요약·해설과 원문, 전문 번역을 서로 분리했습니다. API 이름, symbol, source path는 원문 표기를 사용합니다.
1. 요약·해설
원문의 핵심 논리와 kernel programming 관점의 보충 설명입니다. 아래의 전문 번역과는 별도로 작성했습니다.
2. 영어 원문 전체
번역 기준이 된 Linux v6.18.37 원문입니다. 줄 번호는 이 버전의 파일 좌표입니다.
원문 전체 펼치기
.. SPDX-License-Identifier: GPL-2.0
======
ARCnet
======
.. note::
See also arcnet-hardware.txt in this directory for jumper-setting
and cabling information if you're like many of us and didn't happen to get a
manual with your ARCnet card.
Since no one seems to listen to me otherwise, perhaps a poem will get your
attention::
This driver's getting fat and beefy,
But my cat is still named Fifi.
Hmm, I think I'm allowed to call that a poem, even though it's only two
lines. Hey, I'm in Computer Science, not English. Give me a break.
The point is: I REALLY REALLY REALLY REALLY REALLY want to hear from you if
you test this and get it working. Or if you don't. Or anything.
ARCnet 0.32 ALPHA first made it into the Linux kernel 1.1.80 - this was
nice, but after that even FEWER people started writing to me because they
didn't even have to install the patch. <sigh>
Come on, be a sport! Send me a success report!
(hey, that was even better than my original poem... this is getting bad!)
.. warning::
If you don't e-mail me about your success/failure soon, I may be forced to
start SINGING. And we don't want that, do we?
(You know, it might be argued that I'm pushing this point a little too much.
If you think so, why not flame me in a quick little e-mail? Please also
include the type of card(s) you're using, software, size of network, and
whether it's working or not.)
My e-mail address is: [email protected]
These are the ARCnet drivers for Linux.
This new release (2.91) has been put together by David Woodhouse
<[email protected]>, in an attempt to tidy up the driver after adding support
for yet another chipset. Now the generic support has been separated from the
individual chipset drivers, and the source files aren't quite so packed with
#ifdefs! I've changed this file a bit, but kept it in the first person from
Avery, because I didn't want to completely rewrite it.
The previous release resulted from many months of on-and-off effort from me
(Avery Pennarun), many bug reports/fixes and suggestions from others, and in
particular a lot of input and coding from Tomasz Motylewski. Starting with
ARCnet 2.10 ALPHA, Tomasz's all-new-and-improved RFC1051 support has been
included and seems to be working fine!
Where do I discuss these drivers?
---------------------------------
Tomasz has been so kind as to set up a new and improved mailing list.
Subscribe by sending a message with the BODY "subscribe linux-arcnet YOUR
REAL NAME" to [email protected]. Then, to submit messages to the
list, mail to [email protected].
There are archives of the mailing list at:
http://epistolary.org/mailman/listinfo.cgi/arcnet
The people on [email protected] (now defunct, replaced by
[email protected]) have also been known to be very helpful, especially
when we're talking about ALPHA Linux kernels that may or may not work right
in the first place.
Other Drivers and Info
----------------------
You can try my ARCNET page on the World Wide Web at:
http://www.qis.net/~jschmitz/arcnet/
Also, SMC (one of the companies that makes ARCnet cards) has a WWW site you
might be interested in, which includes several drivers for various cards
including ARCnet. Try:
http://www.smc.com/
Performance Technologies makes various network software that supports
ARCnet:
http://www.perftech.com/ or ftp to ftp.perftech.com.
Novell makes a networking stack for DOS which includes ARCnet drivers. Try
FTPing to ftp.novell.com.
You can get the Crynwr packet driver collection (including arcether.com, the
one you'll want to use with ARCnet cards) from
oak.oakland.edu:/simtel/msdos/pktdrvr. It won't work perfectly on a 386+
without patches, though, and also doesn't like several cards. Fixed
versions are available on my WWW page, or via e-mail if you don't have WWW
access.
Installing the Driver
---------------------
All you will need to do in order to install the driver is::
make config
(be sure to choose ARCnet in the network devices
and at least one chipset driver.)
make clean
make zImage
If you obtained this ARCnet package as an upgrade to the ARCnet driver in
your current kernel, you will need to first copy arcnet.c over the one in
the linux/drivers/net directory.
You will know the driver is installed properly if you get some ARCnet
messages when you reboot into the new Linux kernel.
There are four chipset options:
1. Standard ARCnet COM90xx chipset.
This is the normal ARCnet card, which you've probably got. This is the only
chipset driver which will autoprobe if not told where the card is.
It following options on the command line::
com90xx=[<io>[,<irq>[,<shmem>]]][,<name>] | <name>
If you load the chipset support as a module, the options are::
io=<io> irq=<irq> shmem=<shmem> device=<name>
To disable the autoprobe, just specify "com90xx=" on the kernel command line.
To specify the name alone, but allow autoprobe, just put "com90xx=<name>"
2. ARCnet COM20020 chipset.
This is the new chipset from SMC with support for promiscuous mode (packet
sniffing), extra diagnostic information, etc. Unfortunately, there is no
sensible method of autoprobing for these cards. You must specify the I/O
address on the kernel command line.
The command line options are::
com20020=<io>[,<irq>[,<node_ID>[,backplane[,CKP[,timeout]]]]][,name]
If you load the chipset support as a module, the options are::
io=<io> irq=<irq> node=<node_ID> backplane=<backplane> clock=<CKP>
timeout=<timeout> device=<name>
The COM20020 chipset allows you to set the node ID in software, overriding the
default which is still set in DIP switches on the card. If you don't have the
COM20020 data sheets, and you don't know what the other three options refer
to, then they won't interest you - forget them.
3. ARCnet COM90xx chipset in IO-mapped mode.
This will also work with the normal ARCnet cards, but doesn't use the shared
memory. It performs less well than the above driver, but is provided in case
you have a card which doesn't support shared memory, or (strangely) in case
you have so many ARCnet cards in your machine that you run out of shmem slots.
If you don't give the IO address on the kernel command line, then the driver
will not find the card.
The command line options are::
com90io=<io>[,<irq>][,<name>]
If you load the chipset support as a module, the options are:
io=<io> irq=<irq> device=<name>
4. ARCnet RIM I cards.
These are COM90xx chips which are _completely_ memory mapped. The support for
these is not tested. If you have one, please mail the author with a success
report. All options must be specified, except the device name.
Command line options::
arcrimi=<shmem>,<irq>,<node_ID>[,<name>]
If you load the chipset support as a module, the options are::
shmem=<shmem> irq=<irq> node=<node_ID> device=<name>
Loadable Module Support
-----------------------
Configure and rebuild Linux. When asked, answer 'm' to "Generic ARCnet
support" and to support for your ARCnet chipset if you want to use the
loadable module. You can also say 'y' to "Generic ARCnet support" and 'm'
to the chipset support if you wish.
::
make config
make clean
make zImage
make modules
If you're using a loadable module, you need to use insmod to load it, and
you can specify various characteristics of your card on the command
line. (In recent versions of the driver, autoprobing is much more reliable
and works as a module, so most of this is now unnecessary.)
For example::
cd /usr/src/linux/modules
insmod arcnet.o
insmod com90xx.o
insmod com20020.o io=0x2e0 device=eth1
Using the Driver
----------------
If you build your kernel with ARCnet COM90xx support included, it should
probe for your card automatically when you boot. If you use a different
chipset driver complied into the kernel, you must give the necessary options
on the kernel command line, as detailed above.
Go read the NET-2-HOWTO and ETHERNET-HOWTO for Linux; they should be
available where you picked up this driver. Think of your ARCnet as a
souped-up (or down, as the case may be) Ethernet card.
By the way, be sure to change all references from "eth0" to "arc0" in the
HOWTOs. Remember that ARCnet isn't a "true" Ethernet, and the device name
is DIFFERENT.
Multiple Cards in One Computer
------------------------------
Linux has pretty good support for this now, but since I've been busy, the
ARCnet driver has somewhat suffered in this respect. COM90xx support, if
compiled into the kernel, will (try to) autodetect all the installed cards.
