!! Version 2.4
!!
!! Deutscher Text siehe scc_ger.doc
!!
!! BTW: REAL programmers don't document...
!!


	 SCC.C - Linux driver for Z8530 based HDLC cards for AX.25      

   ********************************************************************

        Copyright 1993-1996 by Joerg Reuter DL1BKE

        portions (c) 1993 Guido ten Dolle PE1NNZ

        for the complete copyright notice see >> Copying.Z8530DRV <<

   ******************************************************************** 


0. Installation of the package
==============================

To compile for a module, type

	make module

for a kernel version, type

	make for_kernel

Note that this will overwrite linux/drivers/char/scc.c and
linux/include/linux/scc.h.

In either case the Makefile will create the utilities and tries
to install them.

Note: make will overwrite any existent z8530drv.conf file in /etc,
      so make sure you have rescued it before!

sccinit  - reads /etc/z8530drv.conf and initializes the driver
sccstat  - shows the status of a channel
sccparam - sets KISS parameters for a channel

To rebuild just the utilities, simply type 'make'


1. Initialization of the driver
===============================

To use the driver, 3 steps must be performed:

     1. loading the module
     2. Setup of hardware, MODEM and KISS parameters with sccinit
     3. Attachment of each channel in the packet software


1.1 Loading the module
======================

Before you can use a module, you'll have to load it with

	insmod scc.o

please read 'man insmod' that comes with modutils.

You should include the insmod in one of the /etc/rc.d/rc.* files,
and don't forget to insert a call of sccinit after that. It
will read  your


1.2. /etc/z8530drv.conf
=====================

To setup all parameters you must run /sbin/sccinit from one
of your rc.*-files. This has to be done BEFORE the start of
NET or axattach. Sccinit reads the file /etc/z8530drv.conf
and sets the hardware, MODEM and KISS parameters. A sample file is
delivered with this package. Change it to your needs.

The file itself consists of two main sections.

1.2.1 configuration of hardware parameters
==========================================

The hardware setup section defines the following parameters for each
Z8530:

chip    1
data_a  0x300                   # data port A
ctrl_a  0x304                   # control port A
data_b  0x301                   # data port B
ctrl_b  0x305                   # control port B
irq     5                       # IRQ No. 5
pclock  4915200                 # clock
board   BAYCOM                  # hardware type
escc    no                      # enhanced SCC chip? (8580/85180/85280)
vector  0                       # latch for interrupt vector
special no                      # address of special function register
option  0                       # option to set via sfr


chip	- this is just a delimiter to make sccinit a bit simplier to
	  program. A parameter has no effect.

data_a  - the address of the data port A of this Z8530 (needed)
ctrl_a  - the address of the control port A (needed)
data_b  - the address of the data port B (needed)
ctrl_b  - the address of the control port B (needed)

irq     - the used IRQ for this chip. Different chips can use different
          IRQs or the same. If they share an interrupt, it needs to be
	  specified within one chip-definition only.

pclock  - the clock at the PCLK pin of the Z8530 (option, 4915200 is
          default), measured in Hertz

board   - the "type" of the board:

	   SCC type                 value
	   ---------------------------------
	   PA0HZP SCC card          PA0HZP
	   EAGLE card               EAGLE
	   PC100 card               PC100
	   PRIMUS-PC (DG9BL) card   PRIMUS
	   BayCom (U)SCC card       BAYCOM

escc    - if you want support for ESCC chips (8580, 85180, 85280), set
          this to "yes" (option, defaults to "no")

vector  - address of the vector latch (aka "intack port") for PA0HZP
          cards. There can be only one vector latch for all chips!
	  (option, defaults to 0)

special - address of the special function register on several cards.
          (option, defaults to 0)

option  - The value you write into that register (option, default is 0)

You can specify up to four chips (8 channels). If this is not enough,
just change

	#define MAXSCC 4

to a higher value.

