
  * OpenBSD needs an explicit "#include <sys/types.h>" before
    "#include <netinet/in.h>".
- fixed lib/secure/src/md5.h: This was broken as it had
  "typedef unsigned long int UINT4;". "unsigned long int" is 8 bytes on x86_64.
  We are using now the standardized types from <inttypes.h>.
- fixed warnings and improved lib/secure/src/md5.c:
  * we are using memcpy() and memset() instead of the open-coded loops as
    suggested 
  * const'ified
  * added function prototypes for static functions
  * moved static functions to the top so that their declaration is before
    their use to allow gcc to inline if only used once.
  It remains ugly - God knows why there are that so many useless
  type-casts.
- Some minor and trivial cleanup: Added two functions to calculate the ETX
  value from a struct tc_edge_entry and a struct link_entry, respectively.
  And use them everywhere.
- added http://meshcube.org/nylon/utils/olsr-topology-view.pl into
  lib/httpinfo/misc as suggested by Jens Nachtigall <nachtigall@web.de>

0.5.4 ---------------------------------------------------------------------

QUAGGA by Immo 'FaUl' Wehrenberg <immo.olsr@do.bundessicherheitsministerium.de>
- updated to svn version 33

BMF PLUGIN  by Erik Tromp <erik_tromp@hotmail.com>
- updated to 1.5.1
- updated to latest plugin interfaces changes and killed warnings (by Bernd
  Petrovitsch <bernd@firmix.at>)

PATCH by Hannes Gredler <hannes@gredler.at> which rewrites the route handling.
To quote him:
----  snip  ----
change list:
- get rid of separate routing tables for HNA and per-node routes, everything is
  now unified in an AVL routing tree (&routingtree)

- introduce walking macros (OLSR_FOR_ALL_RT_ENTRIES()) that hide the internal
  structure of the RIB for making life of the plugin authors easier.

- get rid of different SPF implementations for LQ and non-LQ code paths. a
  non-LQ edge is simply substituted with a cost of 1.0

- get rid of host masks - a new data type olsr_prefix is introduced which is
  basically an ip address plus a prefix length.

  do not install the metric in the kernel FIB - for the kernel its pointless
  if the route gets installed with a metric of N or M.

  we do not need to update the kernel FIB if we have hop count only changes
  (for example if there is a reroute action further downstream)

  the only things which triggers a kernel FIB route update is a next hop
  change (a next hop is neighboring gateway router plus an interface).

  all OLSR routes are installed with a metric of 2

- separate between rt_entry and rt_path - the former is a route installed in the
  kernel with an next hop. the latter is a candidate for best path selection
  after SPF calculation has been done. in the rt_entry we keep a pointer to the
  best_path and also to the next hop that was installed in the kernel FIB.

  we always keep all originator of a route, if a route originator goes away we
  can easy recompute the best path for the route.

  the next hop in the rt_entry gets only updated upon a successful route_add
  call - that way we always remember what next hop to delete.

  stray routes should be history now.

- tweak the linked list toolkit to operate on circular lists.

- get rid of malloc calls for building the kernel update list. the list node is
  now embedded in the rt_entry.

- introduce three queues (add/chg/del) for kernel updates.

- for neighbor route dependency tracking the neighbor routes are queued first or
  last (depending on which queue you work on)

- rework all the plugins which directly manipulate rt entries.

- rework the plugins that read from the routing table (most notably nameserver,
  httpinfo and quagga plugin)

- lots of comments that explains the intentions and purpose of this code-piece.

non RT related stuff:
- use a list rather than a tree for storing the post-SPF results, which further
  improves the raw-SPF runtime.

- add display of SPF runtime (masked behind #ifdef SPF_PROFILING)

- http://gredler.at/download/olsrd/neighbor_routes3.diff: This updates the own
  IP address (read: the main address) after changes (e.g. on
  `ifup wlan0; sleep 1; ifdown wlan0`) and kills the
  olsr_fill_routing_table_with_neighbors() function.
----  snip  ----
And Sven-Ola Tuecke <mail2news@commando.de> fixed an instability issue on interface
up/down operations (see 102-olsrd-rt-refactoring-fixes.patch below) and a missing
initialization.

PATCH by Hannes Gredler <hannes@gredler.at> which "consolidates
the link-state database and the spf-calculation in order
to calculate routes more efficiently".
To quote him (more):
----  snip  ----
- use the link-state (tc) database for SPF calculations rather than
  replicating the notion of vertices and edges for a SPF run.
  this heavily reduces malloc() calls and shrinks the total CPU
  load of the route calculation path between 60%-80%.
----  snip  ----

PATCHES by Sven-Ola Tuecke <mail2news@commando.de> to be found on from
http://download-master.berlin.freifunk.net/sven-ola/nylon/packages/olsrd/files/
- 102-olsrd-rt-refactoring-fixes.patch
  Because you changed a lot of basics: It's time to handle a general
  flaw in the routing system. Plase take a look at chk_if_changed(). This
  will free() any "struct interface" pointer without warning at any time.
  This is why it's possile to SEGV olsrd with a simple "ifdown xxx".
  The patch replaces the (maybe) invalid pointer with an index reference
  "iif_index". You can always ask the OS for a name. Please note, that I do
  not have a working BSD toolchain, so I've placed an #error in the IPv6
  BSD-part where the author/porter has started to hack something funny.

- 110-olsrd-double-wlancard-neigh-hack.patch:
  This is a hack for Nodes having to wifi cards with the same channel,
  bssid, IP-Range etc. If two nodes can see each other by means of two
  possible links (here: two wifi cards with equal config), a bug is  triggered
  with the Neigh-is-SYM detections. This small little hack prevents this.

- 112-olsrd-nameservice-fixemptyname.patch:
  This is an addon to my lat/lon stuff which will prevent olsrd from
  running (oops?) if no hostname is given and the nameservice plugin
  is loaded.