If you have other cards, with support compiled into the kernel, then you can
just repeat the options on the kernel command line, e.g.::
LILO: linux com20020=0x2e0 com20020=0x380 com90io=0x260
If you have the chipset support built as a loadable module, then you need to
do something like this::
insmod -o arc0 com90xx
insmod -o arc1 com20020 io=0x2e0
insmod -o arc2 com90xx
The ARCnet drivers will now sort out their names automatically.
How do I get it to work with...?
--------------------------------
NFS:
Should be fine linux->linux, just pretend you're using Ethernet cards.
oak.oakland.edu:/simtel/msdos/nfs has some nice DOS clients. There
is also a DOS-based NFS server called SOSS. It doesn't multitask
quite the way Linux does (actually, it doesn't multitask AT ALL) but
you never know what you might need.
With AmiTCP (and possibly others), you may need to set the following
options in your Amiga nfstab: MD 1024 MR 1024 MW 1024
(Thanks to Christian Gottschling <[email protected]>
for this.)
Probably these refer to maximum NFS data/read/write block sizes. I
don't know why the defaults on the Amiga didn't work; write to me if
you know more.
DOS:
If you're using the freeware arcether.com, you might want to install
the driver patch from my web page. It helps with PC/TCP, and also
can get arcether to load if it timed out too quickly during
initialization. In fact, if you use it on a 386+ you REALLY need
the patch, really.
Windows:
See DOS :) Trumpet Winsock works fine with either the Novell or
Arcether client, assuming you remember to load winpkt of course.
LAN Manager and Windows for Workgroups:
These programs use protocols that
are incompatible with the Internet standard. They try to pretend
the cards are Ethernet, and confuse everyone else on the network.
However, v2.00 and higher of the Linux ARCnet driver supports this
protocol via the 'arc0e' device. See the section on "Multiprotocol
Support" for more information.
Using the freeware Samba server and clients for Linux, you can now
interface quite nicely with TCP/IP-based WfWg or Lan Manager
networks.
Windows 95:
Tools are included with Win95 that let you use either the LANMAN
style network drivers (NDIS) or Novell drivers (ODI) to handle your
ARCnet packets. If you use ODI, you'll need to use the 'arc0'
device with Linux. If you use NDIS, then try the 'arc0e' device.
See the "Multiprotocol Support" section below if you need arc0e,
you're completely insane, and/or you need to build some kind of
hybrid network that uses both encapsulation types.
OS/2:
I've been told it works under Warp Connect with an ARCnet driver from
SMC. You need to use the 'arc0e' interface for this. If you get
the SMC driver to work with the TCP/IP stuff included in the
"normal" Warp Bonus Pack, let me know.
ftp.microsoft.com also has a freeware "Lan Manager for OS/2" client
which should use the same protocol as WfWg does. I had no luck
installing it under Warp, however. Please mail me with any results.
NetBSD/AmiTCP:
These use an old version of the Internet standard ARCnet
protocol (RFC1051) which is compatible with the Linux driver v2.10
ALPHA and above using the arc0s device. (See "Multiprotocol ARCnet"
below.) ** Newer versions of NetBSD apparently support RFC1201.
Using Multiprotocol ARCnet
--------------------------
The ARCnet driver v2.10 ALPHA supports three protocols, each on its own
"virtual network device":
====== ===============================================================
arc0 RFC1201 protocol, the official Internet standard which just
happens to be 100% compatible with Novell's TRXNET driver.
Version 1.00 of the ARCnet driver supported _only_ this
protocol. arc0 is the fastest of the three protocols (for
whatever reason), and allows larger packets to be used
because it supports RFC1201 "packet splitting" operations.
Unless you have a specific need to use a different protocol,
I strongly suggest that you stick with this one.
arc0e "Ethernet-Encapsulation" which sends packets over ARCnet
that are actually a lot like Ethernet packets, including the
6-byte hardware addresses. This protocol is compatible with
Microsoft's NDIS ARCnet driver, like the one in WfWg and
LANMAN. Because the MTU of 493 is actually smaller than the
one "required" by TCP/IP (576), there is a chance that some
network operations will not function properly. The Linux
TCP/IP layer can compensate in most cases, however, by
automatically fragmenting the TCP/IP packets to make them
fit. arc0e also works slightly more slowly than arc0, for
reasons yet to be determined. (Probably it's the smaller
MTU that does it.)
arc0s The "[s]imple" RFC1051 protocol is the "previous" Internet
standard that is completely incompatible with the new
standard. Some software today, however, continues to
support the old standard (and only the old standard)
including NetBSD and AmiTCP. RFC1051 also does not support
RFC1201's packet splitting, and the MTU of 507 is still
smaller than the Internet "requirement," so it's quite
possible that you may run into problems. It's also slower
than RFC1201 by about 25%, for the same reason as arc0e.
The arc0s support was contributed by Tomasz Motylewski
and modified somewhat by me. Bugs are probably my fault.
====== ===============================================================
You can choose not to compile arc0e and arc0s into the driver if you want -
this will save you a bit of memory and avoid confusion when eg. trying to
use the "NFS-root" stuff in recent Linux kernels.
The arc0e and arc0s devices are created automatically when you first
ifconfig the arc0 device. To actually use them, though, you need to also
ifconfig the other virtual devices you need. There are a number of ways you
can set up your network then:
1. Single Protocol.
This is the simplest way to configure your network: use just one of the
two available protocols. As mentioned above, it's a good idea to use
only arc0 unless you have a good reason (like some other software, ie.
WfWg, that only works with arc0e).
If you need only arc0, then the following commands should get you going::
ifconfig arc0 MY.IP.ADD.RESS
route add MY.IP.ADD.RESS arc0
route add -net SUB.NET.ADD.RESS arc0
[add other local routes here]
If you need arc0e (and only arc0e), it's a little different::
ifconfig arc0 MY.IP.ADD.RESS
ifconfig arc0e MY.IP.ADD.RESS
route add MY.IP.ADD.RESS arc0e
route add -net SUB.NET.ADD.RESS arc0e
arc0s works much the same way as arc0e.
2. More than one protocol on the same wire.
Now things start getting confusing. To even try it, you may need to be
partly crazy. Here's what *I* did. :) Note that I don't include arc0s in
my home network; I don't have any NetBSD or AmiTCP computers, so I only
use arc0s during limited testing.
I have three computers on my home network; two Linux boxes (which prefer
RFC1201 protocol, for reasons listed above), and one XT that can't run
Linux but runs the free Microsoft LANMAN Client instead.
Worse, one of the Linux computers (freedom) also has a modem and acts as
a router to my Internet provider. The other Linux box (insight) also has
its own IP address and needs to use freedom as its default gateway. The
XT (patience), however, does not have its own Internet IP address and so
I assigned it one on a "private subnet" (as defined by RFC1597).
To start with, take a simple network with just insight and freedom.
Insight needs to:
- talk to freedom via RFC1201 (arc0) protocol, because I like it
more and it's faster.
- use freedom as its Internet gateway.
That's pretty easy to do. Set up insight like this::
ifconfig arc0 insight
route add insight arc0
route add freedom arc0 /* I would use the subnet here (like I said
to in "single protocol" above),
but the rest of the subnet
unfortunately lies across the PPP
link on freedom, which confuses
things. */
route add default gw freedom
And freedom gets configured like so::
ifconfig arc0 freedom
route add freedom arc0
route add insight arc0
/* and default gateway is configured by pppd */
Great, now insight talks to freedom directly on arc0, and sends packets
to the Internet through freedom. If you didn't know how to do the above,
you should probably stop reading this section now because it only gets
worse.
Now, how do I add patience into the network? It will be using LANMAN
Client, which means I need the arc0e device. It needs to be able to talk
to both insight and freedom, and also use freedom as a gateway to the
Internet. (Recall that patience has a "private IP address" which won't
work on the Internet; that's okay, I configured Linux IP masquerading on
freedom for this subnet).
So patience (necessarily; I don't have another IP number from my
provider) has an IP address on a different subnet than freedom and
insight, but needs to use freedom as an Internet gateway. Worse, most
DOS networking programs, including LANMAN, have braindead networking
schemes that rely completely on the netmask and a 'default gateway' to
determine how to route packets. This means that to get to freedom or
insight, patience WILL send through its default gateway, regardless of
the fact that both freedom and insight (courtesy of the arc0e device)
could understand a direct transmission.
