19 Apr
2006
19 Apr
'06
10:16 p.m.
It looks like I'm acquiring more bus and tag (s/3x0 parallel channel)
interface peripherals, and got to thinking... it'd be fairly neat if I
could hook a peripheral up to something, like say, a PC, and get it to
talk to Hercules.
It's not terribly difficult to find a cable pinout, but I haven't yet
been able to find a description of the protocol used on the channel
cables (or even what voltage spec, etc, they have).
Does anyone have a copy of this they could make available? Or, is it
already available somewhere that my searching hasn't found? If I had
lots (more) spare time, I might think about reverse engineering
something, but I'd rather start with a spec of what things are supposed
to look like.
Pat
--
Purdue University ITAP/RCAC --- http://www.rcac.purdue.edu/
The Computer Refuge --- http://computer-refuge.org
20 Apr
20 Apr
1:53 a.m.
On Wed, 2006-04-19 at 22:16 -0400, Patrick Finnegan wrote:
Does anyone have a copy of this they could make
available? Or, is it
already available somewhere that my searching hasn't found? If I had
lots (more) spare time, I might think about reverse engineering
something, but I'd rather start with a spec of what things are supposed
to look like.
Have a look at the document ibm/360/A22-6843-3_360channelOEM on
Bitsavers, I think it should tell you most of what you need to know.
--
Lawrence Wilkinson lawrence at ljw.me.uk
The IBM 360/30 page http://www.ljw.me.uk/ibm360
8:34 a.m.
Patrick Finnegan wrote:
It looks like I'm acquiring more bus and tag
(s/3x0 parallel channel)
interface peripherals, and got to thinking... it'd be fairly neat if I
could hook a peripheral up to something, like say, a PC, and get it to
talk to Hercules.
It's not terribly difficult to find a cable pinout, but I haven't yet
been able to find a description of the protocol used on the channel
cables (or even what voltage spec, etc, they have).
Does anyone have a copy of this they could make available? Or, is it
already available somewhere that my searching hasn't found? If I had
lots (more) spare time, I might think about reverse engineering
something, but I'd rather start with a spec of what things are supposed
to look like.
You do know that the ARTIC960 parallel channel adapters Just Work(tm) in
AIX, right?
Peace... Sridhar
12:01 p.m.
Sridhar Ayengar wrote:
Patrick Finnegan wrote:
<snip>
You do know that the ARTIC960 parallel channel adapters Just Work(tm)
in AIX, right?
Peace... Sridhar
are the for hooking an aix box as a device to the mainframe, or vice versa?
I believe there was discussion on the Hercules group, and there was no
support for
running Bus/Tag devices from AIX. It is meant to allow an AIX box to do
something
like support operations as a communications controller or other to the
Mainframe.
I believe that even IBM has put in dual interfaces on devices to allow
them to be
accessed from multiple hosts, such as tape libraries haveing both Bus /
Tag for the
mainframe (or Escon) as well as being scsi targets to the AIX boxes, etc.
Bus/Tag is only one way, there is no way to have a device pretend to be a
Host in the hookup, as far as the protocols and handshakes are concerned.
That is why channel to channel adapters take 4 sets of hoses to do a
connection
between two mainframe, for example. Two (Bus/Tag) each way.
Jim
3:09 p.m.
Jay West wrote:
It was written....
But the P/390 does. And if you need more horsepower than can be
provided by a P/390E, then you can set up an emulated mainframe with
Hercules addressing the peripherals attached to the P/390 over the network.
Peace... Sridhar
are the for hooking an aix box as a device to the
mainframe, or vice
versa?
Well, obviously the idea is to be able to interface b&t connected
devices into hercules.... and since generic clone PC's running hercules
wouldn't support the ARTIC card..... 
3:55 p.m.
Sridhar wrote....
But the P/390 does. And if you need more horsepower
than can be provided
by a P/390E, then you can set up an emulated mainframe with Hercules
addressing the peripherals attached to the P/390 over the network.
