Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!rutgers!mit-eddie!genrad!decvax!ucbvax!C.ISI.EDU!Info-IBMPC
From: Info-IBMPC@C.ISI.EDU.UUCP
Newsgroups: mod.computers.ibm-pc
Subject: Info-IBMPC Digest V6 #10
Message-ID: <8702230331.AA02384@ucbvax.Berkeley.EDU>
Date: Sun, 22-Feb-87 20:08:33 EST
Article-I.D.: ucbvax.8702230331.AA02384
Posted: Sun Feb 22 20:08:33 1987
Date-Received: Mon, 23-Feb-87 18:38:42 EST
Sender: daemon@ucbvax.BERKELEY.EDU
Reply-To: INFO-IBMPC@C.ISI.EDU
Organization: The ARPA Internet
Lines: 1085
Approved: info-ibmpc@b.isi.edu

Info-IBMPC Digest       Sunday, 22 February 1987      Volume 6 : Issue 10

This Week's Editor: Billy Brackenridge

Today's Topics:

     KEYCODES.ASM Program to Display Keyboard Key Codes as Typed
			 Easier RESET switch
			Micro Emacs NANSI.SYS
			    Volume Labels
		  Bug in MicroEmacs NOT in Nansi.sys
		      Wordstar Compatible Editor
		       RAMPAGE RAM Disk Program
 EGA Hardcopy and PRTSC Drivers for Epson and IBM Graphics Printer II
			New version of PAS2C.C
		      How to Order EGA Reference
			   CTTY Help Wanted
			   2nd Drive for jr
		    Public Domain Prolog Available
			  PC-Write (2 Msgs)
			   Ribbon Re-Inkers
		  Another Bug in Microsoft MASM 4.0
			3 Comm 3+ Network Code
		     Tek 40XX Emulation Software
		       AppleTalk for the IBM PC
		8086 Assembler and MS-DOS file handles
			V23 Modem 1200/75 Baud (3 Msgs)
			   Reboot (2 Msgs)
		    REBOOT Program and 640K Limit
			  THRASHER (2 Msgs)

----------------------------------------------------------------------


Date: Sat, 14 Feb 87 17:55:53 EST
From: jcm@ORNL-MSR.ARPA (James A. Mullens)
Subject: KEYCODES.ASM Program to Display Keyboard Key Codes as Typed

Program KEYCODES.ASM reads the keyboard codes produced at the
keyboard i/o port (by installing a special INT 9 handler) and
displays the key codes of every key typed.  All key-make and
key-break codes are shown.  The program is intended for PCs, and for
ATs operating the keyboard in PC-compatibility mode (the normal
keyboard mode).  This program is intended to be an aid to persons
with clone keyboards having poorly documented nonstandard
features/keys.

[KEYCODES.ASM has been added to the info-ibmpc library -wab]

------------------------------


Date: Sun, 15 Feb 87 12:15:09 EST
From: Chris Schmandt <geek@MEDIA-LAB.MEDIA.MIT.EDU>
Subject: Easier RESET switch

I too needed a "reset switch", but did not own the hardware (IBM XT)
to which I wanted to add it.  So I did the following, which may not
be electrically perfect but has worked fine for 1.5 years now.  It has
the advantage of being easily removed without trace.

Theory: momentarily short the +5 output of the power supply to ground,
and you'll reset your PC.

Implementation:  Take a momentary contact pushbutton switch, and solder
a couple of short lengths of solid (not stranded) wire to it.  Connect
one side to ground, by screwing it under one of the mounting screws that
hold the mother board to the chasis (over in the corner of the board
near the power supply).  Make sure the screw goes through the (large)
ground areas on the board.  Take the other lead, and push it into the
white connector on the mother board from the power supply.  You want to
go in the hole closest to the rear of the connector (the wire is orange
on my power supply).   If you use suitable sized solid wire, you can just
shove it right in. Bring the switch out the back, or drill a hole in
the back of the chasis and bolt it down.

As I say, this may not be the best for the power supply, but as I write
in assembler, I have used it *LOTS* of times over the last couple of
years!

chris



------------------------------


Date: Sun, 15 Feb 87 15:39:24 CST
From: pyle@ngp.utexas.edu (Keith Pyle)
Subject: Micro Emacs NANSI.SYS


   The reason that the screen does not clear when MicroEMACS is invoked
and NANSI.SYS is resident is that MicroEMACS does not use the true ANSI
escape sequence to do a clear screen.  Both MicroEMACS 3.6 and 3.7 send
<esc>[J as the clear sequence when the ANSI sequence is <esc>[2J.  Note
that the brain-damaged ANSI.SYS will accept either string as a clear
screen.  I edited my copy of the MicroEMACS source to put out the 'correct'
string and it works fine with NANSI.SYS.  I suspect that the reason that
NANSI.SYS doesn't accept <esc>[J is that there is at least one other
<esc>[nJ ANSI sequence, although I can't remember it at the moment.
ANSI.SYS recognizes only the the one 'J' string.

