Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sage/ceph...
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / Documentation / networking / 6pack.txt
CommitLineData
1da177e4
LT
1This is the 6pack-mini-HOWTO, written by
2
be2a608b 3Andreas Könsgen DG3KQ
67332595 4Internet: ajk@comnets.uni-bremen.de
1da177e4
LT
5AMPR-net: dg3kq@db0pra.ampr.org
6AX.25: dg3kq@db0ach.#nrw.deu.eu
7
8Last update: April 7, 1998
9
101. What is 6pack, and what are the advantages to KISS?
11
126pack is a transmission protocol for data exchange between the PC and
13the TNC over a serial line. It can be used as an alternative to KISS.
14
156pack has two major advantages:
16- The PC is given full control over the radio
17 channel. Special control data is exchanged between the PC and the TNC so
18 that the PC knows at any time if the TNC is receiving data, if a TNC
19 buffer underrun or overrun has occurred, if the PTT is
20 set and so on. This control data is processed at a higher priority than
21 normal data, so a data stream can be interrupted at any time to issue an
22 important event. This helps to improve the channel access and timing
23 algorithms as everything is computed in the PC. It would even be possible
24 to experiment with something completely different from the known CSMA and
25 DAMA channel access methods.
26 This kind of real-time control is especially important to supply several
27 TNCs that are connected between each other and the PC by a daisy chain
28 (however, this feature is not supported yet by the Linux 6pack driver).
29
30- Each packet transferred over the serial line is supplied with a checksum,
31 so it is easy to detect errors due to problems on the serial line.
32 Received packets that are corrupt are not passed on to the AX.25 layer.
33 Damaged packets that the TNC has received from the PC are not transmitted.
34
35More details about 6pack are described in the file 6pack.ps that is located
36in the doc directory of the AX.25 utilities package.
37
382. Who has developed the 6pack protocol?
39
40The 6pack protocol has been developed by Ekki Plicht DF4OR, Henning Rech
41DF9IC and Gunter Jost DK7WJ. A driver for 6pack, written by Gunter Jost and
42Matthias Welwarsky DG2FEF, comes along with the PC version of FlexNet.
43They have also written a firmware for TNCs to perform the 6pack
44protocol (see section 4 below).
45
463. Where can I get the latest version of 6pack for LinuX?
47
48At the moment, the 6pack stuff can obtained via anonymous ftp from
49db0bm.automation.fh-aachen.de. In the directory /incoming/dg3kq,
50there is a file named 6pack.tgz.
51
524. Preparing the TNC for 6pack operation
53
54To be able to use 6pack, a special firmware for the TNC is needed. The EPROM
55of a newly bought TNC does not contain 6pack, so you will have to
56program an EPROM yourself. The image file for 6pack EPROMs should be
57available on any packet radio box where PC/FlexNet can be found. The name of
58the file is 6pack.bin. This file is copyrighted and maintained by the FlexNet
59team. It can be used under the terms of the license that comes along
60with PC/FlexNet. Please do not ask me about the internals of this file as I
61don't know anything about it. I used a textual description of the 6pack
62protocol to program the Linux driver.
63
64TNCs contain a 64kByte EPROM, the lower half of which is used for
65the firmware/KISS. The upper half is either empty or is sometimes
66programmed with software called TAPR. In the latter case, the TNC
67is supplied with a DIP switch so you can easily change between the
68two systems. When programming a new EPROM, one of the systems is replaced
69by 6pack. It is useful to replace TAPR, as this software is rarely used
70nowadays. If your TNC is not equipped with the switch mentioned above, you
71can build in one yourself that switches over the highest address pin
72of the EPROM between HIGH and LOW level. After having inserted the new EPROM
73and switched to 6pack, apply power to the TNC for a first test. The connect
74and the status LED are lit for about a second if the firmware initialises
75the TNC correctly.
