Questions on HD: extensions for RT-11?

From: Jerome Fine <jhfine_at_idirect.com>
Date: Sat Jan 1 10:47:11 2000

>Zane H. Healy wrote:

> Happy 0-|99 to all!!! I might be working, but I'm working from home, so
> just watched my PDP-11/73 roll over about 7 minutes ago, which means the
> first Y2K bug I've seen is it's date function :^(

Jerome Fine replies:

Why were you beside your PDP-11/73 at midnight? Yes, I remember -
there are great at night to cuddle up with (humour). Let's not get back to
that thread!!

Which version of RT-11? I had thought I would get enough time to make Y2K
patches for a number of different versions, but there was never enough interest
and too little time based on the lack of interest. Maybe now that there is actually
a possible use, there may be more interest.

I just spent the last few days making some kludges to the HD: device driver for E11
to allow UNIT=0 to have 8 partitions (0 => 7) in a BIGFILE.DSK (256 MBytes)
on a hard drive. Not really much use at this stage, but the first step on the
road to allowing a full mapping table with 64 partitions. Now that the technical
details are proven, I thought I would ask this list for their thoughts regarding
the range of variables.

(a) Should the mapping table allow for "n" controllers? I thought a maximum of 4?

(b) Should the mapping table allow for "u" units per controller? I thought a
maximum of 4 different units (not the actual unit numbers)?

(c) How many different partitions should the mapping table allow per drive?
I thought a maximum of 4096 (only 64 could be active at one time - this
allows for a drive up to 128 GBytes - a larger drive would be allowed if the
number of controllers and/or different units per controller is reduced - up to
65536 partitions or 2 TBytes for one controller and one unit - would 128 GBytes
be enough for the next 10 years - or more to the POINT can anyone ever keep
track of 4096 partitions - Tim how do you manage with 256 partitions on that
9 GByte SCSI drive?).

(d) Should all 64 partitions from H00: to H77: be bootable?

(e) Is as small an HDX.SYS as possible essential - i.e. place as much code
and tables as possible into high memory?

(f) Is a "SET HD: SHOW" command essential and is an enhancement to
RESORC to initiate that command in a way that is transparent to the user
(after "SHOW DEVICE:HD:") also considered essential?

NOTE: While DU(X).SYS is used as a comparison, I am discussing the features that
would be in HD(X).SYS for a version of E11 which allows more than 32 MBytes of
hard drive disk space to be mounted. If this discussion results in these features also
being added (eventually) to DU(X).SYS, that aspect can be looked at later.

Note that DU(X).SYS allows up to four controllers. That seems a reasonable maximum.

Note that DU(X).SYS allows up to 256 different units from 0 to 255. Does any
controller that anyone knows about (except for host adapters) allow any more than
four units per controller? And is four drives a reasonable maximum even for a
host adapter? The limit of four drives per controller is not meant to say that the
UNIT numbers would be limited to the range 0 to 3. A SET command would
provide an additional mapping to allow any UNIT value from 0 to 65535, such as:
"SET HD: CSR3=160334,UNIT32=8192."
where "UNITpu" is used as UNIT32 for PORT=3,UNIT=2
If anyone has a better suggestion, it would be appreciated.
Thereafter:
"SET H30: PORT=3,UNIT=2,PART=1024."
"SET H34: PORT=3,UNIT=2,PART=1028."
"SET H37: PORT=3,UNIT=2,PART=1031."
might be what one user would require - assuming that a drive of the
appropriate size (about 40 GBytes or larger) was available.

It would be appreciated if anyone who uses V5.5, V5.7 or V5.7 of RT-11
(or will in the future) cares to comment on these questions? If your comments
are even on just one question, that would be helpful.

HAPPY NEW YEAR TO EVERYONE ON THE LIST!

I hope that the actual start of the millennium next year in 2001 will be
just as smooth.

Sincerely yours,

Jerome Fine
Received on Sat Jan 01 2000 - 10:47:11 GMT