Keith Pyle

UUCP: . . .{ihnp4,seismo,ctvax,noao,gatech}!ut-sally!ut-ngp!pyle
ARPA: pyle@ut-ngp

------------------------------


Date: 1987 Feb 16   00:12 EST
From: Bob Babcock   <PEPRBV%CFAAMP.BITNET@wiscvm.wisc.edu>
Subject: Volume Labels

To read a disk volume label, use INT 21H, function 11H.  To write a
label on an previously unlabeled disk, use function 16H (create file).
To change an existing label, use function 17H (rename).  You should
be able to access these from either C or assembler.
 
Beware of using functions 4EH and 4FH to search for the volume label;
they don't work properly (at least under DOS-2).  And don't try to
delete a volume label with a DOS call - under DOS-2 this will damage
the file allocation table entry for some other file, probably the
first hidden file if it's a system disk.

------------------------------


Date: Mon, 16 Feb 87 15:28:18 EST
From: John.Brennen@VI.RI.CMU.EDU
Subject: Bug in MicroEmacs NOT in Nansi.sys


I remember when I fired up MicroEmacs using NANSI.SYS, the screen wasn't
clearing correctly; lots of garbage left over from DOS...

So, I delved into the source code.  The problem was that MicroEmacs sends
the escape sequence ESC [ J to clear the screen.  This works in ANSI.SYS,
due to incomplete ANSI emulation.  However, the documentation clearly
states that ESC [ 2 J is the sequence to clear the screen, and this is
what NANSI.SYS wants.  Yet another example of not taking the documentation
seriously, tsk tsk.

Since this is the only software I've found with this particular problem,
and because I use Epsilon exclusively, I've never fixed uEmacs.  Shouldn't
be difficult at all, if you have the sources, which I believe are available
from SIMTEL20.

	John Brennen		CMU Visual Inspection Lab
	jfb@vi.ri.cmu.edu	Pittsburgh, PA

------------------------------


Date: Mon, 16 Feb 87 15:45:14 est
From: john watters <mcnc!ge-rtp.ge.com!watters@seismo.CSS.GOV>
Subject: Wordstar Compatible Editor

A program called GALAXY is available from OmniVerse
PO Box 2974
Renton, WA 98056-0961
Ph. 206-228-7627
BBS Ph. 206-235-8760


------------------------------


From: hplabs!felix!leviton@seismo.CSS.GOV (Dan'l Leviton)
Date: 16 Feb 87 20:52:19 GMT
Subject: RAMPAGE RAM Disk Program
Organization: FileNet Corp., Costa Mesa, CA


Here follows the MASM 1.0 (yes 1.0!) source for RAMPGDSK.SYS, a ram
disk for EMM memory. It has the advantage of running on an AST
rampage board without the need of loading REX.SYS to use the expanded
memory to emulate extended memory.  RAMPGDSK.SYS uses expanded memory
directly, so all you need to load before it is REMM.SYS.  It takes a
single parameter indicating the number of 16K pages you want it to
use for the ram disk. You can load multiple copies to get more than
one ram disk; each requires a PID. So for example, the lines

device=c:\remm.sys pids=2 depth=1
device=c:\rampgdsk.sys \010
device=c:\rampgdsk.sys \020

in your config.sys file will give you a 160K and a 320K ram disk. Note the
page parameter must have three decimal digits and be preceded by a slash.
If it is omitted, RAMPGDSK.SYS will allocate all available expanded memory
for the ram disk.

Also, the ram disks created by RAMPGDSK.SYS do not lose their contents on a
warm boot.  To make the SYS file, first use MASM, then LINK, then EXE2BIN,
then rename if desired. The resulting file is 1373 bytes in size.

Note the parameters to remm.sys in the above example are for minimizing its
memory usage. It takes a little more memory than the Intel version, because
it is a lot more flexible and self-configuring, and it supports an enhanced
spec.  (Incidently, I wrote the original versions of REMM and REX for AST, 
though they have been working on them since.)

Dan'l Leviton
[RMPGDSK.ASM has been added to the info-ibmpc lending library. -wab]

------------------------------


Date: Tue, 17 Feb 87 17:01:15 MEZ
From: RZ39%DKAUNI11.BITNET@wiscvm.wisc.edu
SUBJECT: EGA Hardcopy and PRTSC Drivers for Epson and IBM Graphics Printer II

 
 
A useful EGA-PrtSc driver for IBM-PC controlled via INT10, similar
to the original GRAPHICS.COM, which is supporting all of the different
EGA modes is available now from Internatioanl Software Distribution,
Hauptstrasse 23, D-6749 Vollmersweiler, West-Germany for about $40.
The EGA-Screen will be printed via <Shift> PrtSc together with any
IBM-Graphics Printer compatible Matrixprinter. The driver is able
to print in normal direction and sideways for better resolution.
We are using this driver and find it's running well.
 
regards Dieter
 
      Dieter Oberle
      Computercenter Micro-BIT
      University of Karlsruhe
      D-7500 KARLSRUHE
      West-Germany
      Phone: 0721/6082067


------------------------------


Date: Tue, 17 Feb 87 21:51:33 EST
From: jcm@ORNL-MSR.ARPA (James A. Mullens)
Subject: New version of PAS2C.C

  NEW VERSION OF PAS2C.C

  This program substitutes C keywords/operators for the Pascal equivalents.
  Over 60 Pascalisms (schisms) are translated into Cisms.

