Linux как операционная система в АТМ, почему нет предложений?
Аналогично с тем, почему в офисной среде все сидят в Windows. Ведь для банков ATM мало чем отличается от обычных пользовательских ПК с административной точки зрения.
А причина в этом проста: нет софта под необходимые нужды, нет необходимой технической поддержки, нет квалифицированного персонала, да и вообще нет необходимости.
Если вы жалуетесь на счет поддержки виндоус, ответьте на вопрос, сколько шли поддержка и обновления Windows XP сколько длились те же самые процедуры для какого-нибудь релиза Ubuntu.
Конечно, Windows XP — Дункан Маклауд по сравнению с любым LTS от Ubuntu, но разница в том, что переход, условно, с 12.04 на 14.04 буде стоить час-другой простоя, а переход с XP на 7 будет стоить столько же плюс стоимость новой лицензии. Или рядовой линуксоид стоит на рынке труда настолько дороже, чем рядовой виндузятник?
Это я не для холивара, мне так же интересно, как автору вопроса.
Валентин: ну так само собой это всё узнаётся перед обновлением, обкатывается на учебных банкоматах, шарится в доках. А разве винду приходят и с бухты барахты накатывают на банкомат? Тоже ведь предварительно ищут дрова, готовят сборки и т. п.
А что за проблемы с шифрованием? Разве банки используют что-то кроме стандартного RSA под соусом TLS? Да, скорее всего более ответственное отношение к ключам и какие-нибудь аппаратные токены, но для современного линукса аппаратный токен — не проблема.
Да, остаётся вопрос интерфейса пользователя и дров к купюроприёмникам и -выдавателям, считывателям карт. Интересно, а графические интерфейсы каждый банк под себя пишет, или есть какие-то фреймворки или конструкторы?
первое и основное — драйвера. Под винду они есть, под Линукс их нет. плюс поддержка Linux ровно также не бесконечная.
К слову — были какие-то ATM от белорусов(?) с линуксами на борту.
А еще остались банкоматы с полуосью! тут поддержки уж как 10 лет никакой нет, а работают 🙂 Часто таким устройствам поддержка и не требуется. Подозреваю, что окупаемость банкомата примерно год. за 3-5 лет можно просто новый поставить! Так что процесс сравним со старыми серверами — их просто заменяют! Отработал свое — в утиль, новый будет с новой операционкой, и не нужно ничего обновлять.
Ну, да, будет зоопарк из операционок, но это небольшая проблема, тем более, что стоимость самой ОС, дай Бог, составит 1% от стоимости банкомата (это наверняка OEM вся обрезанная и embedded, будет около $22 на устройство).
Windows Embedded 8.1 Industry, как и остальные продукты в Embedded-канале, имеет 15-летний цикл жизни. Это позволяет создавать решения на долгую перспективу, не заботясь о том, что вышла новая операционная система и предыдущая стала уже не доступна.
Правильно настроенная Windows XP, не столь критично уязвима из-за малой площади атаки. Подключение к АТМ по сети исключаются на уроне сетевого оборудования и настроенных политик, разрешающих подключаться с ограниченного списка IP-адресов. Запуск ПО так же ограничен белым списком, все что не в нем блокируется. Сделать что-либо более серьезное без подключения клавиатуры или мыши не получится, а доступ внутрь банкомата контролируется и ограничивается. Т.е. при взломе банкомата есть 2 сценария: 1) Получение сетевого доступа — т.е. взлом сети банка и подключение с внутренних ресурсов, или физическое переключение АТМ в свое сетевое оборудование (опять таки физический доступ). 2) Физический доступ к АТМ — если злоумышленник стоит рядом с банкоматом и может подключиться к его ПК, уже совершенно не важно какая ОС на нем стоит т.к. с 99% вероятностью запустят заранее подготовленный LiveCD.