Example for the BayCom USCC:
----------------------------

chip    1
data_a  0x300                   # data port A
ctrl_a  0x304                   # control port A
data_b  0x301                   # data port B
ctrl_b  0x305                   # control port B
irq     5                       # IRQ No. 5 (#)
board   BAYCOM                  # hardware type (*)
#
# SCC chip 2
#
chip    2
data_a  0x302
ctrl_a  0x306
data_b  0x303
ctrl_b  0x307
board   BAYCOM

An example for a PA0HZP card:
-----------------------------

chip 1
data_a 0x153
ctrl_a 0x152
data_b 0x151
ctrl_b 0x150
irq 9
pclock 4915200
board PA0HZP
vector 0x168
escc no
#
#
#
chip 2
data_a 0x157
ctrl_a 0x156
data_b 0x155
ctrl_b 0x154
irq 9
pclock 4915200
board PA0HZP
vector 0x168
escc no

The PE1PET card claims to be compatible to the PA0HZP board but
needs SCC_DELAY #define'd in scc.c.

A DRSI would should probably work with this:
--------------------------------------------
(actually: two DRSI cards...)

chip 1
data_a 0x303
ctrl_a 0x302
data_b 0x301
ctrl_b 0x300
irq 7
pclock 4915200
board DRSI
escc no
#
#
#
chip 2
data_a 0x313
ctrl_a 0x312
data_b 0x311
ctrl_b 0x310
irq 7
pclock 4915200
board DRSI
escc no

Note that you cannot use the on-board baudrate generator off DRSI
cards. Use "mode dpll" for clock source (see below).

This is based on information provided by Mike Bilow and verified
by Paul Healy...


The utility "gencfg"
--------------------

If you only know the parameters for the PE1CHL driver for DOS,
run gencfg. It will generate the correct port addresses (I hope).
Its parameters are exactly the same as the ones you use with
the "attach scc" command in net, except that the string "init" must 
not appear. Example:

gencfg 2 0x150 4 2 0 1 0x168 9 4915200 

will print a skeleton z8530drv.conf for the OptoSCC to stdout.

gencfg 2 0x300 2 4 5 -4 0 7 4915200 0x10

does the same for the BayCom USCC card. I my opinion it is much easier
to edit scc_config.h... 


1.2.2 channel configuration
===========================

The channel definition is divided into three sub sections for each
channel:

An example for /dev/scc0:

# DEVICE

device /dev/scc0	# the device for the following params

# MODEM / BUFFERS

speed 1200		# the default baudrate
clock dpll		# clock source: 
			# 	dpll     = normal halfduplex operation
			# 	external = MODEM provides own Rx/Tx clock
			#	divider  = use fullduplex divider if
			#		   installed (1)
mode nrzi		# HDLC encoding mode
			#	nrzi = 1k2 MODEM, G3RUH 9k6 MODEM
			#	nrz  = DF9IC 9k6 MODEM
			#
rxbuffers 8		# number of rx buffers allocated
			# 		(option, default is 4)
txbuffers 16		# number of tx buffers allocatd
			#		(option, default is 16)
bufsize	384		# size of buffers. Note that this must include
			# the AX.25 header, not only the data field!
			# (optional, defaults to 384)

# Uwaga !!!! mtu 256 buffsize 384 i rxb i txb = 12

# Hardware access parameters (Layer 1), aka 'KISS commands'

txdelay 36              # (see chapter 1.4)
persist 64
slot    8
tail    8
fulldup 0
wait    12
min     3
maxkey  7
idle    3
maxdef  120
group   0
txoff   off
softdcd on                   

The order WITHIN these sections is unimportant. The order OF these
sections IS important. The MODEM parameters are set with the first
recognized KISS paramer...

Please note that you can initialize the board only once after boot. 
You can change all paramters but "mode" and "clock" later with the
Sccparam program or through KISS. Just to avoid securety holes... 

(1) this divider is usually mounted on the SCC-PBC (PA0HZP) or not
    present at all (BayCom). It feeds back the output of the DPLL 
    (digital pll) as transmit clock. Using this mode without a divider 
    installed will normally result in keying the transceiver until 
    maxkey expires --- of course without sending anything (useful).