- 113-olsrd-dyngwplain-pluginvers5.patch:
  This updates the dyngwplain plugin to the new Plugin Iface

- 140-olsrd-arprefreshed.patch:
  This is a new one. Opens a packet socket and listen to UDP(698), extract
  the sender MAC and refreshes the ARP cache whith that. Should speedup
  especially in cases, if you initially try to use a longer routing path which
  normally triggers a "ARP-Lookup-Chain".
- 106-olsrd-nameserviceparams.patch:
  This patch converts more plugins to the new interface version.
- 104-olsrd-policy-routing.patch
  Reworked this one to discard GPL helper functions. Also checked IPv6 and
  re-included the IPC hookup. The patch adds a "RtTable [number]" for
  /etc/olsrd.conf which is simply the Linux
  policy routing table to use. Defaults to 254 (== main).
  This patch was modified/clenaed up by <bernd@firmix.at> to use "#if"
  instead of "#ifdef" as it's more robust against typos.
- 110-olsrd-fixpacketprint.patch, 112-olsrd-nameservice-fixemptyname.patch,
  113-olsrd-txtinfo-fixhttpget.patch, 114-olsrd-timeoutlimit.patch,
  115-olsrd-nameserviceparamfix.patch and
  116-olsrd-fix-pluginparam-addons.patch fixing the compilation warning
  on 64bit and lots of other improvements.
- "Save the fish" patch: Avoid forwarding of packets with too low TTL. This
  kills lots of packet forwarding storms.
  NB: The oneliner was applied by hand by BP and formatted to look (in BPs O)
  more readable.

PATCH by Arnd Hannemann <hannemann@i4.informatik.rwth-aachen.de>
olsr_makefile_make_use_of_exename.patch
- This patch makes sure that the EXENAME variable of Makefile.inc is used
  in Makefile.

PATCHES by John Hay <jhay@meraka.org.za>
- update to new FreeBSD WLAN API
- do not require /bin/bash, use /bin/sh
- Fixed alignment so that olsrd runs on FreeBSD/arm
- allow more interface in an IPv6 subnet on FreeBSD
- use PREFIX and DESTDIR as all the other Makefile.$OS also for FreeBSD
- make txtinfo plugin work with IPv6

PATCH by Andreas Jacobs <jacobs@i4.informatik.rwth-aachen.de>
- fix the loss link quality calculation for "windows size % 4 != 0"

PATCH by Acinonyx <acinonyxs@yahoo.gr>
- Bug fix: include $(TOPDIR)/Makefile.inc at the begin in the Quagga plugin

PATCH by David Cornejo <dcornejo@gmail.com>
- fixed an "+=" of an uninitialized variable (detected with/by the
  scan.coverty.com).

BUG reported by Aaron Kaplan <aaron@lo-res.org>
- BSD-xargs doesn't know "-r".

PATCHES and CLEANUPS by Bernd Petrovitsch <bernd@firmix.at>
- Made a function from the ME_TO_DOUBLE() macro (in src/mantissa.h).
  This saves code throughout the code even on i386 and will even more
  on architectures without floating point units and "-msoft-float".
- And the mathemathics in src/mantissa.h is reformulated to minimize
  floating point operations to save CPU power - especially on embedded
  devices.
- I rewrote the half of src/lq_packet.[ch] which deals with incoming
  packets. This was triggered with performance output of gcc produced
  by Sven-Ola Tuecke at CCCamp07.
  This kills *lots* of (more than) superflous malloc()s and the same
  number of (free()s). And it also kills some code and copying around of
  data.
- Make it compile without warning with flex-2.5.33 (to be found on Fedora 7
  and Gentoo in Sep-2007) again.

- converted the dyn_gw plugin to plugin interface version 5 (which leaves
  the quagga plugion as the last with the old one).
- paving the way to activate -Wshadow, much more to do
- const-ify parameters here and there
- use NULL for pointers (and not "0")
- Killed "extern" declarations in (not generated) .c files

- Based on a patch by Gianni Costanzi <gianni.costanzi@gmail.com> (so credits
  and thanks have to go there):
  added OS_CFLAG_PTHREAD Makefile variable since gcc (on Linux) requests this
  in the manual page.
  Changes/additions:
  - I added definitions to all OS-specific Makefile.$OS with the value similar
    to the value in OS_LIB_PTHREAD (either empty or "-pthread").
  - The variable is added to CPPFLAGS (and not CFLAGS) since CPPFLAGS is used
    for all cpp and gcc calls (and gcc's man page indicates that it sets
    variables for both of them).

0.5.3 ---------------------------------------------------------------------

URGENT BUG FIX
Ignacio García Pérez <iggarpe@terra.es> found a serious bug in the dot-draw
plugin and identified it's source.

BUG FIX
The bmf plugin wanted to kill a not-started thread and this causes a segmentation
violation. This happens if the old plugin interface support is not compiled in
and one uses - or more asccurate: wants to use - the bmf plugin which uses that
ATM. Thanks to Cedric Krier <ced@ced.homedns.org> for reporting this.

DEPENDENCIES FIXED
The dependencies (*.d files) are now properly generated and used. And they also
dependen on the Makefile in the current directory and the Makefile.inc so that
changes there also lead to new compilation.
Remember: We support "ccache" so install it to save lot of meaningless recompiles
if nothing (as seen by the real C compiler, i.e. after the C preprocessor) has
changed.

BUILD FIX
Linux/IA64 and Linux/HPPA also need "-fPIC".

CLEANUPS
- olsr_switch does no longer generate warnings
- olsr_switch is also built with "build_all" and cleaned with "uberclean"
- More "const" all over the place.
- More "static" for local functions and variables.
- Killed an unnecessary "static" variable.

0.5.2 ---------------------------------------------------------------------

URGENT BUG FIX
Sven-Ola Tuecke found a serious bug of the class "show stopper" and sent a patch.