I compensate by giving freedom an extra IP address - aliased 'gatekeeper' -
that is on my private subnet, the same subnet that patience is on. I
then define gatekeeper to be the default gateway for patience.
To configure freedom (in addition to the commands above)::
ifconfig arc0e gatekeeper
route add gatekeeper arc0e
route add patience arc0e
This way, freedom will send all packets for patience through arc0e,
giving its IP address as gatekeeper (on the private subnet). When it
talks to insight or the Internet, it will use its "freedom" Internet IP
address.
You will notice that we haven't configured the arc0e device on insight.
This would work, but is not really necessary, and would require me to
assign insight another special IP number from my private subnet. Since
both insight and patience are using freedom as their default gateway, the
two can already talk to each other.
It's quite fortunate that I set things up like this the first time (cough
cough) because it's really handy when I boot insight into DOS. There, it
runs the Novell ODI protocol stack, which only works with RFC1201 ARCnet.
In this mode it would be impossible for insight to communicate directly
with patience, since the Novell stack is incompatible with Microsoft's
Ethernet-Encap. Without changing any settings on freedom or patience, I
simply set freedom as the default gateway for insight (now in DOS,
remember) and all the forwarding happens "automagically" between the two
hosts that would normally not be able to communicate at all.
For those who like diagrams, I have created two "virtual subnets" on the
same physical ARCnet wire. You can picture it like this::
[RFC1201 NETWORK] [ETHER-ENCAP NETWORK]
(registered Internet subnet) (RFC1597 private subnet)
(IP Masquerade)
/---------------\ * /---------------\
| | * | |
| +-Freedom-*-Gatekeeper-+ |
| | | * | |
\-------+-------/ | * \-------+-------/
| | |
Insight | Patience
(Internet)
It works: what now?
-------------------
Send mail describing your setup, preferably including driver version, kernel
version, ARCnet card model, CPU type, number of systems on your network, and
list of software in use to me at the following address:
I do send (sometimes automated) replies to all messages I receive. My email
can be weird (and also usually gets forwarded all over the place along the
way to me), so if you don't get a reply within a reasonable time, please
resend.
It doesn't work: what now?
--------------------------
Do the same as above, but also include the output of the ifconfig and route
commands, as well as any pertinent log entries (ie. anything that starts
with "arcnet:" and has shown up since the last reboot) in your mail.
If you want to try fixing it yourself (I strongly recommend that you mail me
about the problem first, since it might already have been solved) you may
want to try some of the debug levels available. For heavy testing on
D_DURING or more, it would be a REALLY good idea to kill your klogd daemon
first! D_DURING displays 4-5 lines for each packet sent or received. D_TX,
D_RX, and D_SKB actually DISPLAY each packet as it is sent or received,
which is obviously quite big.
Starting with v2.40 ALPHA, the autoprobe routines have changed
significantly. In particular, they won't tell you why the card was not
found unless you turn on the D_INIT_REASONS debugging flag.
Once the driver is running, you can run the arcdump shell script (available
from me or in the full ARCnet package, if you have it) as root to list the
contents of the arcnet buffers at any time. To make any sense at all out of
this, you should grab the pertinent RFCs. (some are listed near the top of
arcnet.c). arcdump assumes your card is at 0xD0000. If it isn't, edit the
script.
Buffers 0 and 1 are used for receiving, and Buffers 2 and 3 are for sending.
Ping-pong buffers are implemented both ways.
If your debug level includes D_DURING and you did NOT define SLOW_XMIT_COPY,
the buffers are cleared to a constant value of 0x42 every time the card is
reset (which should only happen when you do an ifconfig up, or when Linux
decides that the driver is broken). During a transmit, unused parts of the
buffer will be cleared to 0x42 as well. This is to make it easier to figure
out which bytes are being used by a packet.
You can change the debug level without recompiling the kernel by typing::
ifconfig arc0 down metric 1xxx
/etc/rc.d/rc.inet1
where "xxx" is the debug level you want. For example, "metric 1015" would put
you at debug level 15. Debug level 7 is currently the default.
Note that the debug level is (starting with v1.90 ALPHA) a binary
combination of different debug flags; so debug level 7 is really 1+2+4 or
D_NORMAL+D_EXTRA+D_INIT. To include D_DURING, you would add 16 to this,
resulting in debug level 23.
If you don't understand that, you probably don't want to know anyway.
E-mail me about your problem.
I want to send money: what now?
-------------------------------
Go take a nap or something. You'll feel better in the morning.
3. 한국어 전문 번역
영어 원문의 문단 순서와 의미를 유지한 전체 번역입니다. 코드, 함수명, symbol과 URL은 원문 표기를 유지합니다.
머리말과 driver 계보
1-61`.. SPDX-License-Identifier: GPL-2.0`
ARCnet
참고: ARCnet card manual을 받지 못한 많은 사용자와 같은 상황이라면, jumper 설정과 배선 정보는 같은 directory의 `arcnet-hardware.txt`도 함께 보십시오.
작성자는 다른 방식으로는 아무도 자신의 말을 듣지 않는 듯하다며 다음 두 줄짜리 시로 관심을 끌려고 합니다.
이 driver는 살이 붙어 튼튼해지는데,
내 고양이 이름은 여전히 Fifi라네.두 줄뿐이어도 시라고 부를 수 있을 것이라며, 자신은 영문학이 아니라 컴퓨터 과학을 전공하니 너그럽게 봐 달라고 농담합니다.
핵심 요청은 이 driver를 시험해 성공했든 실패했든, 또는 어떤 결과든 반드시 알려 달라는 것입니다. ARCnet 0.32 ALPHA가 Linux kernel 1.1.80에 처음 포함된 뒤 patch를 따로 설치할 필요가 없어지자 오히려 보고가 더 줄었다고 아쉬워합니다. 성공 보고를 보내 달라는 문구와 원래 시보다 더 나은 운을 맞췄다는 농담도 이어집니다.
경고: 곧 성공·실패 결과를 e-mail로 보내지 않으면 작성자가 노래를 시작할 수도 있다고 합니다. 이 요청이 지나치다고 생각한다면 짧은 항의 메일이라도 보내되, 사용 card 종류, software, network 규모, 작동 여부를 함께 적어 달라고 요청합니다. 당시 주소는 `[email protected]`입니다.
이 문서는 Linux용 ARCnet driver를 설명합니다. Release 2.91은 David Woodhouse `<[email protected]>`가 새 chipset 지원을 추가한 뒤 driver를 정리하면서 만들었습니다. Generic support를 개별 chipset driver와 분리해 source의 `#ifdef` 밀도를 낮췄고, 문서는 완전히 다시 쓰지 않기 위해 Avery의 1인칭 표현을 유지했습니다.
이전 release는 Avery Pennarun의 여러 달에 걸친 간헐적 작업, 다른 사람들의 bug report·수정·제안, 특히 Tomasz Motylewski의 많은 의견과 code로 만들어졌습니다. ARCnet 2.10 ALPHA부터 Tomasz가 새로 작성하고 개선한 RFC1051 지원이 포함되었으며 정상적으로 작동하는 것으로 보였습니다.
.. SPDX-License-Identifier: GPL-2.0
======
ARCnet
======
.. note::
See also arcnet-hardware.txt in this directory for jumper-setting
and cabling information if you're like many of us and didn't happen to get a
manual with your ARCnet card.
Since no one seems to listen to me otherwise, perhaps a poem will get your
attention::
This driver's getting fat and beefy,
But my cat is still named Fifi.
Hmm, I think I'm allowed to call that a poem, even though it's only two
lines. Hey, I'm in Computer Science, not English. Give me a break.
The point is: I REALLY REALLY REALLY REALLY REALLY want to hear from you if
you test this and get it working. Or if you don't. Or anything.
ARCnet 0.32 ALPHA first made it into the Linux kernel 1.1.80 - this was
nice, but after that even FEWER people started writing to me because they
didn't even have to install the patch. <sigh>
Come on, be a sport! Send me a success report!
(hey, that was even better than my original poem... this is getting bad!)