  Revision:

    Version 1.1  17-Feb-87  Changed several keyword translations on the
    advice of James R. Van Zandt <jrv@mitre-bedford.ARPA>.  Added many
    more translations.  Added a source for function strcmpi for the
    unfortunates who don't have this case-insensitive string comparison
    in their C library.

[PAS2C.C has been updated in the info-ibmpc lending library. -wab]

------------------------------


From: Herm Fischer <hermix!fischer@rand-unix.ARPA>
Subject: How to Order EGA Reference
Date: Tue Feb 17 10:03:06 1987

You can get the official IBM reference by calling 1-800-IBM-PCTB, and
asking for part number 6280131, the EGA update section for the 
options and adapters handbook.  They will want to bill $9.95 to
your VISA or American Express card, and my experience is that they
get it to you within a week or two.


------------------------------


From: Cheng-Ping Chang <cpchang%sunybcs%math.waterloo.edu@RELAY.CS.NET>
Date: 17 Feb 87 15:21:23 GMT
Subject: CTTY Help Wanted
Organization: SUNY/Buffalo Computer Science


Could someone kindly show me how to use the CTTY command and how to set
up the wire between two PC's. Should pin 2 and 3 be switched?
Are all 25 pins are required? Should I run MODE COM1:...  before I
issue CTTY ?  I am grateful for any helpful reply.

csnet : cpchang@buffalo
bitnet: cpchang@sunybcs



------------------------------


Date: Wed, 18 Feb 87 08:35:54 PST
From:     TOMASCHKE#GREG%E.MFENET@nmfecc.arpa
Subject: 2nd Drive for jr 

 
 
In the last issue of INFO-IBMPC, Andy Thomas asked about adding a
home-brew second drive to a PCjr.  A procedure for this was detailed
in the Sept 1984 issue of Home Computer Magazine (p82-86). While too
complicated to present here, it involves piggy-backing a couple
of extra chips to the jr's drive controller board, adding about ten
jumper wires, and making a new ribbon cable. The article also included
the software necessary to allow jr to recognize the second drive.
 
I have NOT tried this myself.
 
 
Regards,
Greg.
tomaschke@c.mfenet
tomaschke%c.mfenet@nmfecc.arpa
 

------------------------------


Date: 18 Feb 1987 13:37-EST
Subject: Public Domain Prolog Available
From: DAVSMITH@A.ISI.EDU

In response to the question about a Public Domain Prolog,  there
is one on the PCSIG at the Source.  I can get a copy uploaded if
you wish.  I worked with it briefly, and it seems to function
adequately.

------------------------------


Date: Tue, 17 Feb 87 22:18:40 pst
From: reynolds@ames-prandtl.ARPA (Don Reynolds)
Subject: PC-Write

Having been a satisfied Quicksoft customer since August 1983, and a beta
tester on PC-Write versions 2.60 and 2.70, I have many good things to
say about Quicksoft in particular, and the shareware concept in general.
Shareware vendors tend to be much more responsive to customer problems
than the commercial software sold retail through computer stores (though
PC-Write has been sold in stores as well, such as Software Galaria 
since version 2.60 was released).

Two points:

1) PC-Write source has always been available for the registered user price.
However PC-Write source is not shareware, so it could not qualify for your 
library.  I feel sure Bob Wallace will not mind me circulating the excerpt 
below from the PC-Write 2.71 banner screen:

 --------------------------------------------------------------------------
 A registered copy provides you all the following benefits for only $89:   
  Printed hardbound copy of the full PC-Write manual and Quick Guide.    
  Current PC-Write diskette pair and your unique registration number.    
  PC-Write support service for one year:  includes telephone support,    
  our quarterly newsletter, and two free update or source diskettes.    
  A $25 commission when someone registers and gives your reg. number.    
  Sincere thanks; your support helps us continue to improve PC-Write!
  Hardbound manual $45, diskette pair $16. Quantity prices. Group license.
  When you register, give registration #number above, to credit its owner.  
  PO, COD, Rush: add $5. Overseas add $20. WA orders add 7.9%. Visa/MC ok.  
  Quicksoft, Inc.  206/282-0452.  219 First North #224, Seattle, WA 98109.  



2) The release of PC-Write on SIMTEL-20 noted in Vol. 6, Issue 8 of this 
digest is in files named

	PCWR272A.ARC
and	PCWR272B.ARC

but rest easy.  They contain the perfectly good version 2.71 of PC-Write.
Further, the ED.HLP file is not "Squashed" with Dynamic Ziv-Lempel-Welch 
compression PKWARE used on PKARC 2.0 and PKXARC 2.7 and later.  The files 
on SIMTEL-20 can be unpacked with the earlier version of ARC 5.1 from 
SEAware.  The labeling of the files appears to be an unfortunate error.

