23 Sep
2004
23 Sep
'04
5:11 p.m.
Best rule for SIMH or E11 purposes is
NOT to mix "RM" and "RP" drives on a single RH11/RH70.
--
There was some discussion/questions from Bob Supnik about this
on the simh developers list this week if mixed drives were or
were not supported on the PDP11 operating systems. It sounded
like they were not, so I'm surprised that they may have been
on V9 and earlier.
The problem comes about because the RM and RP massbus controllers
have slightly different register sets.
23 Sep
23 Sep
6:42 p.m.
>>>> "Al" == Al Kossow
<aek(a)spies.com> writes:
Al> Best rule for SIMH or E11 purposes is NOT to mix "RM" and
"RP"
Al> drives on a single RH11/RH70.
Al> --
Al> There was some discussion/questions from Bob Supnik about this on
Al> the simh developers list this week if mixed drives were or were
Al> not supported on the PDP11 operating systems. It sounded like
Al> they were not, so I'm surprised that they may have been on V9 and
Al> earlier.
Al> The problem comes about because the RM and RP massbus controllers
Al> have slightly different register sets.
No they don't.
The RH11/RH70 is the same either way. What changes is the registers
in the disks -- RP and RM disks are very similar but not identical.
Working from memory now -- I believe the correct statement is that
RSTS has always allowed mixing them on either of the two massbus
controllers, for access within the OS. The place you run into issues,
prior to RSTS 9.0, is when booting. Some vague memory says that your
boot device (unit 0) on the primary massbus should be an RP04/5/6 and
not an RM if you mix them.
If you run RSTS 9.0 or later, mix at will and boot at will, it all
works.
paul
9:04 p.m.
> Best rule for SIMH or E11 purposes is NOT to mix
"RM" and "RP"
> drives on a single RH11/RH70.
The thing I was trying to remember about mixed drives and booting is
described by this paragraph in the RSTS V9.0 Update Seminar document
(page 2-7):
Multiple RH controllers for disks (DB and DR) have
always been supported by RSTS. The support for this
has been improved in V9 by allowing the booting of
disks on either controller regardless of drive type.
Previously, there was a restriction if the same drives
were on different controllers, you could only boot the
one on the primary controller.
So the answer for RSTS is (a) mixed drives are supported, (b) if you
mix on both RP/RM massbus controllers, boot the primary unless you're
running V9.0 or later.
paul
24 Sep
24 Sep
12:04 p.m.
Paul,
While I'm willing to accept that mixed drive type strings are
possible and supported, I don't see how the quote below supports that.
At 8:04 PM -0400 9/23/04, Paul Koning wrote:
That's not what I read in the statement above. IIRC, my /70 supports
up to 4 RH70s. The statement above seems to indicate that they
removed a restriction where one must boot from drive 0 of a
particular RH70. The statement itself doesn't indicate that I can
have multiple drive types on the same RH70.
> Best rule
for SIMH or E11 purposes is NOT to mix "RM" and "RP"
> drives on a single RH11/RH70.
The thing I was trying to remember about mixed drives and booting is
described by this paragraph in the RSTS V9.0 Update Seminar document
(page 2-7):
Multiple RH controllers for disks (DB and DR) have
always been supported by RSTS. The support for this
has been improved in V9 by allowing the booting of
disks on either controller regardless of drive type.
Previously, there was a restriction if the same drives
were on different controllers, you could only boot the
one on the primary controller.
So the answer for RSTS is (a) mixed drives are supported, (b) if you
mix on both RP/RM massbus controllers, boot the primary unless you're
running V9.0 or later.
Yes, I agree, this seems to be the case.
Do you have something that explicitly states multiple drive types are
supported on a single RH70?
Thanks,
John
12:47 p.m.
>>>> "John" == John A Dundas,
<John> writes:
John> Paul, While I'm willing to accept that mixed drive type strings
John> are possible and supported, I don't see how the quote below
John> supports that.
John> At 8:04 PM -0400 9/23/04, Paul Koning wrote:
> >> Best rule for SIMH or E11 purposes is NOT
to mix "RM" and "RP"
> >> drives on a single RH11/RH70.