2. Attachment of a channel by your AX.25 software
=================================================

2.1 KA9Q NOS derivates
======================

When the linux has startup, the SCC driver has been initialized,
you can attach the channels in your packet software. This is done
by open the scc devices by using the attach asy command.
The SCC-drivers emulates the scc devices as serial asy ports,
this means e.g. that the baudrate can be set in the attach command.


Example Wampes:

#############################################################################################
# Wampes device attach
# NOTE: Interfacename and the device must be the same!!
# Usage: attach asy 0 0 slip|vjslip|ax25ui|ax25i|nrs|kissui <label> 0 <mtu> <speed> [ip_addr]
#
attach asy 0 0 kissi  scc0 256 256 1200   # Attach SCC channel 1 in 1200 baud
attach asy 0 0 kissi  scc1 256 256 1200   # Attach SCC channel 2 in 1200 baud
attach asy 0 0 kissui scc2 256 256 38400  # Attach SCC channel 3 in 38400 baud
attach asy 0 0 kissui scc3 256 256 9600   # Attach SCC channel 4 in 9600 baud
#              ^^^^
#              for WAMPES 921229 use here: ax25
#

Example JNOS:

############################################
# JNOS device attach
#
attach asy scc0 0 ax25 scc0 256 256 1200
attach asy scc1 0 ax25 scc1 256 256 1200
attach asy scc2 0 ax25 scc2 256 256 300
attach asy scc3 0 ax25 scc3 256 256 4800
#
#


It allows AX.25 communication without a TNC.  Only a MODEM is
needed. The parameters have the same meaning as in KISS mode.
In fact, the AX.25 mode is emulating an extended KISS TNC, so
the same commands can be used to set the parameters of the
interface (see below).

To be able to run fullduplex using an SCC in AX.25 mode, an 
external divider must be available, that divides the baudrate 
generator clock available on the TRxC pin by 32, and puts the 
resulting signal on the RTxC pint of the same channel of the SCC.  
Such a divider is not necessary for normal CSMA packet radio 
operation, but interrupt overhead is slightly reduced if you 
still install it.  

2.2 Kernel AX.25
================

The driver may act as a network device as well. This gives you
the improvement of leaving SLIP and TTY drivers behind and 
hopefully will increase performance.

To attach a SCC channel to the kernel AX.25 simply run

	ifconfig scc0 44.130.20.106 hw ax25 dl1bke-10

(please change the IP address and the callsign to your own)

If you want to use the driver as a TTY driver and link it to
the kernel AX.25 with axattach note that you need to compile
the SLIP driver into your kernel.

3. Adjustment and Display of parameters
=======================================

3.1 Displaying SCC Parameters:
==============================

Once a SCC channel has been attached, the parameter settings and 
some statistic information can be shown using the param program:

dl1bke-u:~$ sccstat /dev/scc0

Parameters:

speed       : 1200 baud
txdelay     : 36
persist     : 255
slottime    : 0
txtail      : 8
fulldup     : 1
waittime    : 12
mintime     : 3 sec
maxkeyup    : 7 sec
idletime    : 3 sec
maxdefer    : 120 sec
group       : 0x00
txoff       : off
softdcd     : on

Status:

HDLC                  Z8530           Interrupts         Queues
-----------------------------------------------------------------------
Sent       :     273  RxOver :     0  RxInts :   125074  RxQueue :    0
Received   :    1095  TxUnder:     0  TxInts :     4684  TxQueue :    0
RxErrors   :    1591                  ExInts :    11776  NoSpace :    0
KissErrors :       0                  SpInts :     1503
Tx State   :    idle

Memory allocated:

Buffer size:     384
rx buffers :       4
tx buffers :       8


The status info shown is:

Sent		- number of frames transmitted
Received	- number of frames received
RxErrors	- number of receive errors (CRC, ABORT)
KissErrors	- number of KISS errors (should be zero...)
Tx State	- status of the Tx interrupt handler: idle/busy/active/tail (2)
RxOver		- number of receiver overruns
TxUnder		- number of transmitter underruns     
RxInts		- number of receiver interrupts
TxInts		- number of transmitter interrupts
EpInts		- number of receiver special condition interrupts
SpInts		- number of external/status interrupts
RxQueue		- number of received packets enqueued for this channel
TxQueue		- number of packets enqueued for Tx
NoSpace		- number of times the receiver buffer pool was found empty


An overrun is abnormal. If lots of these occur, the product of
baudrate and number of interfaces is too high for the processing
power of you computer. If "Space" errors occur, specify a higher
number of buffers in the "scc.h" file.