Best,
Don

------------------------------


Date: 18 Feb 1987 14:28:43 EST
Subject: PC-Write
From: Glen Foster <GFoster@A.ISI.EDU>



The version of PCW announced herein available for FTP from SIMTEL-20
is actually the non-trojan 2.71 version.  It has been serialized by
someone but is the legit version.  I guess the announcement of 2.72
is an inducement to obtain more shareware "commissions" from all us
naive computer users.

I have unarced the program at SIMTEL-20 and have run it successfully on 
a Compaq 286 and HP Vectra with no damage to the respective hard
disks.  Doesn't give you much more than 2.7 did, slightly more 
efficient PostScript support, guesses only words that have the same
initial as the target while spell checking (I liked the old way!)
and some "bug fixes."  Probably not worth the download time if you
already have 2.7.

Glen

------------------------------


To: "R.Rasulis" <smggyuc%BOSTONU.bitnet@WISCVM.WISC.EDU>
Subject: Ribbon Re-Inkers 
Date: Wed, 18 Feb 87 16:41:46 EST
From: 44-28csh@braggvax.arpa


Hi,  Found whats called a MAC INKER.  CAll this toll free: 1-800-547-3303
(in Oregon dial: 503-626-2291)

We are planning to purchase 2 or 3 for our ribbons.  Yep the cost does add up!

This article compliments of Personal Computing "DEC 86" page 192.

The operator will ask you a couple Q's on your ADPE.

CIOA!!

Michael 

------------------------------


Date: Wed, 18 Feb 87 13:08:46 PST
From: Ya'akov_Miles%UBC.MAILNET@MIT-MULTICS.ARPA
Subject: Another Bug in Microsoft MASM 4.0

There appears to be a bug in Microsoft MASM assembler ver 4.0, as
is demonstrated.  Note that the ASSUME directive seems to require
that the segments be specified in a specific order, or forward
references from CALL/JMP do not work properly...

    .Title This does not work
    ASSUME CS:myseg,DS:myseg,ES:myseg,SS:myseg
    myseg  SEGMENT
           call    abcd
    abcd:  ret
    myseg  ENDS
    END

    .Title This works
    ASSUME DS:myseg,SS:myseg,CS:myseg,ES:myseg
    myseg  SEGMENT
           call    abcd
    abcd:  ret
    myseg  ENDS
    END



------------------------------


From: jeff@trwspf.UUCP (Jeff Spahn)
Subject: 3 Comm 3+ Network Code
Date: 18 Feb 87 20:37:00 GMT


I've recently had a very painful experience with 3Com's network software,
and thought I should pass along some warnings to anyone thinking of buying
it.

First, 3Com's 3+Plus(tm) server software takes a LOT of memory.  The name
server (required on every network) needs 400k of memory, the file/print
server needs 382k, and the mailer needs 371k.  When combined these take
less memory; 539k for the file/print server and the name server, 528k for
the name server and mailer, 510k for the file/print server and the mailer,
and 667k for the file/print and name servers and mailer (yes, that's
right, since the maximum memory on the PC is 640k, this configuration is
not supported).  There won't be much memory left for anything else on the
server computer.

Second, 3Com's 3+Plus file/print server is copy protected.  While this by
itself is not bad, it can be very serious if the copy protection software
gets cranky.  A month ago, in mid-January, I attempted to install a 3Com
3+Plus system (name, file, and print servers) on a group of IBM-PC/AT's
and clones with 3Com's Etherlink boards here at work.  I ran into several
snags, and in the course of trying to track down the problem (later found
to be due to a bad cable) reformatted the hard disk.  

According to the documentation, the file/print server can only be
installed at a single ethernet address (obtained from the ethernet board
during installation), and once it is installed it can NOT be de-installed
and installed at another address.  However, it can be re-installed as long
as the ethernet address remains the same.  Or so says the manual.  But
when I went to re-install the file/print server I got a message informing
me that I was trying to install the server on a different ethernet board
than it had originally been installed on, and the software refused to
re-install.

At that point I tried calling the dealer we bought the software from and
3Com to find out what the problem was, which brings me to my third warning
about 3Com's software. It turns out that 3Com no longer provides direct
support to users (unless you have a support contract), so they wouldn't
help me.  After several phone calls to the dealer who sold us the software
and to 3Com it was finally agreed that I would send 3Com the original
floppy disk, they would repair it and send it back, and we would all live
happily ever after.

Would that it were so.  When the floppy came back (a week later, they were
quick), it still refused to install.  Somewhere around this time (my notes
are a bit fuzzy) the installation package also developed the amusing habit
of hanging the system, so that I had to turn off power to reboot.

At that point I told the dealer I wanted to return the software.
(Preferably in exchange for some other, installable, package.  Novell
perhaps?)  They persuaded me to talk to 3Com's local sales representative.
He called (several days later than promised) and said their local engineer
was not available but that he would call me Tuesday and we'd work out a
solution.  It's Wednesday afternoon (2/18), and I still haven't heard a
word.

Now, I probably did make some errors while trying to install the 3Com
software, and some of their people did try to be helpful, but I've been
working with computers a long time and should be able to handle a simple
software installation without it blowing up in my face.  With the
exception of one or two individuals 3Com has been uniformly unhelpful and
uninterested in anything but getting money out of me. (One of the things
they would tell me for free was that there is a new version out which is
SOOO much easier to install, and which only costs an additional $200.)