Рекомендую прочитать статью. Все как обычно упирается в поддержку железа и ПО. Производителям устройств дорого поддерживать XFS стэк ПО для Windows и Linux одновременно, а с учетом что 99% АТМ работают на Windows, то выбор поддерживаемой платформы очевиден. Аналогично с ПО. Если железо не поддерживается, то и совместимость с Linux проверяется только по запросу конкретных банков, и не за бесплатно. Вот и получается, что банку для перехода на Linux необходимо все разработать с нуля, и проверить работоспособность всей системы. Проверка работы на реальном железе в бою опять-таки риск — можно выдать больше денег чем нужно, или меньше. Недовольство клиентов и разбор каждой спорной транзакции очень дорого обходится банку. Плюс к тому же нужно переобучать весь штат сотрудников, или набирать новых, которых тоже нужно обучать. Поэтому вывод прост: «Работает — не трогай».
Источник
Linux atm что это
This document describes how to install, setup, and configure the necessary drivers and tools to support ATM networking under Linux.
For the latest information, please check the ATM on Linux home page .
This document is largely derived from the Usage Instructions document that was included with the ATM on Linux distribution up until version 0.79. That previous document was written by Werner Almesberger > while he was at the Institute for computer Communications and Applications (ICA) .
Copyright 2001 IBM Corporation
A large portion of this document is derived from the Usage Instructions included with the ATM on Linux distribution up to version 0.79 which was released under the BSD License, GNU General Public License (GPL), and GNU Lesser General Public License (LGPL).
There is also a mailing list on which to discuss ATM on Linux. If you have any comments, questions, suggestions, or would just like to get involved, please join the list. You can subscribe and unsubscribe to it at http://lists.sourceforge.net/lists/listinfo/linux-atm-general .
Users are encouraged to continue to use the releases instead of automatically assuming they should grab the latest version out of CVS. However, if you like living on the edge, here is how to do it.
% cvs -z6 -d:pserver:anonymous@cvs.linux-atm.sourceforge.net.:/cvsroot/linux-atm co -P linux-atm
% cvs -z6 -d:pserver:anonymous@cvs.linux-atm.sourceforge.net.:/cvsroot/linux-atm co -r V2_5_0 linux-atm
After you have checked out the source tree, you will need to run the autotools script in the top level directory before you can configure, build, and install from that source tree:
# ./autotools Running aclocal. Running autoconf. Running autoheader. Running automake. automake: configure.in: installing `./install-sh’ automake: configure.in: installing `./mkinstalldirs’ automake: configure.in: installing `./missing’ configure.in: 26: required file `./ltconfig’ not found automake: Makefile.am: installing `./INSTALL’ automake: configure.in: installing `src/lane/ylwrap’ Finished. Now run ‘./configure’ and ‘make’.
If you wish to create a tarred, gzipped distribution file or a RPM distribution file, run make dist or make rpm respectively. The tarred, gzipped file will be placed in the top level of the source tree and the RPM file will be placed in the src/extra/RPMS directory.
Finally, if you would like to receive email including every diff that is committed to the repository as they go in, there is a mailing list called «linux-atm-commits»: http://lists.sourceforge.net/lists/listinfo/linux-atm-commits .
This mailing list should be treated as receive-only. NO discussion or questions are allowed (even of patches which are sent through that list). All discussion should be kept on the linux-atm-general mailing list.
In order to install this package, you’ll need
Perl, version 4 or 5
If you do not wish to futz with extracting and building the source yourself, the ATM tools are also distributed in RPM format. The RPM can be installed as follows:
rpm -ivh linux-atm-x.x.x-x.rpm
First, extract the ATM on Linux distribution:
tar xzvf linux-atm-x.x.x.tar.gz
Documentation (including this HOWTO) in SGML DocBook format
UNI 3.0, UNI 3.1, and UNI 4.0 signaling demon: atmsigd
Debugging tools: delay , ed , encopy , endump , svctor , zndump , and znth
Libraries for applications and demons
Miscellaneous man pages
Extra packages and RPM spec files.