.. warning::
If you don't e-mail me about your success/failure soon, I may be forced to
start SINGING. And we don't want that, do we?
(You know, it might be argued that I'm pushing this point a little too much.
If you think so, why not flame me in a quick little e-mail? Please also
include the type of card(s) you're using, software, size of network, and
whether it's working or not.)
My e-mail address is: [email protected]
These are the ARCnet drivers for Linux.
This new release (2.91) has been put together by David Woodhouse
<[email protected]>, in an attempt to tidy up the driver after adding support
for yet another chipset. Now the generic support has been separated from the
individual chipset drivers, and the source files aren't quite so packed with
#ifdefs! I've changed this file a bit, but kept it in the first person from
Avery, because I didn't want to completely rewrite it.
The previous release resulted from many months of on-and-off effort from me
(Avery Pennarun), many bug reports/fixes and suggestions from others, and in
particular a lot of input and coding from Tomasz Motylewski. Starting with
ARCnet 2.10 ALPHA, Tomasz's all-new-and-improved RFC1051 support has been
included and seems to be working fine!
토론 창구와 다른 driver 정보
62-108이 driver를 어디에서 논의하는가?
Tomasz는 개선된 mailing list를 마련했습니다. 당시에는 본문에 `subscribe linux-arcnet YOUR REAL NAME`을 적어 `[email protected]`로 보내 구독하고, 글은 `[email protected]`로 보냈습니다.
Mailing list archive 주소로 `http://epistolary.org/mailman/listinfo.cgi/arcnet`가 기록되어 있습니다. 현재는 폐지되고 `[email protected]`로 대체된 `[email protected]` 사람들도, 특히 동작 여부가 불확실한 ALPHA Linux kernel을 다룰 때 도움이 되었다고 설명합니다.
다른 driver와 정보
- 작성자의 ARCNET web page: `http://www.qis.net/~jschmitz/arcnet/`
- ARCnet card 제조사 SMC와 여러 card driver: `http://www.smc.com/`
- ARCnet 지원 network software를 만들던 Performance Technologies: `http://www.perftech.com/` 또는 `ftp.perftech.com`
- ARCnet driver가 포함된 Novell DOS networking stack: `ftp.novell.com`
- Crynwr packet driver 모음과 ARCnet용 `arcether.com`: `oak.oakland.edu:/simtel/msdos/pktdrvr`
`arcether.com`은 patch 없이는 386 이상에서 완전하게 작동하지 않고 일부 card도 싫어합니다. 수정판은 당시 작성자의 web page에서 받거나 WWW 접근이 없으면 e-mail로 요청할 수 있었습니다.
Where do I discuss these drivers?
---------------------------------
Tomasz has been so kind as to set up a new and improved mailing list.
Subscribe by sending a message with the BODY "subscribe linux-arcnet YOUR
REAL NAME" to [email protected]. Then, to submit messages to the
list, mail to [email protected].
There are archives of the mailing list at:
http://epistolary.org/mailman/listinfo.cgi/arcnet
The people on [email protected] (now defunct, replaced by
[email protected]) have also been known to be very helpful, especially
when we're talking about ALPHA Linux kernels that may or may not work right
in the first place.
Other Drivers and Info
----------------------
You can try my ARCNET page on the World Wide Web at:
http://www.qis.net/~jschmitz/arcnet/
Also, SMC (one of the companies that makes ARCnet cards) has a WWW site you
might be interested in, which includes several drivers for various cards
including ARCnet. Try:
http://www.smc.com/
Performance Technologies makes various network software that supports
ARCnet:
http://www.perftech.com/ or ftp to ftp.perftech.com.
Novell makes a networking stack for DOS which includes ARCnet drivers. Try
FTPing to ftp.novell.com.
You can get the Crynwr packet driver collection (including arcether.com, the
one you'll want to use with ARCnet cards) from
oak.oakland.edu:/simtel/msdos/pktdrvr. It won't work perfectly on a 386+
without patches, though, and also doesn't like several cards. Fixed
versions are available on my WWW page, or via e-mail if you don't have WWW
access.
설치와 네 가지 chipset driver
109-194Driver 설치
`make config`에서 network device의 ARCnet과 하나 이상의 chipset driver를 선택한 뒤 `make clean`, `make zImage`를 실행합니다. 현재 kernel의 ARCnet driver를 upgrade package로 교체한다면 먼저 `arcnet.c`를 `linux/drivers/net`의 기존 file 위에 복사해야 했습니다. 새 kernel로 reboot할 때 ARCnet message가 보이면 정상 설치된 것입니다.
Chipset 선택지는 네 가지입니다.
1. 표준 ARCnet COM90xx chipset: 대부분 사용자가 가진 일반 card이며, card 위치를 지정하지 않아도 자동 probe하는 유일한 chipset driver입니다. Kernel command line은 `com90xx=[<io>[,<irq>[,<shmem>]]][,<name>] | <name>`, module parameter는 `io=<io> irq=<irq> shmem=<shmem> device=<name>`입니다. `com90xx=`만 지정하면 autoprobe를 끄고, `com90xx=<name>`은 이름만 지정하면서 probe는 허용합니다.
2. ARCnet COM20020 chipset: promiscuous mode(packet sniffing), 추가 진단 정보 등을 지원하는 SMC의 새 chipset입니다. 합리적인 autoprobe 방법이 없으므로 kernel command line에서 I/O address를 반드시 지정해야 합니다.
Command line은 `com20020=<io>[,<irq>[,<node_ID>[,backplane[,CKP[,timeout]]]]][,name]`입니다. Module parameter는 `io=<io> irq=<irq> node=<node_ID> backplane=<backplane> clock=<CKP> timeout=<timeout> device=<name>`입니다. Software에서 node ID를 설정해 card DIP switch의 기본값을 덮어쓸 수 있습니다. COM20020 datasheet가 없고 마지막 세 option의 뜻을 모른다면 필요하지 않으므로 무시해도 됩니다.
3. I/O-mapped mode의 ARCnet COM90xx: 일반 ARCnet card에서도 작동하지만 shared memory를 사용하지 않습니다. 위 driver보다 성능은 낮지만 shared memory를 지원하지 않는 card를 쓰거나, 한 시스템에 ARCnet card가 너무 많아 shmem slot이 모자란 특수 상황을 위해 제공됩니다. Kernel command line에 I/O address를 주지 않으면 card를 찾지 않습니다.
Command line은 `com90io=<io>[,<irq>][,<name>]`, module parameter는 `io=<io> irq=<irq> device=<name>`입니다.
4. ARCnet RIM I card: 완전히 memory-mapped된 COM90xx chip입니다. 당시 지원은 시험되지 않았으므로 보유자는 성공 결과를 작성자에게 보내야 했습니다. Device name을 제외한 모든 option이 필수입니다.
Command line은 `arcrimi=<shmem>,<irq>,<node_ID>[,<name>]`, module parameter는 `shmem=<shmem> irq=<irq> node=<node_ID> device=<name>`입니다.
Installing the Driver
---------------------
All you will need to do in order to install the driver is::
make config
(be sure to choose ARCnet in the network devices
and at least one chipset driver.)
make clean
make zImage
If you obtained this ARCnet package as an upgrade to the ARCnet driver in
your current kernel, you will need to first copy arcnet.c over the one in
the linux/drivers/net directory.
You will know the driver is installed properly if you get some ARCnet
messages when you reboot into the new Linux kernel.
There are four chipset options:
1. Standard ARCnet COM90xx chipset.
This is the normal ARCnet card, which you've probably got. This is the only
chipset driver which will autoprobe if not told where the card is.
It following options on the command line::
com90xx=[<io>[,<irq>[,<shmem>]]][,<name>] | <name>
If you load the chipset support as a module, the options are::
io=<io> irq=<irq> shmem=<shmem> device=<name>
To disable the autoprobe, just specify "com90xx=" on the kernel command line.
To specify the name alone, but allow autoprobe, just put "com90xx=<name>"
2. ARCnet COM20020 chipset.
This is the new chipset from SMC with support for promiscuous mode (packet
sniffing), extra diagnostic information, etc. Unfortunately, there is no
sensible method of autoprobing for these cards. You must specify the I/O
address on the kernel command line.