I don't know if I'll ever get their system running.  But I'm still trying
and will report any major developments to the net.  Are you listening
3Com?

In summary:
    - 3Com's 3+Plus network server software requires a lot of memory.
    - 3Com's 3+Plus network file/print server is copy protected and the 
      copy protection has blown up at least once.
    - Don't expect any free help from 3Com if you have problems with their
      software.  You probably won't get it.

[Our experiences have been similar, however, we are paying for 3 Comm
technical support. It is worth the price, though nearly all of our
calls have been in regards to copy protection stuff. We threw away 
the 3+ code. In addition to the problems you mentioned it disables
interrupts for periods of time longer than a 9600 baud character time.
This blows out all our communications programs. Three Comm's response
has been pure arrogance. (Get an AT or "You don't need communications
programs if you have our software") We restored our old 3Comm ethershare
software. While we would prefer a distributed file system as opposed
to a distributed disk system, the older ethershare meets enough of
our needs while we are waiting for a better solution. I strongly
reccommend AVOID 3 COMM 3+!! -wab]

------------------------------

From:     "Bob Klein" <KL2%NIHCU.BITNET@wiscvm.wisc.edu>
Date:     Fri, 20 Feb 87  09:36:39 EST
Subject:  Tek 40XX Emulation Software

Regarding A. Davison's request for Tektronix 40XX emulation
software, we have had very good experience with PC-PLOT
(MicorPlot Systems Co., 659 Park Meadow Rd. Suite H,
Westerville, Ohio 43081, phone 614-882-4786).  It costs less
than $100.00.  We have used it with EGA but it also supports
Hercules and CGA among others.  It includes several nice features
such as storing different terminal setups, storing a phone number,
and storing a logon string.

------------------------------


Date: Thu, 19 Feb 87 14:55:41 pst
From: Robert Lenoil <lenoil%apple.csnet@RELAY.CS.NET>
Subject: AppleTalk for the IBM PC

(I am definitely an "interested party" in the following product announcement,
 but I think it could be of interest to the Info-IBMPC community.)

Last month, at the Seybold Conference on Desktop Communications, Apple
formally introduced its Desktop Communications strategy, and demonstrated
several new products that fit in with that strategy, most notably the
AppleShare file server.  Also announced was the AppleTalk PC Card, a $395
half-size card for the IBM-PC, which puts the PC on the AppleTalk network.
The card comes packaged with a software driver, and with a menu-driven
utility for converting files created by popular PC applications to PostScript
and sending them to an Apple LaserWriter for printing.  Apple has made the
driver interface available to third party developers, and there already are
programs for such things as electronic mail, printing, and access to files
on Macintosh.  Apple also announced its committment to produce MS-DOS
software for transparent access to AFP (AppleTalk Filing Protocol) file
servers.

I have just unwrapped one of the first finished products to roll off the
assembly line, and I am told that dealer shipments have begun.  For more
information, contact an authorized Apple dealer (call (800) 538-9696 to
find the name of a dealer in your area).  The part number for the AppleTalk
PC Card is M2050.

Robert Lenoil
Apple Computer, Inc.
Network Systems Development


------------------------------


Date:           Sun, 15 Feb 87 23:29:13 PST
From:           Dana Myers <bilbo.dana@LOCUS.UCLA.EDU>
Subject:        8086 Assembler and MS-DOS file handles


In volume 6, issue 9 of this Digest, two items caught my interest. One
was the 'bug' in MASM 4.0 and the other was the continued concern about
the 'MAJOR BUG' in MS-DOS concerning file handles.

1. The 8086 instruction set does not support either addressing mode that
Y'akov Miles was trying to use. It is completely wrong on any version of
the 86 family to expect to be able to fetch a word from memory, add a
register to it, and then fetch that word to a register IN ONE INSTRUCTION.
The 8086 simply does not have that addressing mode.

  Except for the 80386, the DX register cannot be used as a pointer
register as Y'akov seems to think he can.

  It is no bug for MASM 4.0 to interpret 'mov AX, [DI+DS:1234]' as
'mov AX, [DI+1234]'. Also recall that the default segment register used
in an access using the DI register is the DS register, so no segment
override need be generated (yes, I know that a STOS instruction uses ES:DI
but that is not a normal addressing mode).

  If MASM 4.0 assembled 'mov AX, [DX+DS:1234]' as 'mov AX, [DX+1234]',
THAT would be a bug.... there simply is no 8086 addressing mode that uses
DX as an index register... my copy of MASM 4.0 does generate an error.

2. MS-DOS updates directory entries on the closing of a file. MS-DOS reads
directory entries when a file is opened. If a file is opened, written to and
not closed, then another open will get the (non-updated) directory entry.
This has been traditional DOS behaviour. No major bug.


(The above statements are all
my own, and, as such, are prone to error)

Dana H. Myers
Locus Computing Corp.
Santa Monica, CA

Reply to 'bilbo.dana@LOCUS.UCLA.EDU'.

