Re: apple2-v003-linux.tar.gz

[neufeld@caliban.physics.utoronto.ca (Christopher Neufeld)]

In article <331C68E4.4D83@baldmountain.bbn.com>,
Aaron Culliney  <chernabog@baldmountain.bbn.com> wrote:
>Christian Koechli wrote:
>> 
>> I got the new file from ftp.apple.asimov.net unpacked it
>> and tried to compile it on my RedHat 4.1 (kernel 2.0.27) linux box:
>> 
>> # make
>> gcc  -I/usr/include -Wall   -c apple2.S
>> /tmp/cca01775.s: Assembler messages:
>> /tmp/cca01775.s:2163: Error: register does not match opcode suffix
>> /tmp/cca01775.s:2171: Error: register does not match opcode suffix
>> make: *** [apple2.o] Error 1
>> 
>
>Someone else reported this problem with RH4.1 and gcc v.2.7.2.1.  (I
>currently use RH3.0.3, kernel2.0.23, on an i586 with gcc v2.7.2 and
>everything works just fine)
>
   Well, I get the problem with gcc 2.7.2.1 on an i586. It's a simple
thing to fix, though. Here's the patch:

=====  BEGIN PATCH =====

*** apple2.S    Tue Mar  4 12:52:38 1997
--- apple2.S~   Sat Mar  1 15:15:25 1997
***************
*** 1572,1590 ****
     /* ----------------------------------
         HANG instruction
        ---------------------------------- */
  
  op_HANG:
!               decw    PC_Reg
                Continue
  
     /* ----------------------------------
         NOP_2 instruction
        ---------------------------------- */
  
  op_NOP_2:
!               incw    PC_Reg
                Continue
  
     /* ----------------------------------
         NOP_3 instruction
        ---------------------------------- */
--- 1572,1590 ----
     /* ----------------------------------
         HANG instruction
        ---------------------------------- */
  
  op_HANG:
!               decb    PC_Reg
                Continue
  
     /* ----------------------------------
         NOP_2 instruction
        ---------------------------------- */
  
  op_NOP_2:
!               incb    PC_Reg
                Continue
  
     /* ----------------------------------
         NOP_3 instruction
        ---------------------------------- */


===== END PATCH =====

   As you see, the patch looks like it fixes a couple of bugs. The
program counter being treated as a byte value when it should be a word.
I don't suspect that the HANG and NOP_2 operations are often invoked in
real code, so it shouldn't be a big problem. I suspect that it only
causes problems on page boundaries in the virtual machine, if the
byte-ordering is low-high for the PC. (reminiscent of the 6502 indirect
jump bug).
   Oh, this problem was also in v002, I patched it and kept going.

-- 
 Christopher Neufeld - Not a graduate student   neufeld@physics.utoronto.ca
 Home page:  http://caliban.physics.utoronto.ca/neufeld/Intro.html
 "Don't edit reality for the sake of simplicity"