Genie & Lamp Feb97 5/20

[norman.dodge@techall.wa.com]

Hello News!

>>>>>>>>>> CSplit: Version 2.2 >>>>>>>>>>
>>>>>>>>>> CSplit: Begin part 5/20 >>>>>>>>>>
cases.  Depending on the device driver(s) in question, this might work.
\

Here's
the problem tho:  The MLI sets up a command block (device driver
parameter \

list)
prior to calling the device driver.  One of those parameters in the
command block is UNIT NUMBER (found at $43) which is in the format:
%DSSS0000, \

where D
is the drive number minus one, and SSS is the slot number.  OK?  Keep
this \

in
mind.     Next, let's take a look at the hardware side.  Most external
\

cards can
be plugged into any slot and work.  The easiest way to communicate with
a \

card
is to read/write at the slot hardware _base_ address, using the
X-Register \

to
specify the actual slot offset.  For example, the assembly language \

instruction
LDA $C080,X  where X=$10 would access slot 1.  If X=$20, then the access
\

would
be to slot 2, etc.  So, the people who write device drivers, only need a
\

base
address ($C080) and where their card is at (loaded in the X-Reg).  Now,
\

let's
put the two together.

1.  The device driver needs the X-Register to be set to the slot
number. 2.  The MLI issues the slot number as part of the Unit Number.

Hmmm...If you were a device driver, where would you get that slot \

number??? :)
And everythings fine until.... someone _swaps_ the entries in the P8
device driver table and _doesn't_ tell P8 about it :(

Say you have a 3.5 drive on slot 5, and a SCSI card in slot 7, and you
\

swap the
device drivers.  =Potentially= you have the 3.5 driver accessing the
SCSI \

card,
and the SCSI driver accessing the 3.5 drive.  +NOT+ what you want!!!
AND- I don't even want to think about what would happen if the two
swapped entries refered to different devices, i.e. a serial device and a
block device.  \

Yikes!

So, I can be wrong on this, because not all device drivers are written
the \

same
way.  Maybe it'll work for you, but I wanted to at least inform you that
the results may prove disasterous.  Also, since I'm jumping in here,
maybe I \

drew
the wrong conclusion about where this discussion was going to go, and in
\

that
case Sorry!

BTW, I'm currently writing a P8 device driver to control an IDE HD with
\

multiple
partitions, which the user may map into unused P8 slots.  (like the \

RamFAST).
The basic theory is to fill the unused P8 device driver vectors with my
\

device
driver entry point and then translate the different Unit Numbers passed
by \

the
MLI into the correct partition, there by creating virtual drives.  If \

someone
tried to swap the entries, nothing would happen because the entries are
the same-- it's the Unit Number that makes the difference.

So, basicly, hot swapping the device driver entries is a BAD idea, in
my opinion, of course :)

-Sully
(All miss-spellings are my own :)
                    (R.SULLIVAN8, CAT9, TOP23, MSG:7/M645;1)

>>>>>   Hans:
"""""
Sully knows exactly what I was hacking to do, and makes great points.
Don't worry though, I wasn't going to let you walk into the quicksand,
at least \

not
without adequate warning and protection. <g>

Currently, what cards are in slots 5 and 7, respectively, and how many
\

devices
are attached to each card? If slot 5 is the built-in diskport, how many
\

drives
are attached?

Will the program from which you want to access the 'swapped' slot device
\

address
files on it from BASIC with BASIC.SYSTEM commands (e.g. LOAD X, S5,D1),
or \

will
it address it from machine code?

BTW, before you even attempt to swap the $C7 0D for the $C5 4E with a
few \

POKES,
_everything_ needs to be backed up. Plus, unless Sully bales me out with
the answers, I need to re-read Gary Little's explanation about the whole
mapping procedure, as another location (Sully's $BF32,X) may also need
to be \

massaged.

Of course, if you're a curious sort, and a bit impulsive, and if you've
_got everything backed up_, and your estate planning is in order, you
can go \

ahead
with the POKES anyway, and let us know what happens. <g>

Hugh...
                      (H.HOOD, CAT9, TOP23, MSG:8/M645;1)

>>>>>   Hans, Hugh,
"""""
I whipped up a short program that will correctly handle the slot 5/slot
7 \

swap
problem.  This program applies a patch to P8 that will intercept calls
to \

slot
5/7, fix the Unit Number, then passes (redirects) control to the
opposite \

slot.
When the program is launched, it will relocate itself between the BI and
the buffers, so it is protected from other programs. The patch must be
applied \

each
time Prodos is launched, but never twice in the same session.  I'm \

emailing you
guys a copy of both the program and the Merlin source.  I used GSHK
with BinaryII, hope that isn't a problem.

-Sully
                    (R.SULLIVAN8, CAT9, TOP23, MSG:9/M645;1)

<<<<<   Hugh and Sully,
"""""
Actually, that's exactly what I want to streamline -- the HD backup
process. Here's what I've got:

The machine is a Laser 128EX with an expansion box, so I've got two \

available
expansion slots -- slot 5 and slot 7.  Each slot is controlled by a
hardware switch.  If I switch slot 5 "on", I can only access 128K of
ram, and since \

I use
AppleWorks a lot, slot 5 remains dedicated to ram.  If I switch slot 7
\

"on", I
lose access to any 3.5" drive connected to the external drive port (a 5
1/4" drive simply responds to s6,d2, but a 3.5" drive is addressed as
s7,d1 -- \

same
port).  So the problem is that I have three "devices" (1 MB of ram, HD,
\

and 3.5"
drive) competing for two resources, but I only ever need any two of them
\

at the
same time.

In normal use, I have 1 MB of extended ram tying up slot 5 and an Apple
\

High-
Speed SCSI card in slot 7.  For the occasions when I need to transfer a
\

file to
(or from) a 3.5" drive, I simply set up a ramdisk, temporarily move the
\

file (or
disk) to the ramdrive, flip the switch, and reboot. Although this is
fine \
>>>>>>>>>> CSplit: End part 5/20 crc: e1c9 >>>>>>>>>>

 * 2qwk! 2.03 * The first full-featured QWK reader for the Apple.