[dewar@nyu-acf2.arpa,
Bill Marsh of the Naval Ocean Systems Center, San Diego, CA,
and Frank Schwab of the Institut fuer theoretische Physik
all responded about this addressing mode error. The editor apologises
for letting it pass. I make these "wishful thinking" addressing errors
all the time. I am also all to ready to believe any horror stories about
MASM -wab]

------------------------------


Date: Mon, 16 Feb 87 02:50:23 EST
From: David Kirschbaum <kirsch@braggvax.arpa>
To: TWENK%HENTHT5.BITNET@wiscvm.wisc.edu
Subject: V23 Modem 1200/75 Baud

Martin,

I read Billy Brackenridge's response and must agree.  I had a glimmer of
a solution:  horribly abuse the 8250 by switching baud rate to 75 to check
for RX, then switch the poor suffering chip back to 1200 for TX.  However,
I strongly suspect the poor thing won't go along with that for long.

I don't have any documentation for the 8250 that says it'll in fact permit
setting up to 75 baud (lowest indicated is 110 baud), but then there
ain't none in the PC manuals for 19200 baud either, and that works just fine!
Logically, the "magic number" for 75 baud should be 150 baud's divisor
(768) * 2, and that would be easy enough to test.

The main (and probable fatal) problem is that I strongly suspect switching
baud (plus data bits, stop bits, whatever) would destroy the very data
we're trying to receive!

No problem on the TX side:  don't stuff the data byte to the port until the
chip is reset.  The overhead is kinda bad, resetting baud for every byte,
but at 1200 baud I suspect there'd be plenty of time for that stuff.

But still .. the 8250 just wasn't meant for that kind of antics!

If we get any encouraging feedback over the net about this subject,
I'll be glad to try a hack .. already got the testbed software handy,
and my 8250 takes plenty of abuse already!  Will only do "back-to-back"
via ribbon cable with another PC since I don't have such a peculiar modem
nor a remote one to talk with.

Netlandians, any suggestions?  Encouragement?  "Forget it, Clyde,
it won't work!"?

David Kirschbaum
Toad Hall
kirsch@braggvax.ARPA

[The solution is to put all this madness in the modem. The modem can appear
to be 1200 baud in both directions while actually only transmitting 1200/75.
If this is done correctly in the modem any need for special communications
programs go away. -wab]

------------------------------


Date:     Fri, 20 Feb 87 00:58 N
From:     <DRAYER%HGRRUG5.BITNET@wiscvm.wisc.edu>
Subject: V23 Modem 1200/75 Baud


Refering to Martin Beekmans' 1200/75 modem problem in issue nr9:
 
Modems are availabel with a split-speed convertor, employing a buffer so
the PC sees a 1200 FD modem. This buffer is usually 16 bytes long.
 
In my experience VT100 character sequences make this buffer overflow, and
characters are lost, even when CTS/DTR lines are employed. This in turn re-
sults in the receiving end producing error messages or rubbish on the screen.
It's a rather frustrating experience editing full screen, having to count to
five (slowly) after pressing a cursor key...
 
The same small buffer and the 75 bd speed create problems when trying to
download a file, using some protocols/programs. A few do enable a 1200 bd
download, most endup failing dismally with a timeout or too many errors.
 
If you have to pay for the modem out of your own pocket, a modem with 300
bd and 1200/75 is a better choice than only 300 bd. A modem with only
1200/75 bd disables you: uploading is impossible.
If you have enough money: get a modem capabel of 1200/1200 as well as 1200/75,
it's well worth it. 1200 and even 2400 do work on Dutch telephone lines...
usually!
 
                     Nick Drayer jr.
                     Groningen University
                     Holland
                     Earn/Bitnet: drayer@hgrrug5

------------------------------


Date:     Sun, 22 Feb 87 16:26 N
From:     <TNWIJERS%HENTHT51.BITNET@wiscvm.wisc.edu>
Subject: V23 Modem 1200/75 Baud


          Haay,
 
First I want to thank David Kirschbaum and Billy Brackenridge for their
response to my problem.
My problem was to get a communication program which had different send-
receive-speed for instance 1200 receive and 75 send.
I have looked around and found finally a program which had this possibi-
lity. It is called Supercom. It is made here in Holland. Its 64 K big.
The program doesnt contain a vt100-emulator!
Someone (I thought it was David) talked about nansi.sys.
 
Question: I dont know much about ansi.sys and I looked everywhere for
          information but is it possible to modify the program so I
          can run vt100 programs on a main frame?
 
Greetings from Martin Beekmans.
          <TNWIJERS@HENTHT51 Holland>
 
[I sent him a copy of NANSI.SYS sources, but perhaps someone closer could
help. -wab]

------------------------------


Date: Mon, 16 Feb 87 02:49:52 EST
From: David Kirschbaum <kirsch@braggvax.arpa>
To: SCOTTH%GMUVAX.BITNET@wiscvm.wisc.edu
Subject: Reboot

Scott,
There's REBOOT.ASM in INFO-IBMPC's archives also .. does a nice job
indeed.  Tweaks a peculiar byte in DOS that tells it to warm-boot.

