[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [MiNT] Teradesk compile problem



2009/11/30 Vincent Rivière <vincent.riviere@freesbee.fr>:
> Petr Stehlik a écrit :
>>>
>>> No, this is a privileged m68k insn for accessing the usp (User Stack
>>> Pointer) register.
>>
>> Right, when you're in super mode you can access both stacks - usp and
>> ssp...
>>
>> It doesn't explain why assembler would complain about that, though.
>
> Maybe AHCC requires a special directive to enable privileged instructions ?
> Maybe it is ".super" ?
>
> There is a hint in the latest release note :
> http://members.chello.nl/h.robbers/V3B4.TXT
>
> option -*u              default .super
>
>
> --
> Vincent Rivière

Thanks Vincent, I placed the ".super" at the beginning of the second
module in BTST.S, and it compiled.

I also down loaded the new version, but it crashes during "make" after
only a couple of files, probably due too all the warnings that get
generated.. ;-(

Tis cool, was an exercise in how AHCC perfomed on certain PureC
projects, and how well is worked under EasyMint

At least I know how to get supervisor code working in AHCC, and I will
look at adding support later, in the meantime if I want to look at
TeraDesk again, I will use PureC.. I was just hoping to get an
"alternate" PureC dev kit going ..

Thanks for the help guys..

EDIT:
well I spent a bit more time with AHCC and Teradesk (only about
another 5 hours), managed to get all the libraries to compile & link,
had to hack an S file as it "couldn't" include another S file, also
(maybe) fixed what looked like a "code hack" (jmp_a7) to something the
compiler liked (wrapped value in {}). Still even though I managed to
get objects compiled, the linker fails miserably, and the output is
unhelpful (linker: failed 5).

If I read it right, this is the same (or similar) problem as GEM
programs not running on the STe, they must be "well written" and all
"legal like". Anyway, I'll have a look at it another time, but I am at
 least prepared to throw HighWire at AHCC now, most of the "errors"
will just be "code syntax".

It is unfortunate AHCC is now less stable, at least on EasyMiNT
running 1.16.3, so I will give a few ports the GNU treatment (because
cat << EOF works in EasyMiNT) and then bring up my main setup and test
under 1.16.3 and 1.17 (without the SpareMiNT, and unfortunately, cat
<< EOF). I will also try the same compile tests I have spent doing
over the last 24 hours with XaAES, HighWire, and others, try look for
some fixes when I do, learn some more code..

Cheers, and thanks again

Paul