PDP-11/45 RSTS/E boot problem

Noel Chiappa jnc at mercury.lcs.mit.edu
Thu Feb 14 17:46:53 CST 2019


    > From: Paul Koning

    > or a backup team of subsystem experts at the home office to call on.

Actually, the actual hardware problem wasn't too hard for Fritz to find, I
gather, once we knew exactly was failing (the RK11), and how (at 0170000, the
XM incremented). It's not like it was a comes-and-goes kind of problem (it
was quite solid and repeatable), or anything like that, so I'm not sure a
'team of experts' would really have helped.

And the exact details on the failure were also pretty easy to find, once we
got past a lot of other distractions! Once it became clear that the pure text
was getting trashed, it was pretty easy (modulo some confusion caused by the
differing OS images in the 'Ritchie' and 'Wellsch' distros :-) to stop the
machine once it had a) assembled the pure text in main memory, and b) swapped
it out.

Looking at the copy in main memory verified it was good _before_ it was
swapped, looking at the arguments on the stack gave us the disk block it was
written to, and looking at the actual contents of the RK (which PDP11GUI has a
mode to do) showed us the first 02400 bytes were good, and then it was
trash. Bingo!

Then, looking at the RK registers after the transfer had completed showed
something had gone wrong in the hardware - the address left showing
afterward was not the start_address+size - the XM had incremented
inappropriately! And since the start address for the transfer in physical
memory was 0165400, and 0165400+2400 = 0170000, that sounded pretty
suspicious!

So, that all happened pretty quickly. Really, the part that took the longest
was getting past all the confusing noise: my confusion about R5, the
malfunctioning front panel, the different versions of the OS, etc, etc.

	Noel


More information about the cctalk mailing list