Re: [hatari-devel] m68k Linux under Hatari (was: 030 exceptions on auto-increment instructions)

[Eero Tamminen <oak@xxxxxxxxxxxxxx>]

Hi,

On 2/29/20 6:53 PM, Toni Wilen wrote:
> <snip>
>
> > > (There seems to be something fishy in kernel <->
> > > userspace interface, I've seen Linux to use
> > > use exception handling liberally when accessing
> > > all kinds of things, instead of it e.g. checking
> > > for NULL pointers etc.  And for some reasons that
> > > still doesn't work properly in Hatari.)
> >
> > hard to tell what's going on with just this kernel dump.
>
> <snip>
>
> Also enable MMU debugging (mmu_common.h #define MMUDEBUG 1 and
> cpummu030.h #define MMU030_DEBUG 1)
>
> When exactly does this error happen? After kernal has been loaded but
> before any disk accesses or when first user space program starts?

After Busybox init has started from IDE drive
under Atari Falcon emulation.  I normally load
kernel to ST-RAM (to get correct kernel addresses
in debugger), but I guess user-space programs
get loaded to TT-RAM.

All user-space programs don't have issues, but
in the ones with which it happens, its fully
reproducible. The cases I've noticed, are
documented at end of:
https://hatari.tuxfamily.org/doc/m68k-linux-for-hatari.txt

Earlier there were also randomly occurring issues,
but I haven't used new Hatari version enough so
that I could say whether those still happen or
not.


> I have noticed Linux drivers really expect error free hardware and try
> to use all kinds of advanced features, for example neither IDE or SCSI
> originally worked in WinUAE in Linux (or most other unixes) without
> fixing few edge-cases (that if I remember correctly, also made some
> drivers to misdetect hardware revision or type). Native Amiga programs
> and drivers didn't care.

Hatari SCSI emulation isn't yet good enough to
boot Linux, that's why I use IDE.


	- Eero