Re: [hatari-devel] TOS bootup tester results

[Eero Tamminen <oak@xxxxxxxxxxxxxx>]

Hi,

On torstai 25 kesäkuu 2015, Nicolas Pomarède wrote:
> Le 25/06/2015 00:30, Eero Tamminen a écrit :
> > I've been running TOS bootup tester as I'm adding TT-RAM testing
> > support to it, and noticed bootup issues in several configurations.
> > 
> > I haven't yet verified that they all are real problems instead
> > of issues in the tester, except for TOS v3 + 4MB + 0MB TT-RAM issue
> > with latest WinUAE CPU core.  With that config, TOS bombs after memory
> > check. With more ST-RAM, some TT-RAM, or using VDI mode, TOS v3 works.
> 
> tos v3 expects a TT, which means a 68030 + MMU enabled. Did you enable
> MMU in your tests ?

I didn't.


> IIRC I remap the memory banks when TT RAM is used without MMU to get the
> equivalent of 32 bit addressing + simple remapping of the IO region
> (fake minimal mmu like result).
> So, this might explain why TT RAM prevents the crash, but the root cause
> is that you need to enable MMU for TOS v3 (as well as 32 bit adressing)

Enabling MMU doesn't help, but enabling 32-bit addressing makes TOS v3
work also with 4MB of RAM.  I.e. it was actually 32-bit addressing,
not TT-RAM, that helped.

Why things work without 32-bit addressing when there's >4MB of ST-RAM?


	- Eero

PS. I added a commit that disables 32-bit addressing with TOS v2 as
that doesn't seem to be compatible with it.