25 Oct
2012
25 Oct
'12
9:14 p.m.
Message: 5
Date: Thu, 25 Oct 2012 08:17:50 -0400
From: Rick Murphy <rick at rickmurphy.net>
To: <cctech at classiccmp.org>
Subject: Re: Microvax II won't boot, hardware error?
Message-ID: <201210251217.q9PCHosl030262 at rickmurphy.net>
Content-Type: text/plain; charset="us-ascii"; format=flowed
At 06:39 PM 10/24/2012, Kevin Reynolds wrote:
Here's what that machine check stack frame means:
803AF1E8 0000000C 12 bytes of parameters
803AF1EC 00000082 Machine check code = Write bus
error, virtual address
803AF1F0 7FF38058 Virtual address
803AF1F4 02000002 Internal CPU state.
803AF1F8 00001534 PC
803AF1FC 03C00001 PSL
It seems that you are getting hard memory write faults on the stack.
Maybe something is hanging up the Q-bus, or you have failing memory. If
you have diagnostics, running a memory test for a while may get you
somewhere. Otherwise, I'd look into trying to pinpoint the problem by
removing memory modules and see if the system becomes more stable
(assuming you have multiple memory cards, of course).
-Rick
Memory is NOT on the Q-bus, it has it's own dedicated bus on
the C-D connector rows (how many slots are memory only in the
CD rows depends on the specific backplane.) Possibly you
might have a Q-bus board in a memory-only slot, that is certainly
something to check. Otherwise, you may have a bad memory card,
a bad or dirty slot, or possibly a bad CPU board.
Hmm, I've never seen this error, but I think if a peripheral
board is bad, it does not cause a machine check, but maybe I've
never had the specific error. Can you tell what the physical
address was? or, at this point, might virtual=physical?
If so, is 7FF3xxxx in the Q-bus I/O space?
Jon
.... A crash
example is below:
**** FATAL BUG CHECK, VERSION = V5.2 MACHINECHK, Machine check
while in kernel mode
...
KERNEL/INTERRUPT/BOOT STACK
803AF1E8 0000000C
803AF1EC 00000082
803AF1F0 7FF38058
803AF1F4 02000002
803AF1F8 00001534
803AF1FC 03C00001
25 Oct
25 Oct
9:48 p.m.
At 09:14 PM 10/25/2012, Jon Elson wrote:
Memory is NOT on the Q-bus, it has it's own
dedicated bus on
the C-D connector rows (how many slots are memory only in the
CD rows depends on the specific backplane.) Possibly you
might have a Q-bus board in a memory-only slot, that is certainly
something to check. Otherwise, you may have a bad memory card,
a bad or dirty slot, or possibly a bad CPU board.
Ah yes, you're right. Microvax 2, so memory's not on the Q-bus. It's
got to be a bad memory area. Enabling windows or AUTOGEN has just moved
that bad chip to somewhere that's now being used.
Hmm, I've never seen this error, but I think if a
peripheral
board is bad, it does not cause a machine check, but maybe I've
never had the specific error. Can you tell what the physical
address was? or, at this point, might virtual=physical?
If so, is 7FF3xxxx in the Q-bus I/O space?
No, it's a virtual address. You're running with memory management
enabled by that point, so the only way to figure out what physical
address this corresponds to is to look at the page tables, which you
could get from running SDA on the resulting dump.
At this point, it's pretty likely to be a memory fault. Unless you're
going to try to find out what chip to replace, my suggestion to remove
RAM until it stops failing so the bad board can be isolated is probably
the best simple diagnostic.
Eventually, running diagnostics is probably the best thing to try. It's
probably bad RAM but could also be a bad CPU. However, my experience
with bad CPU modules is that they don't get very far along the boot
process before dying.
-Rick
10:02 p.m.
On 10/25/2012 09:14 PM, Jon Elson wrote:
Memory is NOT on the Q-bus, it has it's own
dedicated bus on
the C-D connector rows (how many slots are memory only in the
CD rows depends on the specific backplane.)
Actually the MicroVAX-II uses a PMI interconnect on a daisy-chained
ribbon cable across the tops of the memory and CPU boards, in addition
to the C-D interconnect.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
10:16 p.m.
On Oct 25, 2012, at 10:02 PM, Dave McGuire wrote:
On 10/25/2012 09:14 PM, Jon Elson wrote:
I was gonna say. The PMI interconnect for QBUS doesn't supply enough
address pins for VAX-sized memory, nor is it 32-bit, so MV boards have
their own ribbon cables as a side channel. Not knowing much about VAX
failure modes but having plenty of experiences with touchy ribbon
cables, I'd say that might be a good first bet.