Because I don't have a battery-supported clock in my XT clone
(running PC-DOS 3.1), I hacked REBOOT.ASM to include the DTSAV.ASM code
that saves current date/time to a batch file.  DTSAV produces a wee
little text file called DT.BAT with the current date and time as DOS
commands, e.g.:
DATE 15/02/87
TIME 22:15
All I have to do to reset a fairly recent date/time on REBOOT is to include
the command DT in my AUTOEXEC.BAT.

If you are satisfied with a straightforward reboot, use REBOOT.ASM.
If you want the fancier stuff, yell and I'll send the hack to you.

David Kirschbaum
Toad Hall
kirsch@braggvax.ARPA

------------------------------


Date: Mon, 16 Feb 87 15:27:48 EST
From: John.Brennen@VI.RI.CMU.EDU
Subject: Reboot


In order to reboot a PC, PC clone, PC compatible or whatever, you can
use the following code (to make a .COM file, cut the text between the
lines into a file, say "foo.bar", and on a PC, run "debug < foo.bar").
 ----------------------------------------
a100
xor ax,ax
mov ds,ax
mov [472],wo 1234
jmp ffff:0

nreboot.com
rcx
f
w
q
 ----------------------------------------

The storage of 1234 in absolute memory location 00472 avoids the
memory check.

If you have an AT, AT clone, AT compatible or whatever, you can pulse
the RESET line through software control:
 ----------------------------------------
a100
xor ax,ax
mov ds,ax
mov [472],wo 1234
mov al,fe
out 64,al
jmp 10e

nreboot.com
rcx
10
w
q


	John Brennen		CMU Visual Inspection Lab
	jfb@vi.ri.cmu.edu	Pittsburgh, PA

------------------------------


Date:    02/18/87 17:06:06 GMT+1
From:    UF02%DDAGSI3.BITNET@wiscvm.wisc.edu
Subject: REBOOT Program and 640K Limit

Frank Schwab
069/798-8238
Institut fuer theoretische Physik
Robert-Mayer-Str. 10
D-6000 Frankfurt/M.
This letter is to answer a question in Info-IBMPC V6  #9  and  to
correct a common misunderstanding I want to make clear.
 
 
1.)
First I want to answer the question of Scott Hutchinson:  "How do
I reboot?".  This is not a complicated task.  You  just  have  to
jump  to  where  the 80x86 jumps after a hardware reset.  This is
location 'F000:FFF0'.  But if you just use  a  'JMP'  instruction
you'll  have to go through all the memory testing.  Therefore the
BIOS designers put a small trap door into the  BIOS.   Place  the
value  '1234H'  into  the RESET_FLAG location at '40:72' and when
the BIOS gets control and finds that value it  skips  the  memory
tests.
 
The  whole  program  can be debuged into the PC very easily.  Put
the following debug commands which are  surrounded  by  the  'Cut
here'  marks  into a file called 'reset.dbg' and then start DEBUG
with 'debug < reset.dbg'.  (Please note that the blank line after
'JMP   F000:FFF0'   is   mandatory!). This  will  create  a  file
'reset.com' which does the  resetting.   (Check  this  by  typing
'reset'  at  the  DOS  command  level but save your previous work
before).
 
Those of you which don't want to save the way  they  created  the
program  can also type the commands directly after calling DEBUG.
 
 
---- Cut here ---- (begin)
N RESET.COM
A
MOV AX,40
MOV DS,AX
MOV WORD PTR [72],1234
JMP F000:FFF0
 
R CX
10
W
Q
---- Cut here ---- (end)
 
If you want it a bit more complete you can also use the following
assembly program and build it together with:
 
    MASM RESET;
    LINK RESET;
    EXE2BIN RESET RESET.COM
 
The program is again surrounded by 'Cut here' marks:
 
---- Cut here ---- (begin)
PAGE    65,130
TITLE   RESET --- Requests a System Reset (Ctrl+Alt+Del)
 
;**********************************************************************
;*                         R E S E T   V3.01                          *
;**********************************************************************
;*                                                                    *
;* RESET jumpes to the 80x86-reset-location (F000:FFF0) to perform    *
;* a system reset. Before there is placed a special value into the    *
;* RESET_FLAG (40:72) to indicate that this is a software reset. This *
;* will cause a shorter reset sequence on all IBM-PCs and will also   *
;* reset all other 80x86-machines. The method of using INT 19H to     *
;* reboot has been discarded because it does not reset the interrupts *
;* which causes failure at restart because DOS-resident programs have *
;* been cleared but the corresponding vectors have not.               *
;*                                                                    *
;* ---- This program is hereby donated to the public domain. ----     *
;* ---- (I know that this is not too much of a donation.)    ----     *
;*                                                                    *
;**********************************************************************
;*                                                                    *
;*      Author            :   Frank Schwab                            *
;*      Start             :   16 May 1986                             *
;*      Last Modification :   11 Jun 1986                             *
;*                                                                    *
;*      Buildup           :  MASM RESET;                              *
;*                           LINK RESET;                              *
;*                           EXE2BIN RESET RESET.COM                  *
;*                                                                    *
;**********************************************************************
;*                                                                    *
;* Syntax:      RESET                                                 *
;*                                                                    *
;**********************************************************************
;*
 
 
BIOS_DATA_SEGMENT SEGMENT AT 0040H
 
        ORG     0072H
RESET_FLAG      EQU     THIS WORD
 
BIOS_DATA_SEGMENT ENDS
 
 
BIOS_CODE_SEGMENT SEGMENT AT 0F000H
 
        ORG     0FFF0H
RESET_ADDR      LABEL   FAR                     ;80x86-Reset addr.
 