I'm confused. I thought the P/390 didn't have any bus & tag connectors on
it.... in addition, it's a microchannel card right? Making it less likely to
run in a common PC?
I just skimmed some docs on the P/390 and don't see any way that bus & tag
devices hook up to it?
Jay
4:02 p.m.
Jay West wrote:
Sridhar wrote....
You put a P/390 on the same bus (PCI or MCA) as an ARTIC960 card and the
P/390 can busmaster and take over the ARTIC960 card. The ARTIC960 then
appears as a channel in the mainframe.
Peace... Sridhar
But the P/390 does. And if you need more
horsepower than can be
provided by a P/390E, then you can set up an emulated mainframe with
Hercules addressing the peripherals attached to the P/390 over the
network.
I'm confused. I thought the P/390 didn't have any bus & tag connectors
on it.... in addition, it's a microchannel card right? Making it less
likely to run in a common PC?
I just skimmed some docs on the P/390 and don't see any way that bus &
tag devices hook up to it?
4:47 p.m.
Sridhar Ayengar wrote:
Jay West wrote:
I am not at all familiar with the RTIC libraries, but what I have, I
would not
infer that the libraries can do what the P/390's LIC can do. I remember
discussions with Marty Ziskind at the first show that they came to and they
not only cooked the code for what they did with the I/O for channels,
etc, but also ran their own specialized Token ring code libraries, to allow
them to use unused token ring features for their CTCA emulated between
P/370's ( later P/390 ) (no I'm not talking about the AT/370, but the first
cards were not 390).
So if those libraries have both ends of the hose available, all I'm
saying is
that it would be nice to know where that is, just to see what it does. On
the Hercules group discussion, this was put down as not feasible for reasons
I will have to sift thru a lot of postings there to find. The only hardware
solution was the third party one used by Flex, I think.
Again, this assumes that you want to write your own code and use the
hardware like this. That turns out to be very different than what IBM
can do with the same hardware and their software on the same pile a
lot of times.
Jim
Sridhar wrote....
You put a P/390 on the same bus (PCI or MCA) as an ARTIC960 card and
the P/390 can busmaster and take over the ARTIC960 card. The ARTIC960
then appears as a channel in the mainframe.
5:57 p.m.
jim stephens wrote:
The way the ARTIC960 cards and the P/390 LIC code work are very
different. ARTIC960 isn't simply a library. It's a card with an i960
running custom firmware to emulate a complete channel processor.
You put a
P/390 on the same bus (PCI or MCA) as an ARTIC960 card and
the P/390 can busmaster and take over the ARTIC960 card. The ARTIC960
then appears as a channel in the mainframe.
I am not at all familiar with the RTIC libraries, but what I have, I
would not
infer that the libraries can do what the P/390's LIC can do. I remember
discussions with Marty Ziskind at the first show that they came to and they
not only cooked the code for what they did with the I/O for channels,
etc, but also ran their own specialized Token ring code libraries, to allow
them to use unused token ring features for their CTCA emulated between
P/370's ( later P/390 ) (no I'm not talking about the AT/370, but the
first
cards were not 390). Again, this assumes that you want to write your own
code and use the
hardware like this. That turns out to be very different than what IBM
can do with the same hardware and their software on the same pile a
lot of times.
I know for a fact that IBM's off-the-shelf code can use an ARTIC960
Parallel Channel to talk to a channel-attached peripheral, because I've
used it for that purpose. Both at home and at work.
The ARTIC960 Parallel Channel Adapter was even sold specifically for
that purpose in a combination package with the P/390 PCI.
Peace... Sridhar
4:03 p.m.
Jay West wrote:
Sridhar wrote....
And it comes in PCI too.
Peace... Sridhar
But the P/390 does. And if you need more
horsepower than can be
provided by a P/390E, then you can set up an emulated mainframe with
Hercules addressing the peripherals attached to the P/390 over the
network.
I'm confused. I thought the P/390 didn't have any bus & tag connectors
on it.... in addition, it's a microchannel card right? Making it less
likely to run in a common PC? 