PLUGIN CLEANUP
Plugins handle parameters quite individually (case-sensitive vs case-insensitive,
different parsing functions for the same values, etc.). Plugin Interface version 5
now exports a table of { parameter-name, parsing-function, addr-of-storage } and
it's size. Common functions should be shared.

0.5.1 ---------------------------------------------------------------------

MISC
Upgrade to olsr-bmf 1.5 from Erik Tromp <erik_tromp@hotmail.com>

latitude/longitude support is now in the nameservice plugin done by
Sven-Ola Tuecke <mail2news@commando.de>

added the spf refactoring patch from  Hannes Gredler <hannes@gredler.at> which
saves a noteworthy amount of CPU time. To quote him:
----  snip  ----
1. use of an AVL tree as a min-heap implementation

   as a means for efficient sorting.
   (the etx metric is used as the key in the candidate tree)

2. next-hop propagation

   rather than tracking the previous node in olsr_relax()
   i have changed that model and pre-populate all one-hop neighbors
   with their own IP adress as 'next-hop' and pull that
   pointer up once new paths are explored.

   as a result no walker for counting hops and extracting next-hops
   is required - it turns out at this is slighly more efficient
   than the existing behaviour (even with the cache applied).
----  snip  ----

CLEANUPS
* moved a only locally needed hack from "union olsr_ip_addr" into the only place
  where it is needed in hashing.c

0.5.0 ---------------------------------------------------------------------

APPLIED PATCHES FROM FREIFUNK-FIRMWARE
libsmake, dyngwplain, nameservices+services, tzxtinfo, quagga, quagga-routehandler,
optimize, fixes from Eric Tromp and Sven-Ola Tuecke, CPU-optimize, secure key, HNA-IP

CLEANUPS
removed hardcoded limit on the number of interfaces (at least in the Unix part)
made Makefile targets much more consistent

MISC
put Bmf-1.3 plugin into it

FreeBSD IPv6 SUPPORT

Patch from John Hay applied.

BUGFIXES

0.4.10 --------------------------------------------------------------------

OLSR NETWORK SIMULATOR

An application called olsr_switch(sorry for the nameclash with the win32
GUI) has been added. If olsrd is ran using the -hint option it can now
connect to olsr_switch and be part of a virtual network. This means
that multiple instances can be ran at the same host! The olsr_switch
application allows the user to define link variables on all viritual
links in the network, creating all possible topologies. Read more in
the readme file.

EXPERIMENTAL FISH EYE ALGORITHM

An experimental algorithm for TTL settings on TC messages has
been added. Read more in the corresponding README file.

OPENBSD SUPPORT

olsrd now also compiles on OpenBSD systems. Thanks a lot to Holger
Mauermann for the patches!

BUILDSYSTEM REWRITE

Bruno has rewritten the make system. OS detection is now done
automagically and the general design is much improved.

IP ADDRESS VALIDATION

Olsrd now maintains a set of IP addresses that is considered
invalid. Checks against this set is done on all received
OLSR packets announced originator addresses. The dataset can
be updated dynamically.

PLUGIN INTERFACE

Updated interface version to 3 and added more noticeable warning
for version missmatch.

MATH LIB DEPENDENCY

The dependency on the math lib came from the use of pow(3) in the
mantissa/exponent calculation. The calls to pow in this long forgotten
code, was very unnessecarry. The pow calls are replaced with bitwise
shifts and -lm is no longer needed.

ICMP REDIRECTS ON LINUX

On Linux systems "0" is now also written to the all/send_redirects
procfile and not only the pr. interface ones. Hopefully this resolves
problems reported with systems still sending ICMP redirects.

BUGFIXES

- Fixed init of IPv6 addrtype variable. It used to be uninitialized,
  causing it to default to site-local and have some weird side-effects.
- Fixed crash when changing IPv6 address at run-time
- Fixed a bug that prevented setting willingness 0 in the configfile
- Fixed bad handling of missing command line arguments
- Did various updates to prevent assertion when a neighbor
  has changed IP and sends HELLO before its olsrd precess has detected
  the change.
- Fixed return value check on times(2) calls in the scheduler
- Fixed bug(s) that could cause crash when an interface was brought
  down while olsrd was running.
- Removed the silent (read: unchecked) hard limit of 16 interfaces.
- Compiles now cleanly on 64bit (e.g. x86_64).

0.4.9 ---------------------------------------------------------------------

OPTIMIZATIONS

The MID database is now also indexed by alias addresses. this makes
look-ups of main addresses based on alias addresses _much_ faster.
The mid_lookup_main_addr function alone has gone from >30% of the
total CPU time used by olsrd to <1%.
PLUGIN DEVELOPERS: datatypes for MID entries has been changed.

Routing table calculation in the LQ case now uses balanced binary
trees instead of linked lists, which makes look-ups much faster.

The debug output function olsr_printf() is not used by olsrd
internals any more. Instead the macro OLSR_PRINTF is to be used.
This macro does debug level testing _before_ doing any function
calls(both printf itself and parameters to printf). This is a much
more optimized way of doing debug output. In addition olsrd can
be built without any debug output code if doing 'make NODEBUG=1'.

TIMEOUT OF DATASETS

Validity times for dataset entries are now set using the clock_t type
(used to be struct timeval). This data is now fetched from the times(2)
function instead of gettimeofday(2). This prevents possible problems
if the system time is changed.
PLUGIN DEVELOPERS: Note that datatypes for close to all dataset entries
(link, neighbor, mid, hna, tc etc.) has been changed!!

CONFIGURATION FILE

Multiple interfaces can now use the same interface configuration block.
Example:
Interface "if0" "if1" "if2"
{
        .....
}

Interface configurations now also support the "Weight" directive. When
olsrd finds multiple links to a neighbor it will choose the link
going via the interface with the lowest weight. Olsrd normally
sets this weight dynamically based on interface characteristics,
but if the user wishes to set a fixed weight in a interface(or
a group of interfaces), this directive is now available.