3.2 Setting Parameters
======================


The setting of parameters of the emulated KISS TNC is done in the 
same way in the SCC driver. You can change parameters by using
the command param in NET or NOS

     param <iface> <paramname> <value>

or use the program "sccparam":

     sccparam <device> <paramname> <decimal-|hexadecimal value>

You can change the following parameters:

param	    : value
------------------------
speed       : 1200
txdelay     : 36
persist     : 255
slottime    : 0
txtail      : 8
fulldup     : 1
waittime    : 12
mintime     : 3
maxkeyup    : 7
idletime    : 3
maxdefer    : 120
group       : 0x00
txoff       : off
softdcd     : on


The parameters have the following meaning:

speed:
     The baudrate on this channel in bits/sec

     Example: sccparam scc3 speed 9600

txdelay:
     The delay (in units of 10ms) after keying of the 
     transmitter, until the first byte is sent. This is usually 
     called "TXDELAY" in a TNC.  When 0 is specified, the driver 
     will just wait until the CTS signal is asserted. This 
     assumes the presence of a timer or other circuitry in the 
     MODEM and/or transmitter, that asserts CTS when the 
     transmitter is ready for data.
     A normal value of this parameter is 30-36.

     Example: sccparam scc0 txd 20

persist:
     This is the probability that the transmitter will be keyed 
     when the channel is found to be free.  It is a value from 0 
     to 255, and the probability is (value+1)/256.  The value 
     should be somewhere near 50-60, and should be lowered when 
     the channel is used more heavily.

     Example: sccparam scc2 persist 20

slottime:
     This is the time between samples of the channel. It is 
     expressed in units of 10ms.  About 200-300 ms (value 20-30) 
     seems to be a good value.

     Example: sccparam scc0 slot 20

tail:
     The time the transmitter will remain keyed after the last 
     byte of a packet has been transferred to the SCC. This is 
     necessary because the CRC and a flag still have to leave the 
     SCC before the transmitter is keyed down. The value depends 
     on the baudrate selected.  A few character times should be 
     sufficient, e.g. 40ms at 1200 baud. (value 4)
     The value of this parameter is in 10ms units.

     Example: sccparam scc2 4

full:
     The full-duplex mode switch. This can be one of the folowing 
     values:

     0:   The interface will operate in CSMA mode (the normal 
          half-duplex packet radio operation)
     1:   Fullduplex mode, i.e. the transmitter will be keyed at 
          any time, without checking the received carrier.  It 
          will be unkeyed when there are no packets to be sent.
     2:   Like 1, but the transmitter will remain keyed, also 
          when there are no packets to be sent.  Flags will be 
          sent in that case, until a timeout (parameter 10) 
          occurs.
     3:   Like 2, but the transmitter remains keyed until
	  the network layer resets RTS. The driver will send
	  additional KISS commands to the network layer regarding
	  DCD status and when the TX queue is empty. See chapter
	  5 for details.

     Example: sccparam scc0 fulldup off

wait:
     The initial waittime before any transmit attempt, after the 
     frame has been queue for transmit.  This is the length of 
     the first slot in CSMA mode.  In fullduplex modes it is
     set to 0 for maximum performance.
     The value of this parameter is in 10ms units. 

     Example: sccparam scc1 wait 4

maxkey:
     The maximal time the transmitter will be keyed to send 
     packets, in seconds.  This can be useful on busy CSMA 
     channels, to avoid "getting a bad reputation" when you are 
     generating a lot of traffic.  After the specified time has 
     elapsed, no new frame will be started. Instead, the trans-
     mitter will be switched off for a specified time (parameter 
     min), and then the selected algorithm for keyup will be 
     started again.
     The value 0 as well as "off" will disable this feature, 
     and allow infinite transmission time. 