4:14 p.m.
I'm confused. I thought the P/390 didn't have
any bus & tag connectors on
it.... in addition, it's a microchannel card right? Making it less likely to
run in a common PC?
I just skimmed some docs on the P/390 and don't see any way that bus & tag
devices hook up to it?
On a related note - anyone have the docs for the card that fits inside an
AT to make it look like a Bus and Tag device (namely, the 8232)? Might be
a way to do cheap and cheesy emulation of things using crappy PeeCees.
William Donzelli
aw288 at osfn.org
3:08 p.m.
jim stephens wrote:
Sridhar Ayengar wrote:
That's not the only configuration that works. If you're running
mainframe OS on a P/390, the P/390 can address the ARTIC960 as a
parallel channel inside the OS.
Patrick Finnegan wrote:
<snip>
You do know that the ARTIC960 parallel channel adapters Just Work(tm)
in AIX, right?
Peace... Sridhar
are the for hooking an aix box as a device to the mainframe, or vice versa?
I believe there was discussion on the Hercules group, and there was no
support for running Bus/Tag devices from AIX. It is meant to allow an AIX box to do
something like support operations as a communications controller or other to the
Mainframe. I believe that even IBM has put in dual interfaces on
devices to allow
them to be accessed from multiple hosts, such as tape libraries haveing both Bus /
Tag for the mainframe (or Escon) as well as being scsi targets to the AIX boxes, etc.
There are plenty of devices like that. On the other hand, there are
plenty of devices that aren't.
Peace... Sridhar
5:19 p.m.
On Thursday 20 April 2006 15:08, Sridhar Ayengar wrote:
jim stephens wrote:
If I either the P/390 or ARTIC960 were easily/cheaply available, I might
consider this. But, I would rather create something that can be
cheaply duplicated, and not be fixed to a host hardware platform.
I'd rather run Hercules on a Linux PC (or some UNIX box), or use a real
S/390 (or ES/9000) that I've got than bother with acquiring an
expensive, slow, and hard to find P/390 or R/390.
Pat
--
Purdue University Research Computing --- http://www.rcac.purdue.edu/
The Computer Refuge --- http://computer-refuge.org
Sridhar Ayengar wrote:
That's not the only configuration that works. If you're running
mainframe OS on a P/390, the P/390 can address the ARTIC960 as a
parallel channel inside the OS. Patrick Finnegan wrote:
<snip>
You do know that the ARTIC960 parallel channel adapters Just
Work(tm) in AIX, right?
Peace... Sridhar
are the for hooking an aix box as a device to the mainframe, or
vice versa?
I believe there was discussion on the Hercules group, and there was
no support for running Bus/Tag devices from AIX. It is meant to
allow an AIX box to do something like support operations as a
communications controller or other to the Mainframe.
5:37 p.m.
Patrick Finnegan wrote:
If I either the P/390 or ARTIC960 were easily/cheaply
available, I might
consider this. But, I would rather create something that can be
cheaply duplicated, and not be fixed to a host hardware platform.
The ARTIC960 is definitely available cheaply and easily. I once bought
a box of about fifty. I sold them off piece by piece. And I know
they're still available.
The P/390 hardware/software combination can probably used to leverage
support for the ARTIC960 in Hercules.
I'd rather run Hercules on a Linux PC (or some
UNIX box), or use a real
S/390 (or ES/9000) that I've got than bother with acquiring an
expensive, slow, and hard to find P/390 or R/390.
It's actually pretty surprising the speed you can get with a P/390E card
on some of the harder-to-emulate operations that the P/390E does in
hardware.
In any case, if you feel the compulsion to write support for the
ARTIC960 in Hercules, I'd be happy to lend you a P/390 or two (and
software) to speed you on your way.
Peace... Sridhar
6:12 p.m.
