Date   : Fri, 03 Dec 2004 02:42:30 +0100
From   : John Kortink <kortink@...>
Subject: Re: ...at line ___

On Fri, 3 Dec 2004 01:17:43 +0000 (GMT), Andrew W wrote:

>In <URL:news:local.misc> on Thu 02 Dec, Johan Heuseveldt wrote:
>> 
>> > The next section of code is at &79FE using 50 bytes giving &7A30.
>> 
>> Ah, that could be a clue. The code is wrapping a 256 bytes page
border.
>> Some opcodes take that into account, some don't.
>> So you need to do some checking on that. 
>> 
>> Sorry I can't be more specific. Some basic textbooks are not within
>> reasonable reach.
>
>I thought it might be something like that (in the way that ARM
>instructions need to be 'word-aligned').

It isn't.

There are no issues with alignment on the NMOS 6502s
except for the one Sprow already mentioned (with JMP
(&xxFF)).

There is a second bug in that reads from locations
with indexed absolute addressing that cross a page
boundary (when adding the index) accidentally access
two locations : the target location and the location
one page lower in memory. Most painfully noticed when
the second location is read sensitive, hardly ever
noticable otherwise.

Neither bug is particularly dangerous except in
special cases. The first is fixed in 65C02s. The
second in some 65C02s I think (some still do the
spurious read).


John Kortink

-- 

Email    : kortink@...         
Homepage : http://www.inter.nl.net/users/J.Kortink