The command line options are::
com20020=<io>[,<irq>[,<node_ID>[,backplane[,CKP[,timeout]]]]][,name]
If you load the chipset support as a module, the options are::
io=<io> irq=<irq> node=<node_ID> backplane=<backplane> clock=<CKP>
timeout=<timeout> device=<name>
The COM20020 chipset allows you to set the node ID in software, overriding the
default which is still set in DIP switches on the card. If you don't have the
COM20020 data sheets, and you don't know what the other three options refer
to, then they won't interest you - forget them.
3. ARCnet COM90xx chipset in IO-mapped mode.
This will also work with the normal ARCnet cards, but doesn't use the shared
memory. It performs less well than the above driver, but is provided in case
you have a card which doesn't support shared memory, or (strangely) in case
you have so many ARCnet cards in your machine that you run out of shmem slots.
If you don't give the IO address on the kernel command line, then the driver
will not find the card.
The command line options are::
com90io=<io>[,<irq>][,<name>]
If you load the chipset support as a module, the options are:
io=<io> irq=<irq> device=<name>
4. ARCnet RIM I cards.
These are COM90xx chips which are _completely_ memory mapped. The support for
these is not tested. If you have one, please mail the author with a success
report. All options must be specified, except the device name.
Command line options::
arcrimi=<shmem>,<irq>,<node_ID>[,<name>]
If you load the chipset support as a module, the options are::
shmem=<shmem> irq=<irq> node=<node_ID> device=<name>
Loadable module 지원
195-222Loadable Module 지원
Linux를 구성하고 다시 build할 때 loadable module을 쓰려면 `Generic ARCnet support`와 해당 ARCnet chipset support에 `m`을 답합니다. Generic support는 `y`, chipset support만 `m`으로 선택해도 됩니다.
`make config`, `make clean`, `make zImage`, `make modules` 순서로 build합니다.
Loadable module을 사용한다면 `insmod`로 load하며 command line에서 card 특성을 지정할 수 있습니다. 문서가 기록한 당시의 최신 driver에서는 autoprobe가 훨씬 안정적이고 module에서도 작동하므로 대부분의 수동 지정은 더 이상 필요하지 않았습니다.
예시는 module directory에서 `insmod arcnet.o`, `insmod com90xx.o`, 또는 `insmod com20020.o io=0x2e0 device=eth1`을 실행합니다.
Loadable Module Support
-----------------------
Configure and rebuild Linux. When asked, answer 'm' to "Generic ARCnet
support" and to support for your ARCnet chipset if you want to use the
loadable module. You can also say 'y' to "Generic ARCnet support" and 'm'
to the chipset support if you wish.
::
make config
make clean
make zImage
make modules
If you're using a loadable module, you need to use insmod to load it, and
you can specify various characteristics of your card on the command
line. (In recent versions of the driver, autoprobing is much more reliable
and works as a module, so most of this is now unnecessary.)
For example::
cd /usr/src/linux/modules
insmod arcnet.o
insmod com90xx.o
insmod com20020.o io=0x2e0 device=eth1
Driver 사용과 여러 card
223-261Driver 사용
Kernel에 ARCnet COM90xx support를 포함해 build하면 boot할 때 card를 자동 probe합니다. 다른 chipset driver를 kernel에 포함했다면 앞에서 설명한 필수 option을 kernel command line에 지정해야 합니다.
Linux의 NET-2-HOWTO와 ETHERNET-HOWTO를 읽고 ARCnet을 성능이 높거나 낮은 Ethernet card처럼 생각할 수 있습니다. 다만 HOWTO의 모든 `eth0` 참조를 `arc0`로 바꿔야 합니다. ARCnet은 진짜 Ethernet이 아니므로 device 이름이 다릅니다.
한 컴퓨터의 여러 card
Linux 자체는 이를 잘 지원했지만 ARCnet driver는 작성자의 시간 부족으로 다중 card 지원이 다소 뒤처졌습니다. Kernel에 포함된 COM90xx support는 설치된 모든 card를 자동 탐지하려고 시도합니다.
다른 chipset support를 kernel에 포함한 경우 command line option을 반복합니다. 예: `LILO: linux com20020=0x2e0 com20020=0x380 com90io=0x260`.
Chipset support가 loadable module이면 `insmod -o arc0 com90xx`, `insmod -o arc1 com20020 io=0x2e0`, `insmod -o arc2 com90xx`처럼 서로 다른 module instance 이름을 사용합니다. ARCnet driver가 device 이름을 자동으로 정리합니다.
Using the Driver
----------------
If you build your kernel with ARCnet COM90xx support included, it should
probe for your card automatically when you boot. If you use a different
chipset driver complied into the kernel, you must give the necessary options
on the kernel command line, as detailed above.
Go read the NET-2-HOWTO and ETHERNET-HOWTO for Linux; they should be
available where you picked up this driver. Think of your ARCnet as a
souped-up (or down, as the case may be) Ethernet card.
By the way, be sure to change all references from "eth0" to "arc0" in the
HOWTOs. Remember that ARCnet isn't a "true" Ethernet, and the device name
is DIFFERENT.
Multiple Cards in One Computer
------------------------------
Linux has pretty good support for this now, but since I've been busy, the
ARCnet driver has somewhat suffered in this respect. COM90xx support, if
compiled into the kernel, will (try to) autodetect all the installed cards.
If you have other cards, with support compiled into the kernel, then you can
just repeat the options on the kernel command line, e.g.::
LILO: linux com20020=0x2e0 com20020=0x380 com90io=0x260
If you have the chipset support built as a loadable module, then you need to
do something like this::
insmod -o arc0 com90xx
insmod -o arc1 com20020 io=0x2e0
insmod -o arc2 com90xx
The ARCnet drivers will now sort out their names automatically.
다른 protocol·운영체제와 연동
262-330무엇과 함께 작동하게 만드는가?
NFS: Linux 간에는 Ethernet card를 쓰는 것처럼 다루면 됩니다. `oak.oakland.edu:/simtel/msdos/nfs`에는 DOS client가 있었고, multitasking을 전혀 하지 않는 DOS 기반 NFS server SOSS도 있었습니다.
AmiTCP 등에서는 Amiga `nfstab`에 `MD 1024 MR 1024 MW 1024`를 지정해야 할 수 있습니다. 이는 아마 NFS data/read/write의 최대 block size를 뜻하지만 Amiga 기본값이 실패한 이유는 작성자도 알지 못했습니다.
DOS: Freeware `arcether.com`을 쓴다면 작성자의 web page에 있던 driver patch를 설치하는 것이 좋습니다. PC/TCP를 돕고 초기화 timeout이 너무 빨라 load되지 않는 문제를 완화합니다. 386 이상에서는 이 patch가 사실상 필수라고 강조합니다.
Windows: DOS 절을 따릅니다. `winpkt`를 load하면 Trumpet Winsock이 Novell client나 Arcether client와 정상 작동합니다.
LAN Manager와 Windows for Workgroups: 이들은 Internet 표준과 호환되지 않는 protocol로 card를 Ethernet처럼 보이게 하므로 다른 node를 혼란스럽게 합니다. Linux ARCnet driver 2.00 이상은 `arc0e` device로 이 protocol을 지원합니다. TCP/IP 기반 WfWg·LAN Manager network라면 Linux용 Samba server와 client로도 연동할 수 있습니다.
Windows 95: NDIS 방식 LANMAN driver 또는 ODI 방식 Novell driver로 ARCnet packet을 처리할 수 있습니다. ODI면 Linux의 `arc0`, NDIS면 `arc0e`를 사용합니다. 두 encapsulation을 함께 쓰는 hybrid network는 multiprotocol 절을 참고합니다.
OS/2: SMC ARCnet driver를 사용한 Warp Connect에서는 `arc0e` interface로 작동한다고 전해졌습니다. `ftp.microsoft.com`의 무료 Lan Manager for OS/2 client도 WfWg와 같은 protocol을 사용할 것으로 보였지만 작성자는 Warp에서 설치에 성공하지 못했습니다.
NetBSD/AmiTCP: Linux driver 2.10 ALPHA 이상의 `arc0s` device와 호환되는 이전 Internet 표준 RFC1051을 사용합니다. 더 최신 NetBSD version은 RFC1201도 지원하는 것으로 기록되어 있습니다.