Moreover, interface configurations now support the "LinkQualityMult"
directive. It takes an IP address and a multiplier as arguments.

Example:

Interface "if0"
{
  [...]

  LinkQualityMult 192.168.0.1 0.7
  LinkQualityMult default 0.5

  [...]
}

The idea is to make certain links via the configured interface
artificially worse or better than they really are. The first parameter
specifies the IP address of a neighbour interface. The second
parameter is the multiplier to be applied to the link to the specified
neighbour interface.

After we have derived the link quality from the packet loss among the
packets received from the specified neighbour interface, we multiply
the link quality by the given multiplier. So, the multiplier changes
the LQ value that we use when determining the ETX, which is 1 / (LQ *
NLQ).

If "default" is given as the IP address then the multiplier applies to
all links via the configured interface. Note, however, that olsrd only
assigns one multiplier to each link. It does so in the following way.

* If there is a "LinkQualityMult" line that matches the IP address of
  the link's neighbour interface, then use the multiplier in this
  line.

* If there isn't any matching line, then look for a "default" line. If
  there is a "default" line, then use the multiplier in this "default"
  line.

* Otherwise use 1.0, i.e. do not modify the LQ value at all.

Hence, a line with an IP address has priority over a "default" line.

In the above example, the LQ value of the link between the local
interface if0 and the neighbour interface 192.168.0.1 would be
multiplied by 0.7. All other links between the local interface if0 and
a neighbour interface would be multiplied by 0.5.

IPV6 SOCKETS

The socket option IPV6_V6ONLY is now set on IPv6 sockets in linux.
This means that olsrd will no longer receive IPv4 traffic when
running in IPv6 mode.
This should enable users to actually run one olsrd instance using
IPv6 and one using IPv4 at the same time :)

PLUGINS

The httpinfo, dyn_gw and dot_draw plugins now all include olsrd
headerfiles directly. A plugin should define OLSR_PLUGIN before
including olsrd headers. Doing things this way removed the
redundant datatype definitions that has so often caused trouble
when things has been updated in olsrd.

NETBSD SUPPORT

olsrd now also compiles on NetBSD systems. Use 'make OS=nbsd' to
build. This port is not very well tested and feedback is greatly
appreciated.

EXPERIMENTAL MULTI-INTERFACE CODE FOR FREEBSD

Experimental code using libnet can be compiled in allowing multiple
interfaces to be used in FreeBSD.  You enable this by compiling using
OS=fbsd-ll.  You must have libnet-devel installed for this to work
(/usr/ports/net/libnet-devel).

OTHER UPDATES

- WLAN devices are now detected in FreeBSD
- First specified interfaces IP is used as main address.
- Broadcastflag is not checked on interface if fixed broadcast is
  set in config.
- Removed Linux link layer code as this was not really in use.
- Added -nofork command line option

BUGFIXES

- Link sensing now works according to announced vtime from
  neighbors. Previously links going symmetric->asymmetric
  were not necessarily detected until the holding time expired.
- A bug in route addition/removal that could lead to routes
  using GWs being added before the actual route to the GW
  was set up, has been fixed.
- A bug in the interface selection in route calculation has been
  fixed. New routes are no longer added prior to deleting old ones,
  since this caused trouble.
- A bug in the IPv6 multicast address configuration has been fixed.
  This caused olsrd not to work with global IPv6 addresses.
- A bug in the IPv6 prefix-from IPaddr function has been fixed. This
  bug caused HNA prefixes to always be calculated to 0.
- If a 2-hop neighbor is also a 1-hop neighbor, a bug made olsrd
  prefer a bad direct link to a better link via an MPR.
- If a link or interface lookup failed, olsrd crashed.
- Fixed a crash in the LQ packet generation code based on a
  signed/unsigned integer comparison.


0.4.8 ---------------------------------------------------------------------

LICENSE CHANGE

Olsrd is now distributed under a BSD style license. We believe olsrd will
be better off using this license as it opens up for commercial players
to use olsrd freely in their products. In the end this will lead to
this kind of users doing serious testing and bug fixing of olsrd.

ETX-LIKE LINK QUALITY DETECTION

We can now determine the packet loss on a link by looking at the serial
numbers of the OLSR messages received from a neighbor. This tells us how
many packets get through from our neighbor to us. We use a new extended
kind of HELLO messages (LQ_HELLO messages, LQ = link quality) to broadcast
the link quality that we have determined on our end of the link to our
neighbors. So do our neighbors, and we as well as they end up with an
idea of how good the link is in both directions. From the LQ_HELLOs we
also learn the link quality between our neighbors and our two-hop neigh-
bors. In this way we can select those neighbors as MPRs that have the
best links to our two-hop neighbors. To distribute the link quality
throughout the network, we use a new extended version of TC messages
(LQ_TC messages). They contain the qualities at both ends of each of our
links. Nodes can then run Dijkstra's algorithm to find a path between
themselves and other nodes that minimizes the packet loss.

If the newly introduced "LinkQualityLevel" is set to zero in the
configuration file, link quality is not used and the daemon behaves as
before, i.e. as specified by the OLSR RFC. If this parameter is set
to 1, LQ_HELLOs and LQ_TCs are used instead of HELLOs and TCs, link
quality is measured and MPRs are selected based on the link qualities.
If this parameter is set to 2, the routing table is additionally
calculated based on the link qualities.

Setting "LinkQualityLevel" to a non-zero value BREAKS COMPATIBILITY. You
will then not be able to participate in RFC-conformant OLSR networks
any longer. This is because we use LQ_HELLOs and LQ_TCs instead of HELLOs
and TCs in this case.

It's best practice to set "LinkQualityLevel" to the same value on all
nodes in a network.

