[hatari-devel] IDE bus errors (was: Big problems with bus error on Falcon)

Thomas H. th.huth at gmx.de
Sat Jan 16 11:44:13 CET 2010


> Datum: Thu, 14 Jan 2010 22:47:38 +0100
> Von: Laurent Sallafranque <laurent.sallafranque at free.fr>
> 
> >>
> >> ==> Simply enable an IDE hard disk image an the bus error should be
> >> gone!
> >>     
> >
> > It does go away and the program starts, it doesn't even need DSP.
> >
> > It's "Tron2001 demo.  Game by ICE" from Umich:
> > http://www.umich.edu/%7Earchive/atari/Falcon/Games/tron2001.lzh
> >
> > Seems to non-interactive demo, I'll move it to demos in compatibility
> list
> > and note that it needs IDE image attached and starts in wrong resolution
> > so graphics are wrong.
> >   
> 
> Does it mean that this demo didn't work on a Falcon without hard disk ?
> Or simply that we are too restrictive in our bus error ?
> 

I think it also works on Falcon without hard disk.
The problem is the following:
On (Mega-)ST/STE, maybe also TT, the IDE adaptor was only optional, so the IDE memory space normally generates an bus error when no IDE adaptor (and thus no IDE drive) has been soldered. That's the behaviour that Hatari currently emulates - bus errors when no IDE drive available.

On Falcon, the IDE adaptor is always present (even if no IDE drive has been plugged), so I guess this memory space never generates bus errors on real Falcon.

Now, all TOS versions >= 2.0x correctly check if an IDE adaptor is present by trying to access the memory region and intercepting the bus error in case that it is not there. So there has never been a problem with TOS 4.0x on Hatari yet.

I didn't know that there would be any Falcon programs that blindly tried to access that memory space without trying to intercept bus errors as well. So for Falcon mode, we should maybe try to get rid of bus errors in this memory region. You could try something like to add a "if (ConfigureParams.System.nMachineType != MACHINE_FALCON)" right before the M68000_BusError() statements in ide.c to see whether this helps.

 Thomas



More information about the hatari-devel mailing list