Configuration and rc file examples
Switch fabric control (under construction)
If you are not familiar with building and installing a new kernel, please see the The Linux Kernel HOWTO
After unpacking the kernel distribution, do the usual make config , make menuconfig , or make xconfig in the top-level of your Linux kernel source tree. First, enable
Prompt for development and/or incomplete code/drivers (CONFIG_EXPERIMENTAL)
You should then be able to find the following options:
The burst settings of the ENI driver can be fine-tuned. This may be necessary if the default settings lead to buffer overruns in the PCI chipset. See the on-line help on «CONFIG_ATM_ENI_TUNE_BURST» for a detailed discussion of the implications of changing the burst settings.
Note that the file drivers/atm/nicstar.h contains a few configurable settings for the IDT 77201 driver.
Some drivers can also be used with certain compatible cards. The latest information about compatible cards can be found at ATM on Linux information page.
Then build your kernel and reboot.
If you’ve configured the ENI155p-MF driver, you should see two lines like these (512kB for the -C version, 2048kB for the -S version.):
Note that if you’ve configured only the ATM over TCP driver, there are no messages at startup, because ATM over TCP devices are created later using the atmtcp command.
If you want to enable debugging for options for memory allocations, you need to install MPR before compiling the ATM tools.
If you chose to download the binary RPM package, you can install MPR like so:
rpm -ivh mpr-x.x-x.rpm
If you chose to download the source, extract mpr-x.x.tar.gz like so:
tar xzvf mpr-x.x.tar.gz
cd mpr-x.x ./configure x86-linux make make install
Detection of some general mis-use of malloc and free is automatically performed if the program was compiled with MPR present. Tracing of allocations is enabled by setting MPRPC and MPRFI . See doc/mpr.html or doc/mpr.ps in the MPR distribution for details.
Only little run-time overhead is incurred if memory debugging is included, but those environment variables are not set.
Now, as the final step, configure and build the ATM tools. Configuration is only necessary if your switch uses UNI 3.1 or 4.0, or if it has certain bugs. The configuration options selected by passing the appropriate options to the ./configure script in the linux-atm distribution.
Issue ./configure —help from the top-level directory of the linux-atm distribution to view all possible options.
The ATM tools are built with the following commands:
cd linux-atm-x.x.x ./configure make make install
Some programs are based on large packages that are already distributed outside of the ATM context. For some packages, patches are contained in the ATM on Linux distribution. They are contained in the src/extra directory of the ATM on Linux distribution.
Currently, the following extra packages are available:
dumps network traffic (enhanced for ATM)
ATM name server (based on named 4.9.5)
Note that text2atm automatically uses ANS if available, so ans only needs to be installed on systems providing name server functionality or if ATM-aware maintenance tools nslookup , etc.) are needed.
A script hosts2ans.pl to convert a /etc/hosts.atm file to ANS zone files are provided in the src/extra/ANS/ directory. Its use is described at the beginning of the file.
This section describes device-specific configuration operations, and general diagnostic procedures at the ATM or SONET level. Please see the adapter documentation for details on hardware installation and diagnosis.
If you have no real ATM hardware, you can still exercise the API by using the ATM over TCP «driver». It emulates ATM devices which are directly wired to remote devices (i.e. there is no VPI/VCI swapping).
To establish one (bidirectional) «wire», become root on both systems (or run both sides on the same system to create two connected «interfaces») and run the following command on one of them (let’s call it «a»):
# atmtcp virtual listen
# atmtcp virtual connect address_of_a
Both atmtcp s will report on their progress and the kernel should display messages like:
Link 0: virtual interface 2 Link 1: incoming ATMTCP connection from 127.0.0.1
Link 0: virtual interface 3 Link 1: ATMTCP connection to localhost
Multiple «wires» can be attached to the same machine by specifying a port number (default is 2812). Note that no AAL processing is performed. It is therefore not possible to receive data using a different AAL (e.g. AAL0) than the one with which the data was sent.