When determining the quality of a link olsrd only considers a given
number of most recent OLSR packets received from its neighbors. By
default, olsrd looks at the 10 most recent OLSR packets. This can be
changed via the "LinkQualityWinSize" configuration option. Values
between 3 and 128 are legal.

NEW CONFIGFILE PARSER/FORMAT/SCHEME

A whole new config file syntax and parser is introduced with this
release. The parser is generated using flex and bison and it is
designed to be very modular. The parser can be compiled as either
a standalone binary, a shared library or as part of olsrd. This way
other applications can parse and generate olsrd configuration files
only by linking to the dynamic library.
In the new configfile options can be set pr. interface, plugin
parameters can be set, more IPC options can be set...and more.
The syntax is documented in the olsrd.conf(5) manual page found
in the files/ directory.
Olsrd internals now uses a single struct, as returned by the config
parser, for all configuration in runtime. this means that updating
values in this struct will dynamically update olsrd operation. It
also means that all configuration is now kept in one place instead
of spread all across the place.

FREEBSD AND MAC OSX PORT

This version offers an initial port of olsrd to FreeBSD and Mac OS
X. Like the Windows port it currently does not support IPv6. As this
is a very first try at supporting FreeBSD and Mac OS X, this port is
probably not as reliable as the Linux version. We'd greatly appreciate
feedback and bug reports. To compile you need GNU make. Then simply
run "gmake OS=fbsd" to build the executable.

NEW MAKEFILE(S)

The makefiles used to build olsrd have been rewritten and now includes
dependency file generation using makedep.

DEBUG OUTPUT

Tables are now only printed if changes occur, so there are no longer
any periodical output when using a debuglevel >0. A "heartbeat" is
now printed to STDOUT(if it is a terminal) in the form of a rotating
line to show that olsrd is actually operating.
Adding "ClearScreen yes" to the configuration file clears the screen
each time before the debug output shows updated information. This
makes the debug output easier to read in many cases. "ClearScreen no"
is the default, if no "ClearScreen" directive is given in the
configuration file.

PLUGIN INTERFACE

There are some changes in the plugin interface. It is now at
version 2. Plugin specific parameters from the configfile
using the PlParam option, will now be passed on to plugins.
To receive such parameters a plugin must implement a function:
int
register_olsr_param(char *key, char *value)
to which parameter pairs will be passed.

NO MORE THREADS

Olsrd no longer uses any thread library on its operation. Everything
now happens in the main thread allowing for use of olsrd on platforms
with no thread library.

ACPI SUPPORT IN WILLINGNESS CALCULATION(LINUX)

The willingness calculation for Linux, now also supports the more
modern ACPI proc interface in addition to APM. This is not well
tested and feedback is appreciated!

IPC OPTIONS

Various options regarding IPC connections can now be set in the
configfile. The options are max connections, allowed unicast
hosts and allowed net-ranges.

CODE REWRITES

Much of the message generation and net output code is rewritten.
Sending of partial messages should now work 100%. Also TC, MID
and HNA messages are now cached for a random amount of time
before transmitted - this often leads to the situation where these
messages are sent together with an HELLO message, which is highly
desirable since it saves network resources.
PLUGIN DEVELOPERS must note that the buffers and size variables
used in net output is no longer directly accessible!

NET OUTPUT CODE REWRITTEN

Every interface now has one outputbuffer registered. This allows
messages to be "cached" as explained above, so that multiple
messages are stacked together.
Also the outputbuffer and size are no longer accessible directly,
an API is available to plugins(and olsr code).

OS DEPENDENT NETWORK INTERFACE

The OS dependent network functions are now defined in net_os.h
instead of being mixed up with internal network functions
in net.h.

BUGFIXES

A bug that caused the announced ANSN sequence number to never stop
being increased when a change in the MPR selector set was detected
has been fixed. Another TC related bug that caused timed out TC
entries not to be deleted from the routing table in certain cases
has also been fixed. These bugs could cause unstable routes.
A bug that in many cases caused the wrong routes to be deleted
when using IPv6 has been fixed.


0.4.7 ---------------------------------------------------------------------

NATIVE WINDOWS GUI AND INSTALLER

We now have a native Windows GUI. The GTK+ version is no longer
supported on Windows. The new native GUI is pretty compact as it does
not require the GTK+ runtime DLLs. In addition to monitoring olsrds
state the GUI offers an easy way for novice users to configure
olsrd. It's pretty self-explanatory. Have a look at README-WIN32.txt
for details. The Windows version now also comes with an installer
based on the freely available Nullsoft Scriptable Install System
(NSIS). Looks like the Windows port is becoming more and more
Windows-ish. :-)

WIRELESS INTERFACE DETECTION ON WINDOWS

Like the Linux version the Windows port is now able to tell WLAN
interfaces from wired interfaces. We can now, for example, set
different HELLO intervals for WLAN interfaces and wired interfaces,
respectively.

ROUTE UPDATES

The order of updates of calculated routes is swapped. Now new routes
are added prior to deleting old ones. This could fix issues where
in some very very few cases one could experiment loss of routes.

BUGFIXES

There were quite some bugs in IPv6 operation in 0.4.6. These are
the bugs that were fixed:
- HNA message parsing did not work.
- Some IPv6 addresses were printed as IPv4 addresses in the
  debug output.
- The '-dispin' option now works again.
- A filedescriptor leak in the IPv6 interface detection caused
  olsrd to crash.


0.4.6 ---------------------------------------------------------------------

WINDOWS PORT!

Read the README-WIN32.txt file!

DYNAMIC NETWORK INTERFACE UPDATES

Network interfaces can now be updated and removed/added at
runtime. Olsrd will check all interfaces specified in the config
file or on the command-line, every 5 seconds. If updates
occur, olsrd will register this and update the necessary data.
If interfaces are removed or added, olsrd will detect this
and configure the interfaces. If an interface is removed
olsrd will continue to check the interface, so if say, a PCMCIA
card is removed and later on reinserted(and given the same name),
olsrd will detect this and start running on it again.