76
775. Building and installing the 6pack driver
78
79The driver has been tested with kernel version 2.1.90. Use with older
80kernels may lead to a compilation error because the interface to a kernel
81function has been changed in the 2.1.8x kernels.
82
83How to turn on 6pack support:
84
85- In the linux kernel configuration program, select the code maturity level
86 options menu and turn on the prompting for development drivers.
87
88- Select the amateur radio support menu and turn on the serial port 6pack
89 driver.
90
91- Compile and install the kernel and the modules.
92
93To use the driver, the kissattach program delivered with the AX.25 utilities
94has to be modified.
95
96- Do a cd to the directory that holds the kissattach sources. Edit the
97 kissattach.c file. At the top, insert the following lines:
98
99 #ifndef N_6PACK
100 #define N_6PACK (N_AX25+1)
101 #endif
102
103 Then find the line
104
105 int disc = N_AX25;
106
107 and replace N_AX25 by N_6PACK.
108
109- Recompile kissattach. Rename it to spattach to avoid confusions.
110
111Installing the driver:
112
113- Do an insmod 6pack. Look at your /var/log/messages file to check if the
114 module has printed its initialization message.
115
116- Do a spattach as you would launch kissattach when starting a KISS port.
117 Check if the kernel prints the message '6pack: TNC found'.
118
119- From here, everything should work as if you were setting up a KISS port.
120 The only difference is that the network device that represents
121 the 6pack port is called sp instead of sl or ax. So, sp0 would be the
122 first 6pack port.
123
124Although the driver has been tested on various platforms, I still declare it
125ALPHA. BE CAREFUL! Sync your disks before insmoding the 6pack module
126and spattaching. Watch out if your computer behaves strangely. Read section
1276 of this file about known problems.
128
129Note that the connect and status LEDs of the TNC are controlled in a
130different way than they are when the TNC is used with PC/FlexNet. When using
131FlexNet, the connect LED is on if there is a connection; the status LED is
132on if there is data in the buffer of the PC's AX.25 engine that has to be
133transmitted. Under Linux, the 6pack layer is beyond the AX.25 layer,
134so the 6pack driver doesn't know anything about connects or data that
135has not yet been transmitted. Therefore the LEDs are controlled
136as they are in KISS mode: The connect LED is turned on if data is transferred
137from the PC to the TNC over the serial line, the status LED if data is
138sent to the PC.
139
1406. Known problems
141
142When testing the driver with 2.0.3x kernels and
143operating with data rates on the radio channel of 9600 Baud or higher,
144the driver may, on certain systems, sometimes print the message '6pack:
145bad checksum', which is due to data loss if the other station sends two
146or more subsequent packets. I have been told that this is due to a problem
147with the serial driver of 2.0.3x kernels. I don't know yet if the problem
148still exists with 2.1.x kernels, as I have heard that the serial driver
149code has been changed with 2.1.x.
150
151When shutting down the sp interface with ifconfig, the kernel crashes if
152there is still an AX.25 connection left over which an IP connection was
153running, even if that IP connection is already closed. The problem does not
154occur when there is a bare AX.25 connection still running. I don't know if
155this is a problem of the 6pack driver or something else in the kernel.
156
157The driver has been tested as a module, not yet as a kernel-builtin driver.
158
159The 6pack protocol supports daisy-chaining of TNCs in a token ring, which is
160connected to one serial port of the PC. This feature is not implemented
161and at least at the moment I won't be able to do it because I do not have
162the opportunity to build a TNC daisy-chain and test it.
163
164Some of the comments in the source code are inaccurate. They are left from
165the SLIP/KISS driver, from which the 6pack driver has been derived.
166I haven't modified or removed them yet -- sorry! The code itself needs
167some cleaning and optimizing. This will be done in a later release.
168
169If you encounter a bug or if you have a question or suggestion concerning the
170driver, feel free to mail me, using the addresses given at the beginning of
171this file.
172
173Have fun!
174
175Andreas