>
> The thing I was trying to remember about mixed drives and booting
> is described by this paragraph in the RSTS V9.0 Update Seminar
> document (page 2-7): Multiple RH controllers for disks (DB and DR)
> have always been supported by RSTS. The support for this has been
> improved in V9 by allowing the booting of disks on either
> controller regardless of drive type. Previously, there was a
> restriction if the same drives were on different controllers, you
> could only boot the one on the primary controller.
>
> So the answer for RSTS is (a) mixed drives are supported,
John> That's not what I read in the statement above. IIRC, my /70
John> supports up to 4 RH70s. The statement above seems to indicate
John> that they removed a restriction where one must boot from drive
John> 0 of a particular RH70. The statement itself doesn't indicate
John> that I can have multiple drive types on the same RH70.
Actually, the old restriction wasn't quite as severe as you state it,
but yes, it doesn't directly address my claim. That's why I said "the
thing I was trying to remember..." because I knew there was SOME
restriction relating to drive mixing and the two moving head disk RH
controllers. The text I quoted states that restriction and the fact
it was removed.
> (b) if you mix on both RP/RM massbus controllers,
boot the primary
> unless you're running V9.0 or later.
John> Yes, I agree, this seems to be the case.
John> Do you have something that explicitly states multiple drive
John> types are supported on a single RH70?
Two things: my own memory from working in RSTS development, and
reading the code. I expect it's documented in user manuals, the
sysgen manual for example, but I don't have any of those.
BTW, you're correct to say that four massbus controllers can exist in
an 11/70. At least for RSTS, the rule is that one is for tapes, one
for RS03/04, and two for the RP/RM massbus disks. The tapes massbus
could have on it any mix of massbus tapes, and the RS03/04 massbus
would accept any mix of those two drives, but mixing more than that
wasn't allowed.
Clearly it would have been possible to make the device drivers more
general, so (for example) TU16 and RP04 could live on the same
massbus, or 32 RP/RM disks could be supported, 8 on each -- but there
never was any push for that and the code doesn't work that way. (Some
vague memory says that VMS did have more flexibility in this area --
or was it TOPS-10?)
Similarly, I don't think that RSTS allowed mixing MSCP disks and TMSCP
tapes on the same controller. I don't really know that stuff -- it
may be that no PDP-11 MSCP controller existed that would do disk and
tape on the same controller anyway.
paul
1:06 p.m.
Paul,
At 11:47 AM -0400 9/24/04, Paul Koning wrote:
Thanks for the explanation. I think the code fragment and
explanation you posted to SIMH explains this best.
>>>> "John" == John A Dundas, <John> writes:
John> Paul, While I'm willing to accept that mixed drive type strings
John> are possible and supported, I don't see how the quote below
John> supports that.
John> At 8:04 PM -0400 9/23/04, Paul Koning wrote:
> >> Best rule for SIMH or E11 purposes
is NOT to mix "RM" and "RP"
> >> drives on a single RH11/RH70.
>
> The thing I was trying to remember about mixed drives and booting
> is described by this paragraph in the RSTS V9.0 Update Seminar
> document (page 2-7): Multiple RH controllers for disks (DB and DR)
> have always been supported by RSTS. The support for this has been
> improved in V9 by allowing the booting of disks on either
> controller regardless of drive type. Previously, there was a
> restriction if the same drives were on different controllers, you
> could only boot the one on the primary controller.
>
> So the answer for RSTS is (a) mixed drives are supported,
John> That's not what I read in the statement above. IIRC, my /70
John> supports up to 4 RH70s. The statement above seems to indicate
John> that they removed a restriction where one must boot from drive
John> 0 of a particular RH70. The statement itself doesn't indicate
John> that I can have multiple drive types on the same RH70.
Actually, the old restriction wasn't quite as severe as you state it,
but yes, it doesn't directly address my claim. That's why I said "the
thing I was trying to remember..." because I knew there was SOME
restriction relating to drive mixing and the two moving head disk RH
controllers. The text I quoted states that restriction and the fact
it was removed.