Note that olsrd will now run even if no interfaces are detected.

ROUTE CALCULATION UPDATE

All 1 hop neighbor addresses that are not directly reachable trough
a symmetric link in the link set will now be added with a GW in the
routing table. this GW will be an interface address from a registered
symmetric link to the neighbor.

MID UPDATE

Alias addresses are now registered in the link set processing
if available. This is not to pretty(since the vtime has to
be a mere guess) but it gives faster initial route updates.
If a HELLO is received(at initial registration) from an IP
that is not the same as the one set as main address in the
received HELLO header, then the IP from which the HELLO was
received will be registered as an alias of the main address
set in the HELLO header. The default vtime is set to 15 secs.

CODE RESTRUCTURING

All OS dependent interface configuration code is now located in
OS/ifnet.c(that would be linux/ifnet.c if using GNU/Linux).
The functions are available trough the headerfile src/ifnet.h

CODE CLEANUPS

- Some timer issues fixed
- Got rid of the global socket descriptors
- Updated indexing of network interfaces
- Introduced a global socket to use for ioctl calls(ioctl_s)
- Removed upper limit for interfaces to use
- Point-to-point interfaces are now allowed
- The main select(2) loop now uses a timeout
- A small bugfix in the select fd-set update code
- Some bugfixes in the configuration file parsing
- stdout/stderr are now set to not be buffered
- A bugfix in the MID generation regarding seq. numbering

PLUGIN INTERFCAE

A new set of functions that are called whenever a change in the
interface configuration(update, addition, removal) is made is
introduced.
These functions are added and removed very much like the ptf
functions. Available trough the functions add_ifchgf and del_ifchgf.

OBS OBS OBS!!! PLUGIN DEVELOPERS READ THIS!!!!

The interface struct(declared in interfaces.h) has changed.
Plugin developers must update plugins that uses the interface
struct!

The headerfile olsr_plugin_io.h which contains the plugin interface
commands now contains a revision list where all changes are
stated.


0.4.5 ---------------------------------------------------------------------

PLUGIN: SECURE OLSR

A plugin that generates and checks message signatures is added.
Check lib/secure

PLUGIN: TOPOLOGY GRAPH

A plugin that generates output of the topology in the dot format
is added. See lib/dot_draw

OPTIMIZATIONS

Some optimizations done that should cause fewer route recalculations.

BUGFIXES

Some bugs reported by Takafumi Tanaka fixed.

RESTORATION OF NETWORK SETTINGS

Network settings like disabling of ICMP redirects and spoof filter, are
now restored at exit by olsrd.

RFC COMPLIANCE FIX

Nodes would in certain scenarios retransmit messages originating for
themselves. This was a RFC incompliance. Reported by Ingmar Baumgart.

NON-WLAN TIMERS

No longer setting the HELLO interval of non-wlan interfaces as a multiplier
of the wlan interval. The interval is set seperatly now.
Due to this there are some updates in the configfile imperatives.

CODE CLEANUPS

Some new macros introduced for table insertion and removal and IP
copying and comparison.

FRONT END BUGFIXES

IPv6 bugfixes in the GUI.


0.4.4 ---------------------------------------------------------------------

ROUTE CALCULATION

Routes are no longer added via neighbors declaring a willingness
of WILL_NEVER.

LINK SENSING AND ROUTE CALCULATION

Link sensing and route calculation has been updated to handle
multiple links between hosts in a sane way.

IP SPOOF FILTERING

IP spoof filtering is disabled on the interfaces which olsrd runs.
These settings are not restored!

ICMP REDIRECTS

ICMP redirect message generation is disabled on the interfaces
which olsrd runs.
These settings are not restored!

CONFIGFILE OPTIONS FOR MID AND HNA ADDED

Options to set the MID and HNA emission intervals and
validity time in the configfile added.

IPv6 ADDRESS SCOPE

Option to set what IPv6 address type(scope) to use in the
configfile added. It can be set to either global or site-local.

IPv6 MULTICAST

Options to set site-local and global multicast addresses
in the configfile added.

EMISSION INTERVAL FUNCTION

Functions to set emission intervals at runtime added.

LINK LAYER NOTIFICATIONS

If started with the -llinfo switch olsrd will collect and display
link-layer information on neighbors. This is no longer limited to
one interface. But no action is taken based on this info - and
there is a upper limit of 8 neighbors from which this info can
be collected pr. interface. This limitation is inherited from the
WLAN drivers and is not something imposed by the olsrd code!

IPv6 HNA MESSAGE FORMAT

IPv6 HNA messages now contains 128-bit netmask instead of prefix.

PLUGIN FUNCTIONS

Added "packet transform functions" which allow plugins to alter all
outgoing OLSR traffic.

RESTRUCTURING

Moved most headers out of OS dependent directories. The tunneling
and link-layer notification interface is not yet separated from
the /linux directory.

MANUALPAGE

A manpage(olsrd(8)) describing olsrd has been created. It is installed
when using 'make install'.


0.4.3 ---------------------------------------------------------------------

PLUGIN SUPPORT

Olsrd now supports runtime-loadable plugins! A couple of example plugins
are located in the lib/ directory. Read the README files in the various
plugin directories for more info. A directive to load plugins have been
added to the configfile as well.
Documentation of the plugin interface can be fount at http://www.olsr.org.

The plugin interface is designed to be extendable. But it should _always_
be backwards compatible with the interface used in this release!

LINK-LAYER NOTIFICATION

Some preliminary link-layer notification code has been added. Link quality
can be written to stdout if olsrd is started with the -llinfo switch.
No action is currently taken based on this info. And info will only
be retrieved from the interface named "eth1".
This code is not usable for users as of yet!

