VP/ix serial terminals

[Chris Lewis]

In article <240 at ssbn.WLK.COM> bill at ssbn.WLK.COM (Bill Kennedy) writes:
>I am running AT&T 386 UNIX with Simul-Task 386 (aka VP/ix) in a
>no name clone with a CompuTone AT-4 smart serial card....
>
>I have some anomalies that I can not explain for myself.  If I
>try to run DOS from a terminal on the smart serial card I get a
>pregnant pause, a message
>
>Cannot enable DOS mode for this device
>

How old is your AT-4?  Some of the older AT-4's and Atvantage-X
cards don't support VP/IX - it's largely a function of firmware and
drivers - bug Computone for a new version.  386/ix contains some
documentation in the supplements manual (can't remember the man page
name, but heck, there's only 10 or so mans ...) on the facilities
that a serial driver has to support in order to run VP/IX.  Eg:
an extra couple of ioctl's, and 8250 register emulation... Yup, 8250
emulation!  How else do you think DOS programs change baud rates etc?

>a much longer pause and then my UNIX prompt back.  I can log in with
>my luggable in cu (onto a modem on COM1) and it works fine.  If I try
>to run DOS from a directly connected terminal on COM2 I get a very
>prompt message
>
>Error attempting to open serial device
>Press Enter to continue
>
>and then everything is OK.  I don't get that message when logged in on
>a modem or when I'm on the console (I suppose that's obvious).  Perhaps
>the brain trust can enlighten me.

This is likely something to do with modem control lines (or the
lack of it).  Try putting "CLOCAL" in the gettydefs entry (or use
the ...H modtty settings).

>I'm curious as to why it won't work on any smart card port.  That's
>probably a function of the driver and something the driver must do when
>VP/ix wants to open it.  I hope not because I'm getting ready to replace
>the COM2 driver with one that takes advantage of the NS16550A UART and
>CTS/RTS handshaking.  Can anyone explain what a serial driver must do
>in addition to work with VP/ix?

A few things: 
	- DOS flow control sometimes uses scan codes instead of X-ON/X-OFF
	- the UNIX driver has to schedule fake interrupts for the VP/ix
	  handler and manage some of the other exception handling stuff.
	- VP/ix needs to be able to issue ioctl's to the device that
	  are used to emulate a DOS program's peeks and pokes to a 8250.
	  (eg: baudrate, stop bits etc.)

>Whoops!  I said "finally", I lied.  Does anyone know why VP/ix must
>dongle both floppies (A & B) before it will start?

*Both* drives?  That's a little unusual...  Maybe that's just an artifact
of your controller.  Consider how PC's DOS/BIOS works: when booting the
BIOS ALWAYS goes to the floppy first to see if it's there.  If not, it
sits there trying for a bit then gives up and boots from the hard disk.  
This is *exactly* what VP/ix is doing...  Consider that VP/ix is actually
providing a "virtual PC", *NOT* emulating the BIOS or DOS, and VP/ix can
and will boot vanilla Microsoft DOS right from your floppy if one is there.

>I tried three things I was certain would not work, DAC Easy
>Accounting (ancient uSoft BASIC), the venerable szap from Software
>Toolworks, and MKS vi.  They are all very screen memory oriented and
>szap is very function key oriented.  They all worked!  They even ran
>OK on the DOS partition on drive 0.  I was impressed.

Try some harder ones like programs doing direct video ram writes, Microsoft
Windows, Wordstar, WordPerfect, Lotus, Symphony and I have it from a 
relatively reliable source that a VP/ix session will now run NOVELL client 
with arbitrary ethernet cards!
-- 
Chris Lewis
{uunet!attcan,utgpu,yunexus,utzoo}!lsuc!ecicrl!clewis
Ferret Mailing list: ...!lsuc!gate!eci386!ferret-request
(or lsuc!gate!eci386!clewis or lsuc!clewis)