How do I get it to work with...?
--------------------------------
NFS:
Should be fine linux->linux, just pretend you're using Ethernet cards.
oak.oakland.edu:/simtel/msdos/nfs has some nice DOS clients. There
is also a DOS-based NFS server called SOSS. It doesn't multitask
quite the way Linux does (actually, it doesn't multitask AT ALL) but
you never know what you might need.
With AmiTCP (and possibly others), you may need to set the following
options in your Amiga nfstab: MD 1024 MR 1024 MW 1024
(Thanks to Christian Gottschling <[email protected]>
for this.)
Probably these refer to maximum NFS data/read/write block sizes. I
don't know why the defaults on the Amiga didn't work; write to me if
you know more.
DOS:
If you're using the freeware arcether.com, you might want to install
the driver patch from my web page. It helps with PC/TCP, and also
can get arcether to load if it timed out too quickly during
initialization. In fact, if you use it on a 386+ you REALLY need
the patch, really.
Windows:
See DOS :) Trumpet Winsock works fine with either the Novell or
Arcether client, assuming you remember to load winpkt of course.
LAN Manager and Windows for Workgroups:
These programs use protocols that
are incompatible with the Internet standard. They try to pretend
the cards are Ethernet, and confuse everyone else on the network.
However, v2.00 and higher of the Linux ARCnet driver supports this
protocol via the 'arc0e' device. See the section on "Multiprotocol
Support" for more information.
Using the freeware Samba server and clients for Linux, you can now
interface quite nicely with TCP/IP-based WfWg or Lan Manager
networks.
Windows 95:
Tools are included with Win95 that let you use either the LANMAN
style network drivers (NDIS) or Novell drivers (ODI) to handle your
ARCnet packets. If you use ODI, you'll need to use the 'arc0'
device with Linux. If you use NDIS, then try the 'arc0e' device.
See the "Multiprotocol Support" section below if you need arc0e,
you're completely insane, and/or you need to build some kind of
hybrid network that uses both encapsulation types.
OS/2:
I've been told it works under Warp Connect with an ARCnet driver from
SMC. You need to use the 'arc0e' interface for this. If you get
the SMC driver to work with the TCP/IP stuff included in the
"normal" Warp Bonus Pack, let me know.
ftp.microsoft.com also has a freeware "Lan Manager for OS/2" client
which should use the same protocol as WfWg does. I had no luck
installing it under Warp, however. Please mail me with any results.
NetBSD/AmiTCP:
These use an old version of the Internet standard ARCnet
protocol (RFC1051) which is compatible with the Linux driver v2.10
ALPHA and above using the arc0s device. (See "Multiprotocol ARCnet"
below.) ** Newer versions of NetBSD apparently support RFC1201.
Multiprotocol ARCnet 개요
331-383Multiprotocol ARCnet 사용
ARCnet driver 2.10 ALPHA는 각각 별도의 virtual network device로 세 protocol을 지원합니다.
`arc0`: 공식 Internet 표준 RFC1201이며 Novell TRXNET driver와 100% 호환됩니다. Driver 1.00은 이 protocol만 지원했습니다. 세 protocol 중 가장 빠르고 RFC1201 packet splitting으로 큰 packet도 처리합니다. 특별한 이유가 없다면 이 방식을 강하게 권장합니다.
`arc0e`: Ethernet과 비슷한 packet을 ARCnet 위로 보내는 Ethernet Encapsulation이며 6-byte hardware address도 포함합니다. WfWg·LANMAN의 Microsoft NDIS ARCnet driver와 호환됩니다. MTU 493은 TCP/IP가 요구하는 576보다 작아 일부 operation이 실패할 가능성이 있지만 Linux TCP/IP layer가 대개 자동 fragmentation으로 보완합니다. 더 작은 MTU 탓으로 추정되며 `arc0`보다 조금 느립니다.
`arc0s`: 이전 Internet 표준인 simple RFC1051 protocol입니다. 새 표준과 완전히 호환되지 않지만 당시 NetBSD와 AmiTCP 등은 구표준만 지원했습니다. RFC1201 packet splitting을 지원하지 않고 MTU 507도 Internet 요구값보다 작아 문제가 생길 수 있습니다. `arc0e`와 같은 이유로 RFC1201보다 약 25% 느립니다. Tomasz Motylewski가 지원 code를 기여했고 Avery가 수정했습니다.
원하면 `arc0e`와 `arc0s`를 compile하지 않아 memory를 절약하고 최근 kernel의 NFS-root 설정에서 생길 혼란을 피할 수 있습니다.
`arc0`를 처음 `ifconfig`하면 `arc0e`와 `arc0s`도 자동 생성됩니다. 실제 사용하려면 필요한 virtual device 각각에도 `ifconfig`를 실행해야 하며, 이후 network 구성 방식은 여러 가지입니다.
Using Multiprotocol ARCnet
--------------------------
The ARCnet driver v2.10 ALPHA supports three protocols, each on its own
"virtual network device":
====== ===============================================================
arc0 RFC1201 protocol, the official Internet standard which just
happens to be 100% compatible with Novell's TRXNET driver.
Version 1.00 of the ARCnet driver supported _only_ this
protocol. arc0 is the fastest of the three protocols (for
whatever reason), and allows larger packets to be used
because it supports RFC1201 "packet splitting" operations.
Unless you have a specific need to use a different protocol,
I strongly suggest that you stick with this one.
arc0e "Ethernet-Encapsulation" which sends packets over ARCnet
that are actually a lot like Ethernet packets, including the
6-byte hardware addresses. This protocol is compatible with
Microsoft's NDIS ARCnet driver, like the one in WfWg and
LANMAN. Because the MTU of 493 is actually smaller than the
one "required" by TCP/IP (576), there is a chance that some
network operations will not function properly. The Linux
TCP/IP layer can compensate in most cases, however, by
automatically fragmenting the TCP/IP packets to make them
fit. arc0e also works slightly more slowly than arc0, for
reasons yet to be determined. (Probably it's the smaller
MTU that does it.)
arc0s The "[s]imple" RFC1051 protocol is the "previous" Internet
standard that is completely incompatible with the new
standard. Some software today, however, continues to
support the old standard (and only the old standard)
including NetBSD and AmiTCP. RFC1051 also does not support
RFC1201's packet splitting, and the MTU of 507 is still
smaller than the Internet "requirement," so it's quite
possible that you may run into problems. It's also slower
than RFC1201 by about 25%, for the same reason as arc0e.
The arc0s support was contributed by Tomasz Motylewski
and modified somewhat by me. Bugs are probably my fault.
====== ===============================================================
You can choose not to compile arc0e and arc0s into the driver if you want -
this will save you a bit of memory and avoid confusion when eg. trying to
use the "NFS-root" stuff in recent Linux kernels.
The arc0e and arc0s devices are created automatically when you first
ifconfig the arc0 device. To actually use them, though, you need to also
ifconfig the other virtual devices you need. There are a number of ways you
can set up your network then:
단일 protocol 구성
384-4071. 단일 protocol
가장 단순한 구성은 한 protocol만 사용하는 것입니다. WfWg처럼 `arc0e`만 지원하는 software가 있는 등의 이유가 없다면 `arc0`만 쓰는 편이 좋습니다.
`arc0`만 필요하면 `ifconfig arc0 MY.IP.ADD.RESS`로 주소를 설정하고, host route와 subnet route를 `arc0`에 추가한 뒤 필요한 local route를 더합니다.
`arc0e`만 필요하면 먼저 `arc0`와 `arc0e` 모두에 같은 IP address를 설정하되, host·subnet route는 `arc0e`로 추가합니다. `arc0s`도 `arc0e`와 거의 같은 방식으로 구성합니다.
1. Single Protocol.
This is the simplest way to configure your network: use just one of the
two available protocols. As mentioned above, it's a good idea to use
only arc0 unless you have a good reason (like some other software, ie.
WfWg, that only works with arc0e).