The ZeitNet ZN1221 and ZN1225 adapters use pre-allocated pools of free memory buffers for receiving. Whenever a VC with a certain maximum SDU size is opened for receiving, the corresponding pool is filled with free buffers by the device driver. The adapter removes buffers while it receives data. When the number of remaining buffers falls below a certain threshold, the device driver replenishes the pool again.
The lower and the upper limits for the number of free buffers, and the threshold for adapting to a new data offset (see below for details), can be set using the zntune program. Usage:
The changes are applied to all pools if no pool number is specified. Pool 2 stores 64 bytes packets, pool 3 stores 128 bytes packets, etc. Pools 0 and 1 are currently unused.
The current settings and some usage statistics can be obtained by invoking zntune without specifying new parameters:
zntune [-z] itf [ pool ]
The «Size» column shows the buffer size in Bytes. The «Ref» column shows the number of open VCs using that pool. The «Alarm» column shows how many times the number of free buffers has fallen below the low-water mark since the counters were reset. Similarly, the «Under» column shows how many times an incoming PDU had to be discarded because the corresponding pool was empty.
The columns «Offs», «NxOf», «Count» and «Thres» show the alignment adaption status. «Offs» is the offset of user data the driver currently expects in incoming PDUs. For single-copy, receive buffers are aligned accordingly so that data is received at page boundaries. «NxOf» is the user data offset of the most recently received PDU, where the offset differs from the currently assumed offset. «Count» is the number of PDUs that have been received in sequence with an offset of «NxOf». Finally, «Thres» is the threshold value «Count» has to reach for «NxOf» to become the new current offset.
Use the -z option to reset the «Alarm» and «Under» counters.
Some status information about the ATM subsystem can be obtained through files in /proc/net/atm/ . The file /proc/net/atm/arp contains information specific to Classical IP over ATM, see section CLIP .
All active ATM devices are listed in /proc/net/atm/devices . For each device, the interface number, the type label, the end system identifier (ESI), and statistics are shown. The statistics correspond to the ones available via atmdiag .
Individual ATM devices may register entries of the form type:number (e.g. eni:0 ) which contain device-specific information.
The files /proc/net/atm/pvc and /proc/net/atm/svc list all PVC and SVC sockets. For both types of sockets, the interface, VPI and VCI numbers are shown. For PVCs, this is followed by the AAL and the traffic class and the selected PCR for the receive and the transmit direction. For SVCs, the SVC state and the address of the remote party are shown. SVCs with the interface number 999 are used for special control purposes as indicated in the «State» column.
Furthermore, /proc/net/atm/vc shows buffer sizes and additional internal information for all ATM sockets.
Various counters of the ATM device drivers can be queried with the atmdiag program. See the corresponding man page for details.
The SONET diagnostics tool can be used to monitor link performance and to simulate errors. In order to get current SONET statistics, run it with the ATM interface number as the argument, e.g.
The counters can be reset with the -z option:
The following network failures can be simulated:[1]
insert section errors (B1)
insert line errors (B2)
insert path errors (B3)
force (RX) frame loss
insert loss of signal
insert line alarm indication signal
insert path alarm indication signal
insert header checksum errors
A failure is enabled by adding the corresponding keyword on the command line. The failure is cleared by prefixing the keyword with a minus sign, e.g.
If any diagnostic error insertions are active, their keywords are shown when sonetdiag is used to obtain statistics. Note that some error insertions may be automatically switched off by the hardware.
PVCs can be used for machines that are either connected back to back or via a switch. In the latter case, the cell forwarding has to be manually set up at the switch.
aread / awrite and br / bw are simple programs to access the ATM API. awrite sends the text string passed as its second argument in an AAL5 PDU. aread receives one AAL5 PDU and displays it in hex. Both programs also display the return values of the corresponding system calls and the current values of errno .
bw either sends its standard input or a stream of blocks containing arbitrary data (if a number is passed as its fourth argument) in 8 kB AAL5 PDUs. br receives AAL5 PDUs and writes them to standard output.
