Performed by Jim Leonard (trixter@oldskool.org) using IDA on 20191220

This disassembly/listing is incomplete, so I've included the IDA file if you
want to continue it.

This BIOS, like the Deskpro before it, was famously created using a legal
"clean room" technique.  Some interesting notes found during the disassembly of
the BIOS:

- It appears to delay in a few places by different fixed loop amounts based on
whether the system is running in turbo mode or not.  I would have expected it
to delay using the timer, which would have saved some code and ran the same
regardless of system speed.

- Various routines are spaced apart from each other by "FF" bytes to ensure
various entry points are the same as in IBM's BIOS.

- Many routines don't CALL a subroutine, but instead do this:

  mov bp, offset _continue
  jmp subroutine
_continue:
  ...

subroutine:
  ...
  jmp bp

This is slower and larger than a traditional CALL; RET structure.  I think it
was done in an effort to be more "legal" during the clean-room
reverse-engineering process.  While that might fool a judge, it is a very thin
ruse that doesn't fool any progammer :-)

- Unlike most clone BIOSes that came after it, this BIOS *does* intercept the
ROM BASIC interrupt and attempts to handle it by prompting for a system disk
with BASICA.EXE, and attempts to load it.  It does so with DOS commands, which
is weird to see inside a BIOS, and also means that this functionality only
works while DOS is loaded.

- The string "JSF+PNA·CHL" in the BIOS matches similar strings found in other
Compaq BIOSes... my guess is these are the initials of three people who
worked on it, but I could be wrong.