If you need only arc0, then the following commands should get you going::
ifconfig arc0 MY.IP.ADD.RESS
route add MY.IP.ADD.RESS arc0
route add -net SUB.NET.ADD.RESS arc0
[add other local routes here]
If you need arc0e (and only arc0e), it's a little different::
ifconfig arc0 MY.IP.ADD.RESS
ifconfig arc0e MY.IP.ADD.RESS
route add MY.IP.ADD.RESS arc0e
route add -net SUB.NET.ADD.RESS arc0e
arc0s works much the same way as arc0e.
한 wire에서 여러 protocol 사용
408-4552. 같은 wire에서 둘 이상의 protocol 사용
여기서부터 구성이 복잡해집니다. 작성자의 home network에는 Linux system 두 대와 Linux를 실행할 수 없어 무료 Microsoft LANMAN Client를 쓰는 XT 한 대가 있었습니다. Linux system들은 앞서 설명한 이유로 RFC1201을 선호하며, NetBSD나 AmiTCP system이 없어 `arc0s`는 제한된 test 때만 사용했습니다.
Linux system `freedom`은 modem도 보유하고 Internet provider로 가는 router 역할을 합니다. 다른 Linux system `insight`는 자체 Internet IP address가 있고 `freedom`을 default gateway로 사용합니다. XT `patience`에는 Internet IP address가 없으므로 RFC1597 private subnet의 주소를 할당했습니다.
먼저 `insight`와 `freedom`만 있는 단순 network를 구성합니다. `insight`는 더 빠르고 선호하는 RFC1201(`arc0`)로 `freedom`과 통신하며, `freedom`을 Internet gateway로 사용해야 합니다.
`insight`에서는 `arc0`에 `insight` 주소를 설정하고 자신과 `freedom`으로 가는 route를 추가한 뒤 `freedom`을 default gateway로 설정합니다. 전체 subnet route를 쓰지 않은 이유는 나머지 subnet이 `freedom`의 PPP link 건너편에 있어 routing을 혼란스럽게 하기 때문입니다.
`freedom`에서는 `arc0`에 `freedom` 주소를 설정하고 자신과 `insight` route를 추가합니다. Default gateway는 `pppd`가 구성합니다. 이제 `insight`는 `arc0`로 `freedom`과 직접 통신하고 Internet packet은 `freedom`을 통해 보냅니다.
2. More than one protocol on the same wire.
Now things start getting confusing. To even try it, you may need to be
partly crazy. Here's what *I* did. :) Note that I don't include arc0s in
my home network; I don't have any NetBSD or AmiTCP computers, so I only
use arc0s during limited testing.
I have three computers on my home network; two Linux boxes (which prefer
RFC1201 protocol, for reasons listed above), and one XT that can't run
Linux but runs the free Microsoft LANMAN Client instead.
Worse, one of the Linux computers (freedom) also has a modem and acts as
a router to my Internet provider. The other Linux box (insight) also has
its own IP address and needs to use freedom as its default gateway. The
XT (patience), however, does not have its own Internet IP address and so
I assigned it one on a "private subnet" (as defined by RFC1597).
To start with, take a simple network with just insight and freedom.
Insight needs to:
- talk to freedom via RFC1201 (arc0) protocol, because I like it
more and it's faster.
- use freedom as its Internet gateway.
That's pretty easy to do. Set up insight like this::
ifconfig arc0 insight
route add insight arc0
route add freedom arc0 /* I would use the subnet here (like I said
to in "single protocol" above),
but the rest of the subnet
unfortunately lies across the PPP
link on freedom, which confuses
things. */
route add default gw freedom
And freedom gets configured like so::
ifconfig arc0 freedom
route add freedom arc0
route add insight arc0
/* and default gateway is configured by pppd */
Great, now insight talks to freedom directly on arc0, and sends packets
to the Internet through freedom. If you didn't know how to do the above,
you should probably stop reading this section now because it only gets
worse.
Gatekeeper alias와 protocol 간 routing
456-503이제 LANMAN Client를 쓰는 `patience`를 추가하려면 `arc0e`가 필요합니다. `patience`는 `insight`, `freedom` 양쪽과 통신하고 `freedom`을 Internet gateway로 써야 합니다. Private address는 Internet에서 직접 작동하지 않지만 `freedom`에 이 subnet을 위한 Linux IP masquerading을 설정했습니다.
Provider에게 추가 IP address를 받지 못했으므로 `patience`는 `freedom`·`insight`와 다른 subnet에 있으면서도 `freedom`을 gateway로 사용해야 합니다. LANMAN을 포함한 많은 DOS networking program은 netmask와 default gateway만으로 route를 판단하므로 같은 physical wire에서 직접 통신할 수 있어도 `freedom`이나 `insight`로 가는 packet을 default gateway로 보냅니다.
이를 보완하려고 `freedom`에 private subnet 소속의 추가 alias IP address `gatekeeper`를 줍니다. `patience`의 default gateway는 `gatekeeper`로 설정합니다.
기존 `freedom` 명령에 더해 `ifconfig arc0e gatekeeper`, `route add gatekeeper arc0e`, `route add patience arc0e`를 실행합니다.
이제 `freedom`은 `patience`용 packet을 모두 `arc0e`로 보내고 private subnet의 `gatekeeper` 주소를 source로 사용합니다. `insight`나 Internet과 통신할 때는 Internet IP인 `freedom` 주소를 사용합니다.
`insight`에는 `arc0e`를 구성하지 않았습니다. 구성해도 되지만 private subnet의 별도 IP number가 더 필요합니다. `insight`와 `patience`가 모두 `freedom`을 default gateway로 사용하므로 이미 서로 통신할 수 있습니다.
이 구성은 `insight`를 DOS로 boot해 RFC1201만 지원하는 Novell ODI stack을 사용할 때도 유용합니다. Novell stack은 Microsoft Ethernet Encapsulation과 호환되지 않아 `patience`와 직접 통신할 수 없지만, `insight`의 default gateway를 `freedom`으로 정하면 `freedom`·`patience` 설정을 바꾸지 않고도 두 host 사이를 자동 forwarding합니다.
Now, how do I add patience into the network? It will be using LANMAN
Client, which means I need the arc0e device. It needs to be able to talk
to both insight and freedom, and also use freedom as a gateway to the
Internet. (Recall that patience has a "private IP address" which won't
work on the Internet; that's okay, I configured Linux IP masquerading on
freedom for this subnet).
So patience (necessarily; I don't have another IP number from my
provider) has an IP address on a different subnet than freedom and
insight, but needs to use freedom as an Internet gateway. Worse, most
DOS networking programs, including LANMAN, have braindead networking
schemes that rely completely on the netmask and a 'default gateway' to
determine how to route packets. This means that to get to freedom or
insight, patience WILL send through its default gateway, regardless of
the fact that both freedom and insight (courtesy of the arc0e device)
could understand a direct transmission.
I compensate by giving freedom an extra IP address - aliased 'gatekeeper' -
that is on my private subnet, the same subnet that patience is on. I
then define gatekeeper to be the default gateway for patience.
To configure freedom (in addition to the commands above)::
ifconfig arc0e gatekeeper
route add gatekeeper arc0e
route add patience arc0e
This way, freedom will send all packets for patience through arc0e,
giving its IP address as gatekeeper (on the private subnet). When it
talks to insight or the Internet, it will use its "freedom" Internet IP
address.
You will notice that we haven't configured the arc0e device on insight.
This would work, but is not really necessary, and would require me to
assign insight another special IP number from my private subnet. Since
both insight and patience are using freedom as their default gateway, the
two can already talk to each other.
It's quite fortunate that I set things up like this the first time (cough
cough) because it's really handy when I boot insight into DOS. There, it
runs the Novell ODI protocol stack, which only works with RFC1201 ARCnet.
In this mode it would be impossible for insight to communicate directly
with patience, since the Novell stack is incompatible with Microsoft's
Ethernet-Encap. Without changing any settings on freedom or patience, I
simply set freedom as the default gateway for insight (now in DOS,
remember) and all the forwarding happens "automagically" between the two
hosts that would normally not be able to communicate at all.
하나의 물리 wire 위 두 virtual subnet
504-522도식으로 보면 같은 physical ARCnet wire 위에 두 virtual subnet을 만든 것입니다. 왼쪽은 등록된 Internet subnet의 RFC1201 network이고 오른쪽은 RFC1597 private subnet의 Ethernet Encapsulation network입니다.