The first argument of aread , awrite , br and bw is always the PVC address, i.e. the ATM interface number, the VPI and the VCI number, with a dot between elements. The interface number can be omitted if it is zero. Example:
Note that some adapters only support VPI == 0. Also, the VCI range may be limited, e.g 0 to 1023. The interface number can be obtained from the initialization message the driver printed during startup. atm0 is interface 0, atm1 is interface 1, etc. If the system is equipped with a real ATM adapter (e.g. not only atmtcp ), that adapter is normally at atm0 .
aping receives and sends small AAL5 PDUs on a PVC. It expects that messages it sends are either echoed back or that a similar program on the other side generates a stream of messages. aping reports an error if no messages are received for too long. aping is invoked by specifying the PVC, like aread .
For «real» tests, you should use the modified version of ttcp that comes with this package. The original is available at ftp://ftp.sgi.com/sgi/src/ttcp/ . The following options have been added:
use native ATM instead of UDP/TCP. The address must be in the format [ itf. ]vpi.vci for PVCs, or a valid ATM end system address for SVCs.
use a CBR connection with a peak cell rate of num cells per second. Default is to use UBR.
disable (UDP) checksums
%a ttcp_atm -r -a -s 0.90 %b ttcp_atm -t -a -s 0.90
On adapters where the device driver supports access to raw cells («AAL0»), individual cells can be composed and received with the atmdump program. Here is an example:
Because ATM addresses are inconvenient to use, most ATM tools also accept names instead of numeric addresses. The mapping between names and numbers is defined in the file /etc/hosts.atm . The structure of this file is similar to the /etc/hosts file:
Many ATM tools also attempt to find the corresponding name when displaying an address. When translating from the numeric form to a name, the first applicable name in the file is used.
In addition to ATM addresses for SVCs, also PVC addresses can be stored in /etc/hosts.atm . If different address types are stored under the same name, the first suitable one will be chosen, i.e. if an application explicitly requests only SVC addresses, any PVC addresses will be ignored.
If you have access to the ATM Name Service (ANS, e.g because you’ve installed the ANS extension), you can use it instead of or in addition to the hosts file by specifying the host that runs ANS in the /etc/resolv.conf file.
For performing reverse lookups of E.164 addresses, the list of telephony country codes needs to be known. That list can be obtained from the International Telecommunications Union . The List of ITU-T Recommendation E.164 Assigned Country Codes is currently available in PDF and Word document formats.
Should the URL become out of date, the document should easily be found by searching for the document’s title at the ITU web site.
The script src/lib/pdf2e164_cc.pl in the atm-linux distribution can be used to create the E.164 county codes table with the PDF version of the country code list, e.g.
perl pdf2e164_cc.pl e164_xxx.pdf >/etc/e164_cc
Man pages: atmsigd (8) atmsigd.conf (4)
Note that atmsigd ‘s support for point-to-multipoint is very limited: only operation as a single leaf of a point-to-multipoint tree works.
By default, atmsigd is configured to conform to dynamically configure the UNI version. It can be compiled for UNI 3.0, 3.1, or 4.0 specifically by passing the —with-uni=VERSION to the ./configure script in the top-level directory of the linux-atm source distribution.
Note that atmsigd is configured to be paranoid. If it detects unusual problems, it frequently terminates. This will (obviously) change in the future.
atmsigd also looks for a configuration file at the location specified with the -c option. The default location is /usr/local/etc/atmsigd.conf .
ILMI provides a mechanism for automatic address configuration. If there is no switch or if the switch doesn’t support ILMI, the ATM addresses must be configured manually (see section Manual Address Configuration ). Note that the ILMI demon should not be used on interfaces where addresses are manually configured.
background. Run in a forked child process after initializing.
enables debugging output. By default, ilmid is very quiet.
