Re: [hatari-devel] IDE IO register range access commit

[Thomas Huth <th.huth@xxxxxxxxx>]

Am Mon, 21 Sep 2020 16:24:34 +0200
schrieb Nicolas Pomarède <npomarede@xxxxxxxxxxxx>:

> Le 21/09/2020 à 15:24, Nicolas Pomarède a écrit :
> > Le 21/09/2020 à 15:16, Eero Tamminen a écrit :  
> >> Hi Nicolas,
> >>
> >> In your IDE IO register range access commit:
> >> https://git.tuxfamily.org/hatari/hatari.git/commit/?id=872220ab26d0e8f11f018f5740a8ac96753acac7 
> >>
> >>
> >> I think your check should be for IDE being enabled:
> >> (ConfigureParams.Ide[0].bUseDevice ||
> >> ConfigureParams.Ide[1].bUseDevice))
> >>
> >> Not for Falcon.
> >>
> >>
> >> For example, Linux doesn't seem to work with ACSI,
> >> and Hatari SCSI emulation isn't complete enough,
> >> so I need to use IDE disk image to test m68k Linux
> >> with TT emulation.  
> > 
> > Hi
> > 
> > OK, I can change the test.  
> 
> looking at ide.c, I see that accesses to F00000-F00040 will return a
> bus error if Ide_MmioIsAvailable() return false.
> 
> But Ide_MmioIsAvailable() only does :
>          return ConfigureParams.Ide[0].bUseDevice;
> 
> Regarding Eero's change above, shouldn't it test for
> Ide[1].bUseDevice too ? Or is Ide[0].bUseDevice always true in case
> Ide[1] is also enabled, and in that case testing only Ide[0] is
> enough ?

IIRC IDE0 is always the "master" drive, and you can not have just a
secondary drive on the bus without the master drive... so testing for
IDE0 should be enough.

But: There should also be an additional test for Falcon here, since the
IO registers are *always* available on the Falcon, no matter whether
there is a drive or not. IIRC there are some hard disk drivers (was it
Cecile?) crashing in Hatari since they expect the IDE IO registers to
always be there.

 Thomas