22 Apr
2005
22 Apr
'05
12:12 a.m.
Dont know if this is common knowledge, but I have a VS3100/76 that I threw a
2gig IBM drive in, tried to put VMS on and - it worked. All the stuff I saw
seemed to indicate that 3100s didn't like sysdisks over 1 gig, but my /76
doesn't seem to mind.
There it is if it's useful
Scott Quinn
22 Apr
22 Apr
12:49 a.m.
On 4/22/05, Saquinn624 at aol.com <Saquinn624 at aol.com> wrote:
Dont know if this is common knowledge, but I have a
VS3100/76 that I threw a
2gig IBM drive in, tried to put VMS on and - it worked. All the stuff I saw
seemed to indicate that 3100s didn't like sysdisks over 1 gig, but my /76
doesn't seem to mind.
IIRC the problem isn't _running_ VMS, it's when VMS crashes and uses
old SCSI commands to talk to the disk (which only have enough bits to
address blocks at 1GB or below), and your system disk gets scribbled
on.
I have not owned a 3100, so anyone who knows more feel free to chime
in, but as I understand it, one should put a 1GB disk as DUA0: on a
3100, but one may put larger disks on as data disks, since VMS will
not attempt to write to them when it's going down hard.
-ethan
8:55 a.m.
On Fri, 22 Apr 2005, Ethan Dicks wrote:
I have not owned a 3100, so anyone who knows more feel
free to chime
in, but as I understand it, one should put a 1GB disk as DUA0: on a
3100, but one may put larger disks on as data disks, since VMS will
not attempt to write to them when it's going down hard.
My understanding is that you can use over 1GB for system if critical files
needed during boot end up below the 1GB limit so that the firmware can
address them. You will also need to turn off crash dumps with the
appropriate sysgen parameter or always be sure your SYSDUMP.DMP or
PAGEFILE.SYS that you dump to maps completely under 1GB.
Probably most folks running a VAX aren't planning to log a support call
for their crash dumps anyway....
I've had a 9 and 18 GB disk on my 3100 and once VMS is running all is
fine.
1:14 a.m.
On Apr 21, 2005, at 10:12 PM, Saquinn624 at aol.com wrote:
Dont know if this is common knowledge, but I have a
VS3100/76 that I
threw a
2gig IBM drive in, tried to put VMS on and - it worked. All the stuff
I saw
seemed to indicate that 3100s didn't like sysdisks over 1 gig, but my
/76
doesn't seem to mind.
There it is if it's useful
Scott Quinn
This was at the time very annoying. We wanted to use what turned out
to be affected systems as boot nodes for LAN-based clusters and were
trying to use bigger disks. ("Damned disks work on Viking controllers
on the 3200's! WTF is wrong with these 3100's ?!?!?") Funny now... :-)
The actual problem was that "Early" MicroVAX 3100 and VaxStation 3100
boot ROMs use 6-byte SCSI READ and WRITE commands which limit access
to 1.06GB.
At some point VS3100-76's (and later) came fixed from the factory. I
don't remember when MicroVAX 3100's were fixed. I *think* DEC released
a fix for some models of MV3100's, but never for VS3100's.
VS4000's never had this problem.
To make life easier, a very unofficial fix was created and is still
available here: ftp://ftp.gwdg.de/pub/vms/ka420/
There's a README and the patches are in ka420-rom-patch_010.zip
Mike
2:20 p.m.
Dont know if this is common knowledge, but I have a
VS3100/76 that I
threw a
2gig IBM drive in, tried to put VMS on and - it
worked. All the stuff
I saw
seemed to indicate that 3100s didn't like sysdisks
over 1 gig, but my
/76
doesn't seem to mind.
The console disk driver wraps when addressing logical blocks
beyoned 1.073GB. So if you try to access a block at 2GB
you get the block at ~0.7GB.
The console disk driver is used early on in the boot process
(until enough of OpenVMS has been loaded that it can switch
to its own full-featured driver) and also during crash dumping
(since the console driver is in eprom, it can be relied on not
have become corrupt).
If any of the files needed during boot end up past the 1.073GB
mark, the system will not boot. Event if the files happen to lie
in accessible places initially, they might be moved by a
subsequent upgrade or disk defragmentation.
Much, much worse is the behaviour during a crash. If the
crashdump file happens to be beyond the 1.073GB mark, then
anything actually at the wrapped around address is at risk!
There are ways around this problem, but the only easy one
is to use an RZ26 (1GB disk) as the system disk. That's
not a problem at least for releases up until OpenVMS V7.1.
I stopped getting updates at that point so I don't know
how much bigger VAX releases have become.
The problem is not such an issue with *BSD since you can
effectively control what lives where.
The issue affects all VAXstation 3100s, the
MicroVAX 3100 Models 10 and 20 and the early releases
of the MicroVAX 3100 Models 10e and 20e. I believe that
a fix was produced for the 10e and 20e and ROMs dated
after April 1992 are fine. The OpenVMS FAQ has more details.
So just because it seems to work, does not make it a good idea :-)
Antonio
--
---------------
Antonio Carlini arcarlini at iee.org
7104
days inactive
7104
days old
4 comments
5 participants
participants (5)
-
a.carlini＠ntlworld.com -
ethan.dicks＠gmail.com -
mcesari＠comcast.net -
pt＠new.rr.com -
Saquinn624＠aol.com