IP address to tell switch when asked for one. Can be in either dotted decimal or textual format. By default, ilmid uses some heuristics to select a local IP address.
write diagnostic messages to the specified file instead of to standard error. The special name syslog is used to send diagnostics to the system logger.
configures the ILMI VC to use the specified quality of service. By default, UBR at link speed is used on the ILMI VC.
set UNI version. Possible values are 3.0 , 3.1 , and 4.0 . The dot can be omitted. The default value depends on how ilmid was compiled. Typically, it is 3.0 .
enables extensive debugging output.
disable inclusion of variable bindings in the ColdstartTrap. Some switches (e.g. the LS100) only work if this option is set.
If no interface number is specified, ilmid serves interface 0. You can check whether address registration was successful with the atmaddr command (see below).
The agent supports only the address registration procedures specified in section 5.8 of the ATM Forum’s UNI 3.1 specification. These procedures involve the switch registering the network prefix on the host and the host registering the final ATM address back on the switch. The host accomplishes this by appending an ESI (End System Identifier) and a null selector byte to the network prefix registered by the switch. The ESI is the physical or MAC address of the ATM interface.
If your switch doesn’t support ILMI, you have to set the ATM address manually on the switch and on the PC(s). On the Linux side, make sure that ilmid doesn’t interfere, then use the atmaddr command to set the address(es).
Man pages: atmaddr (8)
Manual configuration of ATM addresses on the switch depends on the brand. On a Fore ASX-200, it can be done with the following command:
conf nsap route new nsap_addr 152 port vpi
conf nsap route new 47000580ffe1000000f21510650020ea000ee000 152 1a2 0 | || |^^ SEL
The entire NSAP address always has to have a length of 40 digits. Note that you can also use addresses with a different prefix and an ESI that doesn’t correspond to any ESI your adapters have. The value of the selector byte (SEL) is ignored.
It is also possible to run with two ATM NICs connected back-to-back, and no switch in between. This is great for simple test environments.
First, if you’re using UTP or STP-5, you need a suitable cable. Our experience with standard 100Base-T back-to-back cables was not good. It appears that the pin-out they use is different. After some false starts, we found that the following cable works:
You can also make up a loopback cable with 1 — 7 and 2 — 8 connected for ultra-cheap setups.
Here we have two machines called «virgil» and «nestor». Substitute your own names as necessary.
One side of the ATM connection needs to use the network version of atmsigd and the other side should use the normal user version. So here on nestor we start atmsigd with:
atmsigd -b -m network
Without a switch, you won’t be able to use ILMI. Instead, create a /etc/hosts.atm file containing two dummy addresses. Our ATM hosts file contains:
These are completely spurious addresses, of course, but as long as you’re not connected to a public or private ATM network, I don’t think it matters. To set the address correctly in the driver, we use:
atmaddr -a virgil-atm
atmaddr -a nestor-atm
The Q.2931 message compiler also generates a pretty-printer for Q.2931 messages. The executable is called q.dump is stored in the src/qgen directory. Note that it is not copied elsewhere by make install .
q.dump expects a sequence of whitespace-separated hex bytes at standard input and outputs the message structure if the message can be parsed. Example:
A demon process is used to generate and answer ARP queries. The actual kernel part maintains a small lookup table only containing partial information.
Man pages: atmarpd (8) , atmarp (8)
atmsigd and ilmid must already be running when atmarpd is started. Use the -b option to make sure they’re properly synchronized, e.g.
#!/bin/sh atmsigd -b ilmid -b atmarpd -b .
#!/bin/sh atmsigd & ilmid & atmarpd & .
The atmarp program is used to configure ATMARP. First, you have to start atmsigd , ilmid , and atmarpd , then create an IP interface and configure it:
# atmarp -c interface_name # ifconfig atm0 local_address possibly_more_options up
# atmarp -c atm0 # ifconfig atm0 10.0.0.3 up
If only PVCs will be used, they can now be created with a command like
# atmarp -s 10.0.0.4 0.0.70
NULL encapsulation is used if the null keyword is specified. Note that ARP requires LLC/SNAP encapsulation. NULL encapsulation can therefore only be used for PVCs.