     Example: sccparam scc0 maxk 20

min:
     This is the time the transmitter will be switched off when 
     the maximum transmission time is exceeded.

     Example: sccparam scc3 min 10

idle
     This parameter specifies the maximum idle time in fullduplex 
     2 mode, in seconds.  When no frames have been sent for this 
     time, the transmitter will be keyed down.  A value of 0 is
     has same result as the fullduplex mode 1. This parameter
     can be disabled.

     Example: sccparam scc2 idle off	# transmit forever

maxdefer
     This is the maximum time (in seconds) to wait for a free channel
     to send. When this timer expires the transmitter will be keyed 
     IMMEDIATLY. If you love to get trouble with other users you
     should set this to a very low value ;-)

     Example: sccparam scc0 maxdefer 240	# 2 minutes


txoff:
     When this parameter has the value 0, the transmission of packets
     is enable. Otherwise it is disabled.

     Example: sccparam scc2 txoff on

group:
     It is possible to build special radio equipment to use more than 
     one frequency on the same bad, e.g. using several receivers and 
     only one transmitter that can be switched between frequencies.
     Also, you can connect several radios that are active on the same 
     band.  In these cases, it is not possible, or not a good idea, to 
     transmit on more than one frequency.  The SCC driver provides a 
     method to lock transmitters on different interfaces, using the 
     "param <interface> group <x>" command.  This will only work when 
     you are using CSMA mode (parameter full = 0).
     The number <x> must be 0 if you want no group restrictions, and 
     can be computed as follows to create restricted groups:
     <x> is the sum of some OCTAL numbers:

     200  This transmitter will only be keyed when all other 
          transmitters in the group are off.
     100  This transmitter will only be keyed when the carrier 
          detect of all other interfaces in the group is off.
     0xx  A byte that can be used to define different groups.  
          Interfaces are in the same group, when the logical AND 
          between their xx values is nonzero.

     Examples:
     When 2 interfaces use group 201, their transmitters will never be 
     keyed at the same time.
     When 2 interfaces use group 101, the transmitters will only key 
     when both channels are clear at the same time.  When group 301, 
     the transmitters will not be keyed at the same time.

     Don't forget to convert the octal numbers into decimal before
     you set the parameter.

     Example: (to be written)

softdcd:
     use a software dcd instead of the real one... Useful for a very
     slow squelch.

     Example: sccparam scc0 soft on



4. Problems 
===========

If you have tx-problems with your BayCom USCC card please check
the manufacturer of the 8530. SGS chips have a slightly
different timing. Try Zilog... I have no information if this
driver works with baudrates higher than 1200 baud. A solution is
to write to register 8 instead to the data port, but this won't
work with the ESCC chips *SIGH!*

I got reports that the driver has problems on some 386-based systems.
(i.e. Amstrad) Those systems have a bogus AT bus timing which will
lead to delayed answers on interrupts. You can recognize these
problems by looking at the output of Sccstat for the suspected
port. See if it shows under- and overruns you own such a system.
If this happens to you feel free to #define SCC_DELAY/SCC_LDELAY
in scc.c. The PE1PET card needs SCC_DELAY defined.

Delayed processing of received data: This depends on

- the kernel version

- kernel profiling compiled or not

- the rather slow receiver in tty_io.c

- a high interrupt load

- a high load of the maching --- running X, Xmorph, XV and Povray,
  while compiling the kernel... hmm ... even with 32 MB RAM ...  ;-)

- NET's speed itself.


