14 Mar
2013
14 Mar
'13
11:03 a.m.
I'm running SimH V3.9 on a PC under Windows 7 using the binaries from
the primary SimH website.
I'm running the OS8 distribution on floppy that is packaged on the Simh
site.
I attached an 'empty' RK disk drive file as RK0, and tried to run RKLFMT
on it to "format" the drive.
The RKLFMT program asks its questions (which drives to format), then
says "ARE YOU SURE?" and expects a Y or N. If N is typed, the questions
about which drives to format are repeated. If Y is typed, the
formatting process begins.
On real PDP8/e hardware, it all does what it is supposed to...the drive
does a bunch of writes, then goes back and reads the sectors to verify
that they were written correctly, then prints out a message indicating
that the format passes are complete, and then starts over with the
"which disks to format" question, at which point you can type ^C to
exit.
However, in SimH, the program starts up and asks which drives to format,
and I say "Y" to drive 0, and "N" to the other 7 drives, then the
"ARE
YOU SURE?" prompt comes up, and I type "Y", and the Y echoes back, and
then it just sits there.
On real hardware, it takes about 80 seconds to format a drive. I waited
MUCH longer than this to see if the simulated RKLFMT would eventually
come back indicating that it had completed. It didn't.
I used ^E to interrupt the execution, and stepped through a few
instructions, and it appears to be hung in a loop waiting for the disk
drive to do something. I checked to see if the disk file holding the
emulated RK05 drive had changed in size, and it was still at 0 bytes.
The RKLFMT program uses the RK8E disk controller's "WRITE ALL" and
"READ
ALL" functions to format and verify the drive. These RK8E commands
ignore header information on the sectors, and just writes/reads the
data, as opposed to the normal READ and WRITE commands that pay
attention to the sector header information.
I am wondering if perhaps the SimH emulation of the RK8E is flawed in
some way such that the READ ALL and WRITE ALL commands don't work
properly, causing RKLFMT to fail.
I can get the emulated RK05 to work by simply issuing a "ZERO RKA0:"
command to OS8, which writes the directory information on the disk,
making it accessible to OS8.
Has anyone run into this before? It isn't a big deal in terms of
usability of emulated RK05s in the PDP 8 SimH implementation, but I'm
wondering if it might be a technical detail in the emulation of the
RK8E that isn't quite correct.
I'd be interesting in hearing anything that folks may have run into or
heard about this.
Thanks,
Rick Bensene