"Jerome H. Fine" <jhfinedp3k at compsys.to> skrev:
> Hopefully, this request is clear enough to be understood. Both the
> software and the hardware portion of the questions are independently
> important, so please answer one aspect even if you can't help with
> the other.
>
> Over the past 30 years of using PDP-11 software (RT-11 over 95%)
> and hardware, I have never had occasion to use a Unibus system with
> more than 256 KB of memory (such as a PDP-11/34).
>
> I would appreciate help in understanding the Unibus Map hardware which
> (if I understand its purpose correctly) is to convert 18 bit addresses given
> to a Unibus controller into 22 bit addresses for real physical memory for
> systems like a PDP-11/84 and perform DMA from / to the hard drive.
>
> Also helpful would be an explanation of the related software used under
> RT-11 along with exactly where the Unibus Map hardware is located
> on a real DEC system (on the CPU board I presume) since the identical
> CPU board is used for both the Qbus and the Unibus with both the
> PDP-11/84 and the PDP-11/94.
I can't answer how RT-11 do things, but I can explain the hardware side of it.
First of all, the Unibus map is not in the CPU board of the 11/84 and 11/94.
It's on the Unibus adapter. KT-whatever.
The whole thing works in a rather simple way, actually. All Unibus devices do
DMA to an 18-bit address. That can never change, so think of this 256 K of
memory as a separate memory area, that don't correspond to your actual physical
memory. It's all fake.
Then divide this whole 256K into 32 equal sized pieces. That will give you 32
pieces of 8K memory ranges. The same as a page on a PDP-11, but that's not that
relevant. :-)
Now, when a device will do DMA, it will do DMA to one, or several of these 32
pages. So you'll have an 18-bit address, of which the high 5 bits tells you
which "page" it will go to. The Unibus Map have 32 registers. One for each of
these "pages". So the high 5 bits of the 18-bit DMA address will select a
register in the Unibus map. The Unibus map will in turn give you a 16-bit value
>from this. This will be the high 16 bits of a 22-bit memory address. The 13 low
bits of the DMA address is added to this 22-bit address to give the final
physical address to do DMA to.
So, in essence, what a driver will do when a Unibus map is present, is to find
some free Unibus map registers (free in the meaning that no other driver is
using them, so you need to keep track of these, as a resource). Or you could
statically allocate some map registers for your driver, if you prefer, and if
all drivers combined don't need more than 31 map registers (I think that the
last register might not be usable, but I'm not sure. It basically gives a
mapping for what otherwise is the I/O page).
Once a driver have decided which map register to use, that will decide what
18-bit address range to use for the DMA. So the 18-bit address is calculated
>from the physical address truncated, and the address for the range covered by
the map register. The map register is loaded with the corresponding physical
address high bits, and then you fire the I/O.
Johnny
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: bqt at softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
I remember the Havant IBM Disk plant.
I went on a visit there once.
It was one enormous clean room.
The drives they made had a name - something like Eagle, Sparrow or Wren
Rod
-----Original Message-----
From: cctech-bounces at classiccmp.org
[mailto:cctech-bounces at classiccmp.org] On Behalf Of Sridhar Ayengar
Sent: 08 August 2007 04:05
To: On-Topic and Off-Topic Posts
Subject: Re: IBM "Type 0667" ESDI disk information?
jim wrote:
> Sridhar Ayengar wrote:
>
>> Mr Ian Primus wrote:
>>
> Is there any country of origin on the 667? There was a division in
> Havant that manufactured drives in this time frame, that IBM sold to
> the plant managers, that is now Xyratex.
If memory serves, those series were made in several different places,
including California and Mexico. Probably one or two other places too.
Peace... Sridhar
Well, I did some checking, and my one vax does _not_
have an Emulex ESDI controller - it's a Dilog. I
downloaded the manual from Bitsavers. I'm going to
start playing with it and see if I can get it to
format some of these spare ESDI disks.
One disk I have a couple of is this IBM drive. It's
labeled as "P/N 90X8528 70MB ESDI", and on the side as
"Type: 0667". Googleing has turned up little, although
I found some info on cyls/heads/sectors. There is a
block of dip switches on the drive, and it would be
nice to figure out what they do...
-Ian
I'm going to risk the rath, but I just know someone on this list has
delt with this.
After replacing my Ryobi 14.4v nicd charger 2 times (yes 2), I discovered
I have a dead battery and it's killing the charger(s). Apparently it can
do this rather quickly, as I only left it on the charger for about an hour.
(one web site says it only takes a matter of minutes - apparently the cheapo
charger does not protect itself)
So, ok. Since I have 2 dead ones I might as well try to fix it. It just
as a simple circuit board with a rectifier, 4 devices, 3 leds and bunch
of discretes.
Q1 A1015 PNP transistor (2SA1015)
Q2 2N3904 NPN transistor
Q3 XL08 ST 40 .8A SCRS
Q4 X0403DE ST SCR
Anyone ever fix one? what fails?
I'm going to guess the big to-225 style SCR failef, since I'll bet that
modulates the current to the battery.
-brad
I have a Heath controller board, 85-2222-1. I think that's an RX01 or RX02
controller for DEC. I was using it with my Shugart 801. Anyone know what it
actually is and if it has any value?
-T
-----
205. [Philosophy] Life: a sexually transmitted disease that is fatal.
--... ...-- -.. . -. ----. --.- --.- -...
tpeters at nospam.mixcom.com (remove "nospam") N9QQB (amateur radio)
"HEY YOU" (loud shouting) WEB: http://www.mixweb.com/tpeters
43? 7' 17.2" N by 88? 6' 28.9" W, Elevation 815', Grid Square EN53wc
WAN/LAN/Telcom Analyst, Tech Writer, MCP, CCNA, Registered Linux User 385531
> Message: 11
> From: Adrian Burgess <classiccmp at discordance.org.uk>
> Subject: Re: UK Classic Computer event (Was VCF UK)
> To: "General Discussion: On-Topic and Off-Topic Posts"
>> Can we have a quick poll to see who lives where?
>
> Half way between Nottingham and Lincoln.
>
UK, Aldershot, Hampshire
Mike.
I should know this, but I can't easily find the answer...
I have a BBC B+ with (1770)DFS (V2.x)( and ADFS (V3.5x) (or at least I
think those version numbers are right). It is cabled up to a single 80
cylinder double-sided disk drive.
I wish to read some files of a 40 cylinder (DFS) disk. I seem to remember
there was a command that would cause the drive to double-step, but only
allow reading. What is it?
-tony
Hi,
I was looking through my collection yesterday and was testing out an old
Intel SDK85 system. I noticed the main processor is stamped 8085, which
means this chip is either an 8085 or the "A" was not printed properly and
the chip is really an 8085A.
I have a few 8085 systems and a box of 8085 CPUs. I checked them all and
every one of them is an 8085A. I don't think I have ever seen an 8085,
they've all been 8085A types.
Other than the stamp on the chip, is there anyway I can tell the difference
between an 8085 and an 8085A? Furthermore, is the 8085 rare, and I should
put the chip away, or is it nothing special?
Why did Intel bring out the 8085A? Were there issues with the 8085?
Seeyuzz
River
