ZX-News #07 06 октября 1996

Программирование

Programmers - On the bug in Shadow Service Monitor'e (Scorpio)

              Programmers



        © Alexander Mayorov, 1996.



    About Bugs in SHADOW SERVICE MONITOR


  For many times, vlzamyvaya with
Scorpion all kinds of protection, I ran across
to "glyuchok, but somehow it was lazy with it
understand. Then one day, putting all
the case, I "caught" the bug.

The thing that working with the register <R>
in skorpionovskom monitor not implemented
absolutely correct!

Through this all "ksorki" Register with <R> need to go 
completely or trace mode or in real time. And you can not part 
ottrassirovat the monitor, while the rest continue to perform 
in real-time.


Just want to say that this bug is present on all versions of 
skorpoina what I could find. A version of these:



        Normal ROM: v2.7b, v2.95

          ProfPZU: v3.42, v3.92


       Now specifically about the bug:

     "Buggy" command LD R, A and LD A, R.


      Write ma-vermillion program from:

      ----------------------------

           DI

           LD A, # 00

           LD R, A


      and obtain the following results:

 - When performing this programmki

   mode trace <R> will:


   in the STS - R = # 00, as it should be


   in MONITOR - R = # 01 - glitch!

 -In real time:
 (Ie put a breakpoint after LD R, A)


   in the STS - R = # 00, as expected


   in MONITOR - R = # 02, better than nowhere!


        And now, a team of LD A, R:

        ------------------------
 STS: After you run the battery, as well as in the case <R> 
will find the value of the register before the regeneration 
team plus two. (Which corresponds to reference data)


 MONITOR: After the command register contents <R> increase by 
two, but the battery will register value before the command <R> 
PLUS ONE! (Ie One unit LESS what should be)



   If someone does not understand vysheskazanogo,

  then look below narisovanuyu plate:



        the command LD A, R


               Register R Battery


  to team # 00 with no difference


  in the STS-e # 02 # 02


  in MONITOR-e # 02 # 01



         In my view everything is clear ...


  I hope that you have already guessed why
"Ksorki" normally take place in the
trace, despite these errors, but
any case, try to explain:


  Initially, the register <R> sent some
number. Due to an error, there (in the case
<R>) Is the number plus one. Then,
when the number of <R> sent to the battery, it gets there 
DECREASE One unit, and since (-1) + (+1) = 0, all there seems 
to be and in order. But if the middle "ksorki" you

tired of clicking on "True Video" or tired of waiting (SLOW 
...) and you're putting a breakpoint, run the program in real 
time, then ... 


       Another valid point:


  If you log in shadow businesses by Magik-y or
by breakpoint-in you will see the register <R>
certain number (eg # 25), do not believe my eyes! There must be 
a number on TWO LESS! And so that was a <R>

what we need, say shadow businesses "r = r-2".
But do not forget that the 7-bit should not be changed.

"What does this do we do?" - You ask.


   On this there are several answers:

1) Do not break into programs with the help of shadow 
businesses. 

2) Wait, what if this bug fix ...

3) The hacking programs to adjust the register value <R>. Ie at 
the entrance of shadow businesses reduce <R> at One unit can 
engage and tracing. And before you leave shadow businesses must 
again reduce <R> at One unit. 

That such things ...

P.S. Perhaps I am wrong somewhere! Check
all by yourself!

P.P.S. As you see, not only in ZxNews are glitches ... :)


  Sincerely, * MAS * from Omega group.


        __________________________