Subject: Re: RFC: g/c fuibyte(),fuiword(),suibyte(),suiword() ?
To: Jaromir Dolecek <jdolecek@NetBSD.org>
From: Robert Elz <kre@munnari.OZ.AU>
List: tech-kern
Date: 07/28/2003 10:19:03
    Date:        Sun, 27 Jul 2003 16:43:47 +0200 (CEST)
    From:        Jaromir Dolecek <jdolecek@NetBSD.org>
    Message-ID:  <200307271443.h6REhlNc026319@s102-n054.tele2.cz>

  | If they should be kept, can someone provide me with information
  | what those functions are supposed to do, so that the fetch(9)
  | and store(9) manpages could be updated?

No comment on whether they're still needed or not (but if fuword()
et al, still are, then fuiword() really should be retained too, and
used correctly).

I was around that far back (though never used PDP-11's much), and cgd
is completely correct - on (the right kind of) PDP-11, an address
meant either of two differnet locations (in the same address space)
depending upon whether it was being used for an instruction fetch, or
some other kind of use.   If you wanted to alter the instruction that
would be fetched from address X, you needed to use suiword() (or
fuiword() to get it) - otherwise you'd operate on the data which was
at a quite different location with the same address.

The joys of 16 bit address spaces...

kre