 
BIOS_CODE_SEGMENT ENDS
 
CSEG    SEGMENT 'CODE'
 
        ASSUME CS:CSEG,DS:BIOS_DATA_SEGMENT,ES:NOTHING,SS:CSEG
 
        ORG 0100H
 
RESET:
        MOV     AX,SEG BIOS_DATA_SEGMENT        ;Signal that this is
        MOV     DS,AX                           ;a keyboard reset.
        MOV     RESET_FLAG,1234H
 
        JMP     BIOS_CODE_SEGMENT:RESET_ADDR    ;O.k., do it
 
 
WHO     DB      'RESET  V3.01  11 Jun 1986  Frank Schwab'
 
CSEG    ENDS
 
        END    RESET
---- Cut here ---- (end)
 
 
2.)
When I read the last digest I found Ron Belanger speaking of  the
"DOS 640K limit".  I really wonder where it is!  Those weird 640K
are NOT a DOS limit!  (exclamation mark).  DOS uses the 80x86  in
its  so  called  'real  mode' (which is the only one for 8088 and
8086, of course).  In this mode the processor can  address  2**20
bytes which is one Megabyte.  DOS uses the full address space and
so it can also use 1MB.
 
The restriction is imposed ONLY by the PCs  memory  design  where
the  screen  buffers  and  the  ROM  begin at address 'A000:0000'
(which happens to be 640K, oh wonder) and eat up the  space  from
there  to  the  top  of  the address space.  I'm pretty sure that
people would start to wonder when DOS should try to use the video
buffers  for  its programs.  So, people get this DOS limit out of
your head and call it the PC limit because this is what it is!
 
 
Frank Schwab
Institute for Theoretical Physics
University of Frankfurt
Germany
UF02@DDAGSI3 (this is an EARN node so you might need to add '.BITNET')

------------------------------


Date: Mon 16 Feb 87 13:08:33-PST
From: Cliff Yamamoto <CYAMAMOT%GUMBY@USC-ECLC.ARPA>
Subject: THRASHER



	In response to the question in the previous issue about 
Thrasher, yes it really works.  Although as the documentations mentions,
it would not be a program you'd like to run everyday.  Being unsure 
whether or not I should run it with my cache program I ran it with and
without.  I'm not sure of the exact timings that the others obtained,
but my lowest values "clustered" about 16 buffers with the cache and
17 buffers without it.  

	I say "clustered" because the lowest timings that I obtained around
the 16 buffers area were also obtained at other buffer settings.  For
example I obtained 34 second reports at buffers = 8,9,10,16,17,18,19,23 & 28.
With the cache I obtained 15 second reports at buffers = 7,10,15,16,17 & 20.

	If anyone would provide additional insight to my reasoning I would
be glad to hear about, but in the end after obtaining my data I reasoned that
because of the clustered set around 16-19 w/o the cache and 15-17 with, I
decided that 16 buffers would be best suited for my system.

	It seems that 15 or 16 thus far has been the ideal value, although
the entire test is at the whim of the 100K test that is created.  However, in
short, Thrasher seems like a good program although I think if we really wanted
to be picky, having the tenths of seconds displayed would be helpful.  

Cliff Yamamoto


------------------------------


Date: 16 Feb 87 12:26 PST
From: Gobbel.pa@Xerox.COM
Subject: THRASHER


I tried running Thrasher both with and without the disk cache program
that I usually use (Super PC-Kwik).  Without the cache, the fastest runs
are around 400 seconds, with the number of buffers a multiple of the
number of sectors on a track for my disk (9, 18, 27, etc.).  With the
cache, all runs with more than 2 buffers take about 45 seconds!  The
main reason for this amazing difference is that with the cache, the disk
is not used at all after the initial file creation.  While I'm a big fan
of this cache program, my feeling is that what this really indicates is
that Thrasher is not a very realistic test.  A better way to do this
sort of thing would be to look at the pattern of disk accesses generated
by real programs, and simulate that.  Also, I'm not thrilled with the
way that Thrasher beats on my poor disk drive.

There do seem to be a couple of helpful bits of information out of all
this: 1) it helps to set the number of buffers to be a multiple of the
number of sectors on one track for your disk, and 2) a good disk cache
program is really nice to have.  The one I'm using apparently (judging
from reviews) works much better than Lightning, the other widely used
commercial disk cache.

Disclaimer: I have no connection with the makers of any of the stuff
I've just mentioned.

-Randy

[Many people have sent outputs from their thrasher runs. I haven't printed the
raw data as there were no results much different from these summaries. -wab]

------------------------------

End of Info-IBMPC Digest
************************
-------