`freedom`은 왼쪽 RFC1201 쪽 주소, `gatekeeper`는 오른쪽 private 쪽 alias입니다. 두 interface는 같은 router system 안에서 IP masquerade 경계를 사이에 두고 연결됩니다. `insight`는 왼쪽 subnet, `patience`는 오른쪽 subnet에 붙고, `freedom`의 별도 link가 Internet으로 이어집니다.
For those who like diagrams, I have created two "virtual subnets" on the
same physical ARCnet wire. You can picture it like this::
[RFC1201 NETWORK] [ETHER-ENCAP NETWORK]
(registered Internet subnet) (RFC1597 private subnet)
(IP Masquerade)
/---------------\ * /---------------\
| | * | |
| +-Freedom-*-Gatekeeper-+ |
| | | * | |
\-------+-------/ | * \-------+-------/
| | |
Insight | Patience
(Internet)
작동할 때 보고할 내용
523-537작동한다면 다음에는 무엇을 하는가?
Driver version, kernel version, ARCnet card model, CPU type, network의 system 수, 사용 software 목록을 포함해 설정을 설명하는 mail을 `[email protected]`로 보내 달라고 요청합니다.
작성자는 받은 모든 message에 때로 자동 응답을 포함해 답장을 보냈습니다. E-mail 경로가 이상하고 여러 곳으로 forwarding되는 경우가 많으므로 합리적인 시간 안에 답이 없으면 다시 보내 달라고 합니다.
It works: what now?
-------------------
Send mail describing your setup, preferably including driver version, kernel
version, ARCnet card model, CPU type, number of systems on your network, and
list of software in use to me at the following address:
[email protected]
I do send (sometimes automated) replies to all messages I receive. My email
can be weird (and also usually gets forwarded all over the place along the
way to me), so if you don't get a reply within a reasonable time, please
resend.
작동하지 않을 때와 buffer 진단
538-573작동하지 않는다면 다음에는 무엇을 하는가?
성공 보고와 같은 정보에 더해 `ifconfig`, `route` 출력과 마지막 reboot 이후 나타난 `arcnet:`으로 시작하는 관련 log entry를 mail에 포함합니다.
직접 고치려 해도 먼저 mail을 보내 이미 해결된 문제인지 확인하는 것이 좋습니다. 여러 debug level을 시험할 수 있으며 `D_DURING` 이상으로 강한 test를 할 때는 `klogd` daemon을 먼저 종료하는 편이 좋습니다. `D_DURING`은 송수신 packet마다 4~5줄을 표시하고, `D_TX`, `D_RX`, `D_SKB`는 실제 packet 전체를 출력해 양이 매우 큽니다.
2.40 ALPHA부터 autoprobe routine이 크게 바뀌었습니다. 특히 `D_INIT_REASONS` debug flag를 켜지 않으면 card를 찾지 못한 이유를 알려 주지 않습니다.
Driver가 실행 중이면 root로 `arcdump` shell script를 실행해 언제든 ARCnet buffer 내용을 볼 수 있습니다. 해석하려면 `arcnet.c` 상단 부근에 열거된 관련 RFC가 필요합니다. `arcdump`는 card가 `0xD0000`에 있다고 가정하므로 다른 address라면 script를 수정합니다.
Buffer 0과 1은 수신, buffer 2와 3은 송신에 쓰며 양쪽 모두 ping-pong buffer를 구현합니다.
Debug level에 `D_DURING`이 포함되고 `SLOW_XMIT_COPY`를 define하지 않았다면 card reset 때마다 buffer를 상수 `0x42`로 지웁니다. Reset은 보통 `ifconfig up` 또는 Linux가 driver 고장으로 판단할 때만 일어납니다. 송신 중에는 buffer의 미사용 부분도 `0x42`로 지워 packet이 실제 사용한 byte를 쉽게 식별하게 합니다.
It doesn't work: what now?
--------------------------
Do the same as above, but also include the output of the ifconfig and route
commands, as well as any pertinent log entries (ie. anything that starts
with "arcnet:" and has shown up since the last reboot) in your mail.
If you want to try fixing it yourself (I strongly recommend that you mail me
about the problem first, since it might already have been solved) you may
want to try some of the debug levels available. For heavy testing on
D_DURING or more, it would be a REALLY good idea to kill your klogd daemon
first! D_DURING displays 4-5 lines for each packet sent or received. D_TX,
D_RX, and D_SKB actually DISPLAY each packet as it is sent or received,
which is obviously quite big.
Starting with v2.40 ALPHA, the autoprobe routines have changed
significantly. In particular, they won't tell you why the card was not
found unless you turn on the D_INIT_REASONS debugging flag.
Once the driver is running, you can run the arcdump shell script (available
from me or in the full ARCnet package, if you have it) as root to list the
contents of the arcnet buffers at any time. To make any sense at all out of
this, you should grab the pertinent RFCs. (some are listed near the top of
arcnet.c). arcdump assumes your card is at 0xD0000. If it isn't, edit the
script.
Buffers 0 and 1 are used for receiving, and Buffers 2 and 3 are for sending.
Ping-pong buffers are implemented both ways.
If your debug level includes D_DURING and you did NOT define SLOW_XMIT_COPY,
the buffers are cleared to a constant value of 0x42 every time the card is
reset (which should only happen when you do an ifconfig up, or when Linux
decides that the driver is broken). During a transmit, unused parts of the
buffer will be cleared to 0x42 as well. This is to make it easier to figure
out which bytes are being used by a packet.
Runtime debug level 변경
574-590Kernel을 다시 compile하지 않고 debug level을 바꾸려면 `ifconfig arc0 down metric 1xxx`를 실행한 뒤 `/etc/rc.d/rc.inet1`을 다시 실행합니다. 여기서 `xxx`가 원하는 debug level입니다.
예를 들어 `metric 1015`는 debug level 15를 설정합니다. 당시 기본값은 7이었습니다.
1.90 ALPHA부터 debug level은 debug flag의 binary 조합입니다. 따라서 level 7은 `1+2+4`, 즉 `D_NORMAL+D_EXTRA+D_INIT`입니다. `D_DURING`을 포함하려면 16을 더해 level 23으로 만듭니다.
이 설명이 이해되지 않는다면 굳이 더 알 필요가 없을 수 있으니 문제를 e-mail로 보내 달라고 마무리합니다.
You can change the debug level without recompiling the kernel by typing::
ifconfig arc0 down metric 1xxx
/etc/rc.d/rc.inet1
where "xxx" is the debug level you want. For example, "metric 1015" would put
you at debug level 15. Debug level 7 is currently the default.
Note that the debug level is (starting with v1.90 ALPHA) a binary
combination of different debug flags; so debug level 7 is really 1+2+4 or
D_NORMAL+D_EXTRA+D_INIT. To include D_DURING, you would add 16 to this,
resulting in debug level 23.
If you don't understand that, you probably don't want to know anyway.
E-mail me about your problem.
돈을 보내고 싶을 때
591-594돈을 보내고 싶다면 다음에는 무엇을 하는가?
낮잠이나 자고 나면 아침에는 기분이 나아질 것이라는 농담으로 문서를 끝냅니다.
I want to send money: what now?
-------------------------------
Go take a nap or something. You'll feel better in the morning.
요약·해설
arcnet.rst:1-594이 문서는 초기 Linux ARCnet driver의 설치와 운용 기록입니다. 현재 build 명령과 mailing-list 주소는 역사적이지만, COM90xx·COM20020 차이, `arc0`·`arc0e`·`arc0s`의 encapsulation 관계, 같은 wire에서 여러 protocol을 route하는 원리는 분명하게 보존할 가치가 있습니다.
네 hardware mode와 필수 설정을 비교합니다.
서로 호환되지 않는 ARCnet encapsulation을 별도 interface로 노출합니다.
Generic layer 뒤에 필요한 chipset module instance를 올립니다.
원문 L508-519 ASCII diagram의 node와 forwarding 관계를 재구성했습니다.
외부 stack의 encapsulation에 맞춰 Linux virtual device를 고릅니다.
진단 시 buffer 용도와 출력량을 함께 확인합니다.