When using SVCs, some additional configuration work may be necessary. If the machine is acting as the ATMARP server on that LIS, no additional configuration is required. Otherwise, the ATM address of the ATMARP server has to be configured. This is done by creating an entry for the network address with the option arpsrv set, e.g.
Note that the ATMARP server currently has to be started and configured before any clients are configured.
The kernel ATMARP table can be read via \path. The table used by atmarpd is regularly printed on standard error if atmarpd is started with the -d option. If atmarpd is invoked without -d , the table is written to the file atmarpd.table in the dump directory (by default /var/run ; can be changed with -D ), and it can be read with atmarp -a .
Besides Classical IP over ATM, LAN Emulation (LANE) can be used to carry IP over ATM. LANE emulates the characteristics of legacy LAN technology, such as support for broadcasts. LANE server support is described in the src/lane/USAGE file in the linux-atm distribution.
Man pages: bus (8) , lecs (8) , les (8) , and zeppelin (8)
If you plan to run more than one LANE clients, LANE service or LANE clients and LANE service, you need to specify different local ATM addresses for each demon. Since all the LANE demons use similar service access points (SAPs) they need different ATM addresses to differentiate between connections.
Just as with CLIP, the LANE client consists of two parts: a demon process called zeppelin which takes care of the LANE protocol and kernel part which contains LANE ARP cache.
atmsigd and ilmid must already be running when zeppelin is started. When zeppelin starts, the kernel creates a new interface which can then be configured:
# zeppelin possibly_more_options & # ifconfig lec0 local_address possibly_more_options up
In the example below, two LANE clients are started. The first client uses default interface lec0 , default listen address and tries to join the default ELAN. The other LANE client gets interface lec2 assigned to it, binds to local address mybox3 , tries to join ELAN called myelan and will bridge packets between ELAN and Ethernet segments. Address mybox3 is defined in /etc/hosts.atm . Rest of the bridging can be configured by reading the Bridging mini-HOWTO. [bridge-howto ]
By default, zeppelin uses interface lec0 , binds to local ATM address using selector byte value 0, tries to contact LECS using Well-Known LECS address, joins the default ELAN as defined by the LECS, accepts the MTU size as defined by the LES and will not act as an proxy LEC. These parameters can be tailored with command line options which are defined in zeppelin (8) .
zeppelin will automatically join any ELANs which use higher MTU than the default MTU of 1516 bytes. The MTU of the LANE interface will adjust itself according to the MTU of the current ELAN.
The state of the LANE ARP cache entries can be monitored through /proc/net/atm/lec . For each entry the MAC and ATM addresses and status is listed. If the entry has an active connection, the connection identifiers are also listed.
The LANE service ( lecs (8) , les (8) , and bus (8) ) is configured using configuration files. The configuration file syntax is listed on the respective manual pages.
A more detailed description of Linux LANE services is discussed in Marko Kiiskil�’s Master’s Thesis [kiis ].
The Linux MPOA client continues the tradition of user space — kernel divided ATM services. The demon process called mpcd processes MPOA control packets while the kernel holds MPOA ingress and egress caches and does the packet forwarding.
Man page: mpcd (8)
atmsigd and ilmid must already be running when mpcd is started. Since MPOA detects IP layer flows from LANE traffic, you need to have zeppelin running before MPOA can function. However, the order in which zeppelin and mpcd is started is not fixed. You can kill any of the demons at your will and restart it later without need to restart the other demon. The easiest way to disable MPOA is to kill the running mpcd .
The MPOA demon needs two different local ATM addresses which it uses when initiating and receiving data and control connections. The addresses can be the same as with e.g. zeppelin but must be different among other mpcd demons. By default, mpcd does not retrieve configuration information from the LECS. The necessary command line options and an example of using LECS are shown on the mpcd manual page. The manual page also lists the rest of the available options.
The contents of MPOA ingress and egress caches can be monitored through the /proc/net/atm/mpc file.
The Linux MPOA client also supports CBR traffic class for shortcuts SVCs instead of default UBR. The QoS specifications for future shortcuts can be set and modified using /proc/net/atm/mpc .