Sridhar Ayengar wrote:
Patrick Finnegan wrote:
<snip>
for what it's worth, the P/390 only ever ran tape drives and only drove
selector channel protocol. It never ran any DASD due to protocol
limitations, and the fact that it was too slow to be useful, when emulated
storage thru the token ring ran faster.
My friend who has both versions only remembered seeing a version
of the Host version of the cards and cables for the Micro Channel
version, and not for the PCI. Univac supported the host version on
the PCI card set.
IBM systems with P/390's were more likely to be in Escon farms than
in bus / tag. The R/390's may have been more likely to be in sites with
SSA available, which IBM was pushing at the time the P/390's were
in vogue. Therefore less likely to see Escon versions of that hardware
or the Bus/Tag versions.
I don't think this is a bad idea, but I don't think that the solution exists
in a form that can be bought, even if IBM had it in some product.
Jim
6:21 p.m.
jim stephens wrote:
My friend who has both versions only remembered seeing
a version
of the Host version of the cards and cables for the Micro Channel
version, and not for the PCI. Univac supported the host version on
the PCI card set.
I have the PCI version, so I know it exists, but yes, the MCA version
was a lot more common.
IBM systems with P/390's were more likely to be in
Escon farms than
in bus / tag. The R/390's may have been more likely to be in sites with
SSA available, which IBM was pushing at the time the P/390's were
in vogue. Therefore less likely to see Escon versions of that hardware
or the Bus/Tag versions.
If memory serves, the ARTIC960-based ESCON adapter still goes for $big.
They're popular for adding ESCON channels above capacity in the
MP/3000, in shops that can't afford to trade in their MP/3000's for
z800's or z890's.
Peace... Sridhar
11:16 p.m.
On Thursday 20 April 2006 17:37, Sridhar Ayengar wrote:
It sounds tempting, but I think I'd rather make something new, than use
something that's not being produced anymore.
Pat
--
Purdue University ITAP/RCAC --- http://www.rcac.purdue.edu/
The Computer Refuge --- http://computer-refuge.org
Patrick Finnegan wrote:
So any idea where I can find one? I haven't ever seen one before, but I
didn't know they existed either. Do they have some specific name or
model #, other than "artic960"?
If I either the P/390 or ARTIC960 were
easily/cheaply available, I
might consider this. But, I would rather create something that can
be cheaply duplicated, and not be fixed to a host hardware
platform.
The ARTIC960 is definitely available cheaply and easily. I once
bought a box of about fifty. I sold them off piece by piece. And I
know they're still available. The P/390 hardware/software combination can probably
used to leverage
support for the ARTIC960 in Hercules.
I'm not sure how easy it'll be to figure out how the P/390 drives the
ARTIC960 card, without something like a PCI bus analyzer... I'd rather
build something like this from specs if I can, instead of trying to
reverse engineer it.
I'd rather
run Hercules on a Linux PC (or some UNIX box), or use a
real S/390 (or ES/9000) that I've got than bother with acquiring an
expensive, slow, and hard to find P/390 or R/390.
It's actually pretty surprising the speed you can get with a P/390E
card on some of the harder-to-emulate operations that the P/390E does
in hardware.
In any case, if you feel the compulsion to write support for the
ARTIC960 in Hercules, I'd be happy to lend you a P/390 or two (and
software) to speed you on your way.
4:19 p.m.
New subject: IBM Bus & Tag channel protocol specs? (more begging)
That is why channel to channel adapters take 4 sets of
hoses to do a
connection
between two mainframe, for example. Two (Bus/Tag) each way.
I suppose it is a good time for me to beg for Bus and Tag cables, if
anyone has any extras. Sometimes they can be found buried in computer
room raised floors, because nobody wanted to fish them out.
William Donzelli
aw288 at osfn.org
6668
days inactive
6669
days old
18 comments
6 participants
participants (6)
-
aw288＠osfn.org -
jwest＠classiccmp.org -
jwstephens＠msm.umr.edu -
ljw-cctech＠ljw.me.uk -
pat＠computer-refuge.org -
ploopster＠gmail.com