> (b) if you mix on both RP/RM massbus
controllers, boot the primary
> unless you're running V9.0 or later.
John> Yes, I agree, this seems to be the case.
John> Do you have something that explicitly states multiple drive
John> types are supported on a single RH70?
Two things: my own memory from working in RSTS development, and
reading the code. I expect it's documented in user manuals, the
sysgen manual for example, but I don't have any of those. BTW, you're correct to say that four massbus
controllers can exist in
an 11/70. At least for RSTS, the rule is that one is for tapes, one
for RS03/04, and two for the RP/RM massbus disks. The tapes massbus
could have on it any mix of massbus tapes, and the RS03/04 massbus
would accept any mix of those two drives, but mixing more than that
wasn't allowed.
Didn't remember those details. Thanks.
Clearly it would have been possible to make the device
drivers more
general, so (for example) TU16 and RP04 could live on the same
massbus, or 32 RP/RM disks could be supported, 8 on each -- but there
never was any push for that and the code doesn't work that way. (Some
vague memory says that VMS did have more flexibility in this area --
or was it TOPS-10?)
Do the restrictions specify _which_ RH is for tape, RS, disk? Or
just that only two for RP/RM, 1 for tape, 1 for RS and it can be any
of the four?
Similarly, I don't think that RSTS allowed mixing
MSCP disks and TMSCP
tapes on the same controller. I don't really know that stuff -- it
may be that no PDP-11 MSCP controller existed that would do disk and
tape on the same controller anyway.
I don't remember a DEC controller that did both MSCP and TMSCP
(though I'm certain someone will correct me if I'm wrong), so this
may be the motivation there. (Though almost certainly Emulex or
someone made a controller that would do both.)
Thanks,
John
1:57 p.m.
>>>> "John" == John A Dundas,
<John> writes:
John> Paul, At 11:47 AM -0400 9/24/04, Paul Koning wrote:
> BTW, you're correct to say that four massbus
controllers can exist
> in an 11/70. At least for RSTS, the rule is that one is for
> tapes, one for RS03/04, and two for the RP/RM massbus disks. The
> tapes massbus could have on it any mix of massbus tapes, and the
> RS03/04 massbus would accept any mix of those two drives, but
> mixing more than that wasn't allowed.
John> Didn't remember those details. Thanks.
> Clearly it would have been possible to make the
device drivers
> more general, so (for example) TU16 and RP04 could live on the
> same massbus, or 32 RP/RM disks could be supported, 8 on each --
> but there never was any push for that and the code doesn't work
> that way. (Some vague memory says that VMS did have more
> flexibility in this area -- or was it TOPS-10?)
John> Do the restrictions specify _which_ RH is for tape, RS, disk?
John> Or just that only two for RP/RM, 1 for tape, 1 for RS and it
John> can be any of the four?
The former. The tape RH is at 772440, the RS RH is at 772040, and
the other two (776700 and 776300) are RP/RM. That roughly matches
what the PDP11 architecture handbook shows for assignment. However,
it says that the 776300 address is "alternate RS/RP/RM/TJ when for
RSTS it's only RP/RM. It also says that there's a fifth RH11/RH70
address, 775400 for the TU78. RSTS doesn't support that address and
doesn't seem to support the TU78 (too fast for a PDP11, I think).
Instead it knows that address as a "KXJ11-CR" whatever that may be.
paul
25 Sep
25 Sep
6:41 a.m.
On Fri, 24 Sep 2004, Paul Koning wrote:
Similarly, I don't think that RSTS allowed mixing
MSCP disks and TMSCP
tapes on the same controller. I don't really know that stuff -- it
may be that no PDP-11 MSCP controller existed that would do disk and
tape on the same controller anyway.
Correct. They're functionally different,
so cant be mixed. Even
the aftermarket controllers with multiple channels (e.g. the Emulex
UC08 dual-channel SCSI-to-(T)MSCP) requires one protocol per channel.
--f
7531
days inactive
7533
days old
7 comments
4 participants
participants (4)
-
aek＠spies.com -
dundas＠caltech.edu -
pkoning＠equallogic.com -
waltje＠pdp11.nl