- Dave
Memory is NOT on the Q-bus, it has it's own
dedicated bus on
the C-D connector rows (how many slots are memory only in the
CD rows depends on the specific backplane.)
Actually the MicroVAX-II uses a PMI interconnect on a daisy-chained
ribbon cable across the tops of the memory and CPU boards, in addition
to the C-D interconnect.
11:16 p.m.
On 10/25/2012 10:16 PM, David Riley wrote:
On Oct 25, 2012, at 10:02 PM, Dave McGuire wrote:
It's much worse for the KD32 (MicroVAX-I). Their memory is actually
on the Qbus, so not only do 32-bit transfers require two data cycles,
but Qbus is ALREADY a multiplexed bus! SSSSllllooooowwwww.
I love the KD32 anyway though. :) It was my first VAX, when I had
worshipped VAXen from afar for many years. And you know what? It
worked great. It was nowhere near as fast as my PDP-11/73, but the
additional functionality of VMS and a 32-bit CPU, and the GUI made life
so much nicer that I really didn't mind the performance hit. I still
have a MicroVAX-I...I should really get a working system disk for it and
get it running.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
On 10/25/2012 09:14 PM, Jon Elson wrote:
I was gonna say. The PMI interconnect for QBUS doesn't supply enough
address pins for VAX-sized memory, nor is it 32-bit, so MV boards have
their own ribbon cables as a side channel. Memory is NOT on the Q-bus, it has it's own
dedicated bus on
the C-D connector rows (how many slots are memory only in the
CD rows depends on the specific backplane.)
Actually the MicroVAX-II uses a PMI interconnect on a daisy-chained
ribbon cable across the tops of the memory and CPU boards, in addition
to the C-D interconnect. 
26 Oct
26 Oct
4:28 p.m.
On Thu, Oct 25, 2012 at 8:16 PM, Dave McGuire <mcguire at neurotica.com> wrote:
It's much worse for the KD32 (MicroVAX-I). Their memory is actually
on the Qbus, so not only do 32-bit transfers require two data cycles,
but Qbus is ALREADY a multiplexed bus! SSSSllllooooowwwww.
I love the KD32 anyway though. :) It was my first VAX, when I had
worshipped VAXen from afar for many years. And you know what? It
worked great. It was nowhere near as fast as my PDP-11/73, but the
additional functionality of VMS and a 32-bit CPU, and the GUI made life
so much nicer that I really didn't mind the performance hit. I still
have a MicroVAX-I...I should really get a working system disk for it and
get it running.
-Dave
I also have a MicroVAX I (Currently housing a KA650, but I still have the
original boardset). It would be fun to get it running VMS -- the MicroVAX
I tops out at VMS 5.1, though. Where can one find media (and licenses) for
older versions of VMS?
- Josh
5:14 p.m.
I also have a MicroVAX I (Currently housing a KA650,
but I still have the
original boardset). It would be fun to get it running VMS -- the MicroVAX
I tops out at VMS 5.1, though. Where can one find media (and licenses) for
older versions of VMS?
- Josh
The oldest VMS I could find is 5.2. Original licensing is quite troublesome to
deal with (trust me, I know). The hobbyist licenses should work though.
Kevin
5:46 p.m.
On 10/26/2012 04:28 PM, Josh Dersch wrote:
I also have a MicroVAX I (Currently housing a KA650,
but I still have the
original boardset). It would be fun to get it running VMS -- the MicroVAX
I tops out at VMS 5.1, though. Where can one find media (and licenses) for
older versions of VMS?
I, umm, may have some. =)
Note that LMF was introduced with 5.0, so before that you don't have
to worry about it.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
28 Oct
28 Oct
1:28 a.m.
On 10/26/2012 2:46 PM, Dave McGuire wrote:
On 10/26/2012 04:28 PM, Josh Dersch wrote:
I, er, would love to know more about this :).
Thanks,
Josh
I also have a MicroVAX I (Currently housing a
KA650, but I still have the
original boardset). It would be fun to get it running VMS -- the MicroVAX
I tops out at VMS 5.1, though. Where can one find media (and licenses) for
older versions of VMS?
I, umm, may have some. =)
Note that LMF was introduced with 5.0, so before that you don't have
to worry about it.
-Dave
4259
days inactive
4261
days old
8 comments
6 participants
participants (6)
-
derschjo＠gmail.com -
elson＠pico-systems.com -
fraveydank＠gmail.com -
mcguire＠neurotica.com -
rick＠rickmurphy.net -
tpresence＠hotmail.com