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

Re: [MiNT] [Aranym-dev] [ARANYM.XIF] Hangs with 1.16.1 MiNT kernel



On Sun, 2006-01-29 at 14:41 -0500, Standa Opichal wrote:
> Hi!
> 
> Well, Frank, here you go! :)
> 
> IMHO this is actually a bug in our binutils adaptation. When I save
> only the d4 it is enough to get FreeMiNT running. And it is caused
> by this the following binutils-2.13.2.1-mint.patch (from the RPM):
> 
> ...
> +  /* Generate the jump instruction to the entry point.  In m68k
> +     assembler mnemnonics it looks more or less like this:
> +
> +       move.l  exec_bytes->e_entry(pc), d4
> +       jmp     0(pc,d4.l)
> +
> +     Sorry for the wrong syntax.  As a real assembler addict I
> +     never actually use an assembler.  I edit my binaries manually
> +     with a hex editor, looks much cooler and it strengthens your
> +     abstraction abilities.  */
> +
> +  exec_bytes->g_jump_entry[0] = 0x28;
> +  exec_bytes->g_jump_entry[1] = 0x3a;
> +  exec_bytes->g_jump_entry[2] = 0x00;
> +  exec_bytes->g_jump_entry[3] = 0x1a;
> +  exec_bytes->g_jump_entry[4] = 0x4e;
> +  exec_bytes->g_jump_entry[5] = 0xfb;
> +  exec_bytes->g_jump_entry[6] = 0x48;
> +  exec_bytes->g_jump_entry[7] = 0xfa;
> ...
> 
> 
> The question why had Guido used 'd4' for binutils and not e.g. 'd0'
> or directly 'bra exec_bytes->e_entry(pc)' which would not use any
> register....
> 
> I tracked this down by the aranym's flightrecorder feature! Thanks
> Milan for this one!
> 
> Standa
> 
> PS: I would rather see a new FreeMiNT release + new binutils
> package (with the fixed libbdf stuff I have reported) than 100 of
> whatever so meaningless as currency. :)

Well, 
I can help.

I mean we have working gcc 3.3.6, latest binutils is fine.  I can
implement this bug fix and fix the strip and stack tools included.  So
why not build freemint from scratch on the new stuff?  It'd be a fun
exercise for the new sparemint website's distinct repositories ;)  The
only problem is that if for instance your system ran with 68020-60 libs
then the resulting binaries that you build that linked those libs for
68000 would not work....  Other things would be the UVM port that Mikro
abandoned/delayed and shared libs, both of which I'm not competent
enough yet to do.  It'd be nice to wait until this is done.

But honestly.  Does anyone use freemint on 68000?  I know it's a fun
exercise and all but I think most 68000 users use MagiC because it's
more trim.

Duped to mintlist.

This new freemint is something I'd like to get done with everyone's
blessings.  We can kind of establish a roadmap and start building
packages :)

Thanks,
Mark


Attachment: signature.asc
Description: This is a digitally signed message part