Re: [hatari-devel] Apex 2.10d question

[Thomas Huth <th.huth@xxxxxxxxx>]

On 19/12/2020 18.47, Eero Tamminen wrote:
> Hi,
> 
> On 12/19/20 9:31 AM, Laurent Sallafranque wrote:
>> Eero, in the compatibility list comment area, Apex 2.10d says that it has
>> worked with an early version of hatari.
>>
>> I've tried with hatari 1.4, 1.6.2, 1.8 and current version and it always
>> crash on startup (for various reasons that were fixed by the time).
>>
>> Do you remember which version make it run ?
>>
>> (I'd like to bissect it in case we could find the regression).
> 
> I found from Git history that my comment wasn't
> correct.  Original one for Hatari v1.3 was:
> ----------------------------------------------
> Crashes with a "MOVE.W $fff20000,D0" instruction.
> It's unclear why this works on a real Falcon.
> The separate image viewers work fine.
> ----------------------------------------------
> 
> I.e. the main program has never worked.
> 
> I had replaced that entry with one for v2.41, and
> later I added the demo version back with incorrect
> comment.  That will be fixed when I commit my
> compatibility list updates.
> 
> If you can check whether it really works on
> real Falcon that would be great!

IIRC I once tested that version on my Falcon, and it worked, indeed - but in
very mysterious ways: If I just ran the program, there was no bus error
here. But if I loaded it into a debugger first (IIRC I used Bugaboo), and
set a breakpoint in front of the instruction, I got a bus error indeed
instead when trying to execute the instruction.

See also the old mail thread:

-
https://listengine.tuxfamily.org/lists.tuxfamily.org/hatari-devel/2015/10/msg00038.html

It's just completely unclear when this address is causing a bus error and
when it's ignored.

 Thomas