MPR UPDATES

Neighbors with willingness set to WILL_ALWAYS is now always added. The
MPR selection is optimized as described in RFC3626 section 8.3.1
point 5.

MPR BUGS

There was a couple of bugs in the MPR selection code. All fixed.

LINUX 2.6 BUG

Users could not run olsrd on multiple interfaces using the Linux 2.6 kernel
series. This is now fixed.

MINOR BUGS

One small bug in link hysteresis initialization and one in HNA route
deletion.

MESSAGE-SEQUENCENUMBER INCOMPLIANCE

Message-seqnos was implemented on a pr. messagetype and pr. interface
basis. This is not RFC3626 compliant and has been updates. All messages
and interfaces now uses a global seuencenumber.

RANDOM SEQUENCENUMBERS

Sequencenumbers are now initialized using a random value.

DAEMON MODE

Fixes for running olsrd in daemon mode(debug 0).

OPTIMIZATIONS

Optimizations based on profiling implemented.

SANITATION

Message size is checked for every message before passing it to the
appropriate message parsing function. If the size is bigger than the
remaining size of the olsr packet then the message is discarded.

GUI FRONT-END

Some minor modifications done to the gui.

INTERNAL CHANGES

Lots of changes in handling of registration and unregistration of
scheduler, socketparser, parser and local hna set.
Some restructuring and movement of functions.


0.4.2 ---------------------------------------------------------------------

No public release

0.4.1 ---------------------------------------------------------------------

No public release

0.4.0 ---------------------------------------------------------------------

LINK HYSTERESIS

Link hysteresis as described in the RFC added.

TC REDUNDANCY

TC redundancy as described in the RFC added

MPR REDUNDANCY

MPR redundancy as described in the RFC added

SCHEDULER REWRITE

The scheduler has been rewritten to register scheduled
events(function pointers) dynamically.

PARSER REWRITE

The parser has been rewritten to register parse functions
on a pr. messagetype dynamically

MALLOC WRAPPER

A wrapper for the malloc(3) syscal has been introduced.
olsr_malloc(size_t, const char *) should ALWAYS be used.

CODE RESTRUCTURING

Lots of restructuring in headerfiles.
Linux spesific code moved to subdirectory linux/
FreeBSD code to be put in freebsd/
LOTS of code restructured due to the rexrite of the scheduler
and parser.

MPR CALCULATION

Optimizing the MPR set as suggested in the RFC section
8.3.1 point 5 added.

BUGFIXES

Some bugs in MPR selection fixed.


0.3.8 ---------------------------------------------------------------------

CODE REWRITES

HNA code completely rewritten!
The old code was really bad! Now it is much cleaner implemented and
should work fine.
One- and two-hop neighbor list structures rewritten. They now use
the same hashed double-linked lists as the rest of the information
sets use.

CODE RESTRUCTURING

Some code restructuring which resulted in the new files: timer.h,
mantissa.c, mantissa.h, hashing.c, hashing.h, linux/net.c
linux/net.h, linux/kernel_routes.c, linux/kernel_routes.h and some
renamed files

BUG FIXES

The bug causing errormessages on exit is fixed.
The bug that caused initial HNA routes in the GUI to display a hopcount
of 0 fixed.
IPv6 HNA netmasks were all added as /128. It's fixed now.

IPv6

Turns out sitelocal address had to be preferred to avoid IPv6 header
messing up link-sensing. Bug fixed


0.3.7 ---------------------------------------------------------------------

CODE REWRITES

The entire TC set and MPR selector set code rewritten. Hashing and
double linked lists added to the MID code.
Lots of HNA code has been rewritten as well.

HNA bugfix

When a node lost connectivity to a HNA GW all HNA entries to that gw
was deleted. This has been fixed - and HNA entries only gets deleted
on timeout now.
Other HNA bugs have been fixed as well. But the whole HNA implementation
needs a rewrite!

Holdingtime BUGfix

There was a bug that caused holdingtime not to be properly recalculated
when using non-default emission intervals.

Forwarding and symmetric neighbors

Seems I had misread the RFC on what a 1-hop symmetric neighbor is.
Now the link-set is checked when a check for a symmetric neighbor
is done. Not the 1 hop neighbor set as it used to be. This way
the hack when receiving MID messages could be removed.

IPv6

Global addresses are now preferred. Sitelocal addresses are only used
if no global addresses are found. There have been problems where
the address set in the IP header of IPv6 packets do not match the
address chosen by OLSR.

GUI - 0.2.5

The GUI is finally updated! It can now handle piggybacked messages
and link sensing HELLO message format.


0.3.6 ---------------------------------------------------------------------

HNA support using IPv6

HNA now works using IPv6 as well! Just add the network address and
prefix in the HNA6 section in the configfile. The IPv6 HNA packets differ
some from the IPv4 as they don't send the netmask but the prefix.


NEW FILENAME FOR THE BINARY AND CONFIG FILE

They are now called olsrd and olsrd.conf. I figured I'd use the olsrd
name before somebody else started using it ;-)


Config file

New options: DEBUG, HNA6 and IPC-CONNECT added to config file.


Output

A printf-wrapper has been introduced - debug level output handling
is much more uniform now.
The daemon should run fine in detached mode now(DEBUG 0)

GUI front-end

The GUI front-end can now connect and disconnect at any time while the
daemon is running, if started using the -ipc switch or with IPC set to
'yes' in the configfile.


BUGS

A small HNA bug fixed.


0.3.5 ---------------------------------------------------------------------

Forwarding jitter

Forwarding jitter added. Messages are piggybacked if they arrive(and are to
be forwarded) while there is data to be forwarded buffered.
Added some bugfixes to support parsing of piggybacked messages as well.

Route calculation

Fixed route calculation so that 2 hop neighbors are added. This smooths
changes from 1 hop to 2 hop neighbor out. The old way one had to wait for
a TC before the 2 hop neighbors were added.

