19 Aug
2006
19 Aug
'06
4:04 p.m.
Thanks for all the info -- it'll definitely come in handy. I now have a
complete image of the PERQ's hard disk (available on my website as well)
I am not sure how much of the disk structure you can directly get from
the POS system calls. For example, there are 4 words of filesystem
pointers in the sector _header_, that's why the DMA controller is
somewhat complicated (it can be set up to transfer those pointers to one
buffer and the sector data to another part of memory altogether). My
guess is you can't directly get those header words using POS calls.
It may not matter. From what I rmemberm the infromation is stored
elsewhere too (probably in the file descriptor blocks). THe pointers in
the headers are used for speed and to help recover the filesystem in the
event of disk problems.
and I'm working on some software to allow me to
traverse the
partitions/filesystems in the image to extract software. Figure that
could come in handy at some point.
My experience of the POS documentation is that (a) it was written to
refresh the memory of somebody who already knew the machine inside out
and (b) important information is missing -- things like the _exact_
meaning of the bits in some descriptor word, etc.
Do you have the POS source code? I thin I hace most of the Pascal (not
microcode, though) sources for one version.
I wish modern hard drives lasted as long as this thing has -- 25+ years
old and only a handful of bad sectors...
Which drive do you have? The 14" SA4000 unit used in the 1 and 1a seems
to suffer from drive belt trouble and little else. The 8" Micropolis 1203
used in the 2T1 has problems with the servo analogue electronics -- I've
had to replace op-amps and the power amplifier chip in those. The 5.25"
units used in the 2T2 and 2T4 suffer from the same problems as PC hard
drives of the time, because that's what they are :-)
-tony