Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!usc!samsung!xylogics!bu.edu!snorkelwacker.mit.edu!ai-lab!rice-chex!bson
From: bson@rice-chex.ai.mit.edu (Jan Brittenson)
Newsgroups: comp.sys.handhelds
Subject: Re: Some HP-48 Internals Answers
Message-ID: <13652@life.ai.mit.edu>
Date: 4 Mar 91 05:44:18 GMT
References: <13605@life.ai.mit.edu> <27d099e0:2282.1comp.sys.handhelds;1@hpcvbbs.UUCP>
Sender: news@ai.mit.edu
Organization: nil
Lines: 61

In a posting of [3 Mar 91 06:40:04 GMT]
   akcs.dnickel@hpcvbbs.UUCP (Derek S. Nickel) writes:

 > The value 028FC is a "prolog signature", not an address of some code.
 > Try unassembling (not unthreading) at, for example, 02911 and you will
 > see that the fist two instructions (D=D-1 and some HST stuff) are
 > unusual", if fact the HST stuff is a NOP.  But those first five
 > nibbles are used by other parts of the OS to identify a prolog object.

This is what it looks like:

028fc   Type:
028fc 31a0     move.p2	#a, c
02900 b6a      sub.b	c, a
02903 808c     jump.a	@a
02907 82c20    data.a	span10nib	; ; #2c28
0290c e8030    data.a	L_0308e
02911   type_Short:
02911 cf820    TYPE
02916 487      brcs	L_0298f
02919 1c4      sub.a	5, d1
0291c 141      move.a	a, @d1
0291f 142      move.a	@d0, a
02922 164      add.a	5, d0
02925 808c     jump.a	@a
02929 94c20    data.a	span21nib	; ; #2c49
0292e f7030    data.a	L_0307f
02933   type_Real:
02933 cf820    TYPE
02938 465      brcs	L_0298f
0293b 1c4      sub.a	5, d1
0293e 141      move.a	a, @d1
02941 142      move.a	@d0, a
02944 164      add.a	5, d0
02947 808c     jump.a	@a

   Where is the "D=D-1 and some HST stuff"? (D=D-1 f = DEC.f D; "f" is
the field)

   The prefix at #2911 is #28fc. If A contains #2911, then the
threading instruction "JUMP.A @A" will jump to #28fc. At #28fc, 10
(decimal) will get subtracted from A, after which it will contain
#2911-10 = #2907. The "JUMP.A @A" then jumps indirect #2907, which is
#2c28, where a spanning routine for 10-nybble objects resides. D0
(which points to wherever the #2911 was found, i.e. where the object
is) is advanced by 10 past the 10-nybble object. Similarly, the ML
code at #2c48 advances D0 by 21, which is the size of a real.


 > Is this clear (I fear its not).  My point is that 028FC is not an
 > address but actual machine code.  (unassembling at address 028FC only
 > leads to madness - trust me - I tried it before I realized what was
 > going on).

- Mr. Spock, any madness readings?
- No Sir. Just plain logic.

						-- Jan Brittenson
						   bson@ai.mit.edu

					   Read my lisp: no new classes!