2 hop neighbor set

All neighbors received in HELLO messages are added to the two hop
neighborhood. This helps smoothing out the 1-to-2 hop transition as well.

MID calculation

MID calculation had to be updated due to the registering of two hop neighbors
that are already one hop neighbors. They should not be considered when calculating
two hop coverage.

Optimizations

Deletion of possible one hop neighbors registered on non main-addresses when
first registering MID info from a node removed. This can not occur due
to the fact that registering of neighbor nodes are done on main addresses
from link-sensing
Option to not forward messages on wired links on which they arrived removed.
This is no longer usable to any degree when adding support for piggybacking
of forwarded messages is to be implemented.

Cleanups

Some unused parameters and commandline options were removed.



0.3.3 ---------------------------------------------------------------------

TC MESSAGEING

Added sending of empty TC messages if all MPR selectors are removed. Empty
TC messages are sent for a TC_HOLD_TIME period as described in section
9.3 in the RFC.

HNA

Fixed HNA route calculations for IPv6. I forgot to upgrade this in 0.3.2

OPTIONS

-hnaint and -midint command line optionas added to set the interval
of HNA and MID generation.

FORWARDING

As an optimization messages have not been forwarded on the received interface
if this is registered as a non-WLAN interface. This causes problems for
people using Ethernet-to-WLAN adapters. This optimization is now turned
off by default. It can be activated using the -nofwlan option.

BUGS

THERE ARE SOME NASTY BUGS IN 0.3.2!
Fixed a bug that caused 2 hop neighbors not to time out.
Fixed a bug that caused a 1 hop neighbor timeout not to recalculate
the routing table.
Fixed a small memleak in the MID set
Fixed some smaller bugs.

IPv6 functioning is not well tested in this release.

0.3.2 ---------------------------------------------------------------------

HNA

Updated HNA to be RFC compliant. This means that all received HNA tuples
are registered and only the subset of unique entries(net/netmask) with
the smallest hopcount is inserted into the routing table.

Config file

User can specify configfile name using the -f switch.

Bugs

Fixed a link-sensing bug that caused problems when using MID nodes.

0.3.1 ---------------------------------------------------------------------

MPR calculation

MPR calculation is now based on the willingness announced by nodes.

Neighbor set

The neighbor set does not have timeouts on entries any more. Creation
and deletion of neighbor entries is done from the link-set as suggested
in the RFC.

Cleanups

Gotten rid of a lot of code that became more or less obsolete due to
all the changes introduced in 0.3.0.

Bugs

Fixed a nasty bug concerning the sequence numbering of TC packets.
Fixed some bugs regarding MPR calculation - and probably introduced
a whole new species when implementing the willingness-based calculation ;-)


0.3.0 ---------------------------------------------------------------------

Due to the extensive amount of updates this release is versioned 0.3.0

Duplicate table and forwarding

The duplicate table functionality is totally rewritten in 100% RFC
compliance. The forward algorithm specified in the RFC is implemented
this meant rewriting much of the packet processing code. The daemon
also forwards unknown packettypes now.

Link sensing and neighbor processing

Link sensing and neighbor processing was done in the same operation in the
old code. Link sensing is now a mechanism of its own which is 100% RFC
compliant. Neighbor processing and HELLO generation has been partly
rewritten to be RFC compliant. Neighbors are now registered as either
SYN or NOT_SYM. The NOT_NEIGH, SYM_NEIGH and MPR_NEIGH values are only
used when building HELLO messages. Links are in one of the following states:
UNSPEC_LINK, ASYM_LINK or SYM_LINK.
HELLO messages now advertise link status of neighbors on the interface
the message is transmitted.

Willingness

A node now dynamically calculates willingness based on powersource and
possibly battery power. If the node is AC powered it announces a willingness
of 6. If batterypowered the willingness is calculated as:
P / 26
Where P is percentage of power left.

Configuration file

The daemon now tries to read its configuration from the file /etc/uolsrd.conf
A "standard" configuration file is installed when doing 'make install'.
This file should be edited to fit your needs.
Command-line options can still be used to override the configuration form
the file(ore if no config file is present).
The -f switch can also be used to specify a configuration file.

Willingness

If AdvancedPowerManagemant(APM) is supported on your system the willingness
of the node is dynamically calculated based upon the powerstatus of the node.
MPR calculation based upon willingness is not done as of yet.

TC ANSN

A bugs in the TC processing fixed.

HNA

Some serious HNA bugs fixed. These bugs caused a node to always choos the
gateway with the larges hopcout :) More seriously - a bug caused looping
of HNA messages in certain scenarios.

Package sequence numbers

These were added. Not much use as of now - but they're there(as specified
in the RFC).

Broadcast address

Users can now specify the broadcast address to use. Useful if one wishes to
use the 255.255.255.255 broadcast.



0.2.5 ---------------------------------------------------------------------

Htime and Vtime

Htime and Vtime processing and handling added. Holding times used to be
a static value for all nodes. Now every node can signal its vlidity time
to others.
To achieve this mantissa/exponent calculation functionality was added.
Messages now include Vtime(and Htime in HELLO) values in mantissa/exponent
format instead of just zero. These values are as stated above, used as holding
time for nodes.
the GUI should be upgraded as well...

Jitter

The use of jitter in message generation intervals vere fixed.


0.2.4 ---------------------------------------------------------------------

Seqno bugs

Some bugs concerning sequence numbers in HELLO and MID message handling
was fixed. Newly registered nodes in the neighbor and MID sets were
initialized with a seqno of 0. When using a "wraparound" sequencenumber
check(as defined in olsr_protocol.h) this can produce errors if a new
node joins a network where existing nodes has seqnos >0x00FF.

IPv6 related bugs

Some checks and copying of node addresses where done using sizeof(u_olsr32_t)
which is 32 bits. This caused only the 32 firs bits of IPv6 addresses to
be copied and checked.