Kernel panics (based on excerpts from /linux/README)


- if a bug results in a message like

	unable to handle kernel paging request at address C0000010
	Oops: 0002
	EIP:   0010:XXXXXXXX
	eax: xxxxxxxx   ebx: xxxxxxxx   ecx: xxxxxxxx   edx: xxxxxxxx
	esi: xxxxxxxx   edi: xxxxxxxx   ebp: xxxxxxxx
	ds: xxxx  es: xxxx  fs: xxxx  gs: xxxx
	Pid: xx, process nr: xx
	xx xx xx xx xx xx xx xx xx xx

  or similar kernel debugging information on your screen or in your
  system log, please duplicate it *exactly*.  The dump may look
  incomprehensible to you, but it does contain information that may
  help debugging the problem.  The text above the dump is also
  important: it tells something about why the kernel dumped code (in
  the above example it's due to a bad kernel pointer)

- in debugging dumps like the above, please look up what the EIP value 
  means.  The hex value as such doesn't help me or anybody else very much: 
  it will depend on your particular kernel setup.  What you should do is 
  take the hex value from the EIP line (ignore the "0010:"), and look it up 
  in the kernel namelist to see which kernel function contains the offending 
  address.

  To find out the kernel function name, you'll need to 

         less /linux/System.map

  This will give you a list of kernel addresses sorted in ascending
  order, from which it is simple to find the function that contains the
  offending address.  Note that the address given by the kernel
  debugging messages will not necessarily match exactly with the
  function addresses (in fact, that is very unlikely), so you can't
  just 'grep' the list: the list will, however, give you the starting
  point of each kernel function, so by looking for the function that
  has a starting address lower than the one you are searching for but
  is followed by a function with a higher address you will find the one
  you want.  In fact, it may be [IS!] a good idea to include a bit of
  "context" in your problem report, giving a few lines around the
  interesting one. 

  There is a small program included in the kernel source named
  "ksymoops". It resides in linux/scripts. Copy the error report
  from /var/adm/messages into a seperate file, strip-off the
  syslog-prefix from each line and, taken that you named the
  file "foo",

	cat foo | linux/scripts/ksymoops linux/System.map

  and you'll get some output that may help you and will help
  me to find the problem. But only if it shows a problem within
  the routines of the driver, of course... ;-)

- alternately, you can use gdb on a running kernel. (read-only; i.e. you
  cannot change values or set break points.) To do this, first compile the
  kernel with -g; edit arch/i386/Makefile appropriately, then do a "make
  clean". You'll also need to enable CONFIG_PROC_FS (via "make config").

  After you've rebooted with the new kernel, do "gdb vmlinux /proc/kcore".
  You can now use all the usual gdb commands. The command to look up the
  point where your system crashed is "l *0xXXXXXXXX". (Replace the XXXes
  with the EIP value.)

  gdb'ing a non-running kernel currently fails because gdb (wrongly)
  disregards the starting offset for which the kernel is compiled.



If you can't solve a problem, send me

- a description of the problem,
- information on your hardware (computer system, scc board, modem)
- your kernel version
- the output of sccstat -p scc# ("#" is the No. of the channel)
- the settings of "speed", "clock" and "mode" for that channel
  in /etc/z8530drv.conf if sccstat -p doesn't show them.


And always remember: 
The 1.2.* series should run reliable. This driver perhaps does NOT!
The 1.3.* series is for alpha tests again --- you get the idea...


5. Appendix
===========

The fullduplex mode 3 works as follows: 

1. The protocol layer sends a KISS command 

	[PARAM_RTS] [TX_ON] 

   to key up the transmitter. Then it sends the data frame(s) to
   the driver. When all frames are sent, the driver will send the 
   KISS command 

   	[PARAM_HWEVENT] [HWEV_ALL_SENT]. 

   The protocol layer may now key down the transmitter with 

	[PARAM_RTS] [TX_OFF]

   or send more frames. Note that the maxkeyup timer may expire and
   key down the transceiver before everything is sent.

2. The driver sends

	[PARAM_HWEVENT] [HWEV_DCD_ON]
   or
	[PARAM_HWEVENT] [HWEV_DCD_OFF]

   if the status of the DCD changes.

3. The protocol layer can send

	[PARAM_HWEVENT] 0

   the driver will reply with one of the DCD status messages.

Note that KISS command doesn't necessarily mean that the enclosed data
is SLIP encoded. In network driver mode the driver does not encode/decode 
SLIP, but it will still distinguish between a data and a command packet 
by the leading byte, just like in 'real' KISS mode.